LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if reqoested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDue.p65~p.15

 

REPORTED NEEDS OF LOW-INCOME MOTHERS: IMPACT ON SERVICE
UTILIZATION AND INTERVENTION PROGRAM OUTCOMES

By

Lauren Michel Rosenkoetter Barton

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Psychology

2002

 

ABSTRACT

REPORTED NEEDS OF LOW-INCOME MOTHERS: IMPACT ON SERVICE
UTILIZATION AND INTERVENTION PROGRAM OUTCOMES

By

Lauren Michel Rosenkoetter Barton

This study investigated individual differences in reported needs within a
population of low-income pregnant women. Of particular interest were observed
differences between the extent of concern identified by participants (reported
need) and that which might have been expected by providers based on more
objective characteristics of each participant’s self-described situation at the
outset of intervention (expected need). Path analysis was employed to test the
validity of two conceptual models. One model examined factors predicting
reported needs during pregnancy. The second model investigated the influence
of expected and reported needs on individuals’ postnatal utilization of two
different home-visiting support services. Subsequent quality of parenting
interactions, home environments provided, and infant mental development also
were included in the second model as potential outcomes of increased service
utilization. Findings indicated that a continuum of individual differences exists in
need reporting behavior among pregnant low-income women. Reported needs
were based on substantive expected needs, but these two constructs were not
equivalent. In addition to expected needs, increased stressor responses
(depressive symptoms and perceived stress) and higher levels of personal

control (mastery and global self-esteem) both exerted significant direct positive

effects on reported needs. Higher levels of reported needs reflected a greater
level of “readiness” for services that consisted of both a sense of necessary
immediacy to cope with the concern and feelings of personal control to address
the situation. This critical difference between expected and reported needs also
was evidenced in subsequent client response to need. Unlike expected needs,
reported needs predicted higher levels of postnatal service utilization behavior in
both treatment groups. In addition to levels of reported needs, intensity of
service use also was influenced by perceived program helpfulness, competing
time demands stemming from maternal employment, and accessibility of
services. Implications are discussed regarding family-centered care practices,
adaptation of preventive interventions to help low-income women identify

expected needs, and sample selection for outcome evaluations of interventions.

Firs

Copyright by
LAUREN MICHEL ROSENKOETTER BARTON
2002

Dedicated in loving memory of Donna Barton

ACKNOWLEDGEMENTS

Although this dissertation only bears the name of one author, it would not
have been possible without the support of numerous people. The research
participants themselves graciously gave their time, their trust, and often their
tears to share their stories with us. These women taught me more than they will
ever know. I am grateful. Perhaps through this paper and future work, I may
begin to give voice to their experiences and help the community services around
them better reach and respond to others in similar situations.

Thanks also to the team of people so heartily invested in every phase of
design and implementation of the Michigan Maternal Health Services Study. The
data collectors on this project could write many books about “lessons learned”,
and yet they never wavered in their commitment to listening with their hearts and
insuring that participants’ experiences could be accurately translated and truly
heard in the language of project data files. For this, and their tireless efforts
chasing and tracking study participants with respect and creativity, I thank them.

I am also indebted to the team of individuals, spearheaded by Leanne
Kallemeyn, who devoted countless hours to not just accurately cleaning, but also
to clarifying the meaning, of project data. Such self-motivated care and concern
with every last detail gives me the utmost confidence about using our numbers to
speak on behalf of our participants. Special thanks to Jason Davis as well for his
efforts collecting the program service use data on which this study depends.

Finally, Marcia Gebben was instrumental to the effective development and

implementation of the Michigan Maternal Health Services Study. Her dedication
to continuously identifying the next essential steps, asking the tough questions,
organizing a practical plan, and finding a way to methodically get those things
done well was such a source of stability that it was easy to take for granted in the
midst of all of the other challenges posed by the study. A heartfelt thanks for
these efforts as well as for her ongoing willingness to listen, show, and share so
much with me over these past years.

Immeasurable gratitude is also due to those who were formative in my
professional development throughout the graduate school process. Special
thanks to my advisor, Hiram Fitzgerald, for his candor and willingness to answer
even the dumbest of questions with careful respect. His dedication to go beyond
talking about assets to incorporating a strengths-based approach to mentorship
gave me the confidence I needed for the challenges I faced. I would also like to
extend my appreciation to Lee Anne Roman for her sustained commitment to
support my growth in conceptualizing and conducting community-based
research. Through her efforts I have had the opportunity to learn about how to
nurture fragile relationships with community partners, finesse the politics of turf
battles, and overcome systemic barriers that obstruct opportunities for low-
income families. Such lessons cannot be gleaned from books alone. Next, I
would like to thank Alexander von Eye for his patience and willingness to explain
things again and again and again until some of the Greek statistical jargon finally
made sense. The probability of encountering another statistician with such high

levels of both knowledge and patience was exceedingly low. Thank you!

I also would like to acknowledge my appreciation to each of the members
of the MOMS Research team. They provided me with valuable insights about
interdisciplinary perspectives on problems and constantly challenged me to
redefine and reconceptualize my assumptions and ideas. Additional thanks also
are due to Pennie Foster-Fishman and Rachel Schiffman for their valuable
feedback and shared insights about the complexities of community-based
research.

Finally, acknowledgements would not be complete without thanking my
family and friends for their tireless support. Friends were never too far away or
too busy to share their smiles, sympathy, or suggestions. These gifts were a
source of great support and a considerable contribution to the development of
this document. Special thanks to all the members of the Barton family whose fun
visits helped erase the stress at hand and who were the first to cheer about each
step of progress along the way. My gratitude also extends to my sister, Kristen
Mathisen, and my brother, Timothy Rosenkoetter. Kristen’s empathy and caring
was never far away with notes and calls timed to say just the right thing.
Timothy’s immersion in similar toil was a source of much comfort over many
nocturnal telephone chats. My grandparents’ encouragement and support also
was greatly appreciated. Next, I wish to thank my parents for their continued
dedication to helping me define and achieve my dreams. Their steadfast
determination, hard work, and persistence have set admirable examples for me.
For this, and for all that they have given me and challenged me to do over the

years, I cannot thank them enough.

Last, but definitely not least, I wish to thank my husband, John. Years ago
when I spoke of a degree and dissertation he had no idea how my choice would
change his life. Yet, despite the many sacrifices and frustrations, he has been
unconditionally supportive of me and of my work. For knowing when to listen
and when to act, how to encourage and if to distract, for sometimes challenging
and always loving, I thank you. This study truly was a team effort.

This study was supported by grant MCJ-260743 from the Maternal and
Child Health Bureau (Title V, Social Security Act) Health Resources and Services

Administration, Department of Health and Human Services.

PREFACE

The problem statements in numerous research documents describe
groups of individuals in need or at high risk for experiencing less optimal health
and developmental outcomes. Often, these target groups are defined rather
matter-of-factly. Programs or policies in place are evaluated to see if they can
reduce the clients’ needs or ameliorate the negative outcomes that are typically
observed in that population. Hypotheses are tested to determine which specific
needs place individuals at highest risk and if a causal path can be charted.

Some studies define individuals in need as those with a specific set of
predetermined situational life factors that suggest a high need should exist. In
contrast, other studies define individuals in need based on the self-perceptions of
the individuals themselves. Often research with these two types of
methodologies is compared and findings are debated without careful attention to
the sampling difference between them. What kinds of differences might result
from these alternative approaches?

This dissertation provides a starting point for thinking about the
implications of these different ways of defining a sample of individuals in need.
Specifically, it investigates individual differences in the extent to which people
report needs that are consistent with expectations based on more objective,
criteria-based definitions of need. Do reported needs essentially duplicate
expected needs? If not, what factors contribute to differences in reported needs?
Can individual differences in reported needs be anticipated based on

psychosocial characteristics of individuals? Do reported needs reflect a

“readiness” to use home-visiting services? Is individual variation in reported
needs predictive of differential programmatic participation rates and of positive
parenting and child development outcomes?

A richer understanding of individual differences in reporting needs may
offer important information for improving family-centered service delivery.
Investigating reported needs may help identify people who are most likely to
benefit from existing services and may provide a starting point for developing
alternative interventions that prepare people to benefit from available services.
In addition, this research may lead to more precise inferences about the
effectiveness of preventive intervention programs in particular populations by
providing better clarity about the characteristics of the research samples from
which we generalize. Considerable variation exists between the “high need”
individuals in our communities. Listening to and learning from how individuals

report their needs may help us consider what actions may contribute to solutions.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................. xvi

LIST OF FIGURES .......................................................................................... xvii
LIST OF APPENDICES ................................................................................... xviii
OVERVIEW ..................................................................................................... 1

CHAPTER 1: Predicting Reported Needs in Low-Income Pregnant Women
INTRODUCTION

Interventions Targeting Low-Income Pregnant Women .............................. 7
Two Approaches to Defining High Need Groups ........................................ 10
Clarifying Terminology ................................................................................ 14
Reporting Perceived Needs ........................................................................ 15
Conceptual Model for Predicting Reported Needs in Low-Income
Pregnant Women ..................................................................................... 18
The Impact of state on reported needs ................................................. 19
Relationships between perceived ability to cope and
reported needs ................................................................................... 22
Self-disclosure of perceived needs ....................................................... 27
Summary ............................................................................................... 31
METHODS
Sample: Michigan Maternal Health Services Study Recruitment
Sample and Primary Study Design ......................................................... 32
Treatment conditions ............................................................................. 32
Treatment A: Maternal and Infant Support Services
Standard of Care (SOC) ............................................................... 33
Treatment B: Nurse-Community Health Worker Team Care
(Nurse-CHW Team care) .............................................................. 34
Eligibility ................................................................................................ 34
Recruitment ........................................................................................... 36
Sample: Dissertation Subsample ............................................................... 37
Eligibility ................................................................................................ 37
Sample characteristics .......................................................................... 38
Data Collection ........................................................................................... 40
Measurement of Variables .......................................................................... 40
Perceived stress .................................................................................... 41
Depressive symptoms ........................................................................... 42
Social support ....................................................................................... 43
Availability of social network ............................................................ 44
Received support from social network ............................................. 44

xii

TABLE OF CONTENTS (Cont)

CHAPTER 1: Predicting Reported Needs in Low-Income Pregnant Women

METHODS (Cont)
Measurement of Variables (Cont)

Perceived personal control .................................................................... 45
Mastery ............................................................................................ 45
Self-esteem ...................................................................................... 46
Mothering ability esteem .................................................................. 47
Perceived personal control construct ............................................... 48

Negative network orientation ................................................................. 49

Comfort with self-disclosure .................................................................. 51

Prior involvement with Child Protective Services .................................. 51

Expected and reported need indices ..................................................... 51
Reported needs index ...................................................................... 52
Expected needs index ...................................................................... 52
Need index characteristics ............................................................... 53

Data Analysis .............................................................................................. 54
Missing data imputation ........................................................................ 54
Analytical methods ................................................................................ 55

RESULTS

Manifest Model ........................................................................................... 57

Model adaptations ................................................................................. 57

Model fit ................................................................................................. 58

Observed effects ................................................................................... 59

Structural Model ......................................................................................... 61

Model adaptations ................................................................................. 61

Model fit ................................................................................................. 64

Measurement models ............................................................................ 64

Observed effects ................................................................................... 66

DISCUSSION

Influences on Reported Needs ................................................................... 68
Expected needs .................................................................................... 68
Stressor responses ............................................................................... 69

l Depressive symptoms ...................................................................... 72

Perceived personal control .................................................................... 73
Mothering ability esteem .................................................................. 76
Personal control and appraisal processes ....................................... 76

l Cognitive ability ..................................................................................... 81

Social support ....................................................................................... 82

Disclosure ............................................................................................. 84

Summary .................................................................................................... 87

xiii

TABLE OF CONTENTS (Cont)

CHAPTER 2: Predicting the Influence of Reporting Tendencies on
Service Utilization and Intervention Program Outcomes
INTRODUCTION
Conceptual Model for Predicting the Impact of Maternal Reporting

on Service Utilization and Intervention Program Outcomes ..................... 89
Relationships between needs and service utilization ............................ 90
Maternal availability and service utilization ............................................ 91
Perceived program helpfulness and service utilization .......................... 92
The impact of state on service use and parenting interactions .............. 93
Predicting the quality of parent-child interactions: Influences of

intervention, child cues, and the mothers’ self-reporting of needs ...... 95
Predictors of mental development in children .................................. 98
Intervention group differences ............................................................... 100
Summary ............................................................................................... 103
METHODS
Sample ....................................................................................................... 104
Data Collection ........................................................................................... 104
Measurement of Variables .......................................................................... 104
Expected and reported need indices ..................................................... 104
Degree of match between expected and reported needs ...................... 105
Perceived stress .................................................................................... 105
Depressive symptoms ........................................................................... 105
Perceived helpfulness of services ......................................................... 106
Postnatal employment status ................................................................ 106
Postnatal residence in Kent County ...................................................... 106
MSS/ISS postnatal service use intensity ............................................... 107
Child’s mental development .................................................................. 107
Child’s test behavior .............................................................................. 108
Quality of parenting interactions ............................................................ 108
Clarity of child’s cues ............................................................................ 109
Quality of home environment ................................................................ 110
Child’s health ......................................................................................... 111
Data Analysis .............................................................................................. 1 13
Missing data imputation ........................................................................ 1 13
Analytical methods ................................................................................ 114
RESULTS
Manifest Model ........................................................................................... 1 15
Model adaptations ................................................................................. 1 15
Model fit ................................................................................................. 1 15
Observed effects ................................................................................... 116

xiv

TABLE OF CONTENTS (Cont.)

CHAPTER 2: Predicting The Influence Of Reporting Tendencies On
Mother-Infant Interactions And Child Development Outcomes

RESULTS (Cont)
Exploratory Analyses .................................................................................. 1 17
Service use ........................................................................................... 118
Infant health .......................................................................................... 1 19
Infant sex ............................................................................................... 1 19
Maternal education ................................................................................ 1 19

DISCUSSION

Influences on Service Utilization ................................................................. 121
Expected needs .................................................................................... 121
Reported needs ..................................................................................... 123
Maternal availability to receive services ................................................ 125
Postnatal employment ..................................................................... 125
Time points in Kent county ............................................................... 127
Service helpfulness ............................................................................... 128
Perceived stress and service utilization ................................................ 130
Perceived stress and parenting interactions .......................................... 132
Depressive symptoms ........................................................................... 134
Maternal and Child Program Outcomes ..................................................... 136
Maternal interactions ............................................................................. 1 37
Home environment ................................................................................ 142
Infant mental development .................................................................... 147
Child performance factors ..................................................................... 149
Test day behavior ............................................................................ 149
E Clarity of cues .................................................................................. 150
Poor infant health ............................................................................. 151
Summary .................................................................................................... 153

CHAPTER 3: Reported Needs Among Low-Income Pregnant Women:
' Discussion and Implications

Study Aims ....................................................................................................... 155
Is the Distinction Between Expected and Reported Needs Important? ............ 155
What Influences Reported Needs? .................................................................. 158
What Contributes to MSS/ISS Service Utilization? .......................................... 160
How Does Service Utilization Relate to Outcomes? ........................................ 161
Why Focus on Needs in a Strengths-Based Setting? ...................................... 162
Study Limitations and Recommendations for Future Research ....................... 163
Closing Comments ........................................................................................... 165
REFERENCES ................................................................................................ 225

XV

LIST OF TABLES

Table 1. Comparison of treatment characteristics across study
intervention groups .......................................................................................... 167

Table 2. Characteristics of Medicaid-eligible mothers in the

Michigan Maternal Health Services Study (n=613) sample

and in dissertation subsample (n=438) at study enrollment .......................... 171
Table 3. Author-generated expected and reported need indices .................... 173
Table 4. Comparison of percentages of participants in each

intervention group reporting needs or concerns on each of

the reported need index variables ................................................................. 177

Table 5. Enrollment characteristics of study participants by
intervention group (n = 438) .......................................................................... 178

xvi

LIST OF FIGURES

Figure 1. Conceptual model 1 predicting reported needs in
low-income pregnant women ........................................................................ 182

Figure 2. Path analysis findings testing conceptual model 1
using manifest variables and estimating parameters with
maximum likelihood approach ....................................................................... 183

Figure 3. Path analysis findings testing conceptual model 1
using manifest variables and estimating parameters with
generalized least squares approach ............................................................. 184

Figure 4. Path analysis findings testing conceptual model 1
using manifest variables and estimating parameters with
maximum likelihood approach; model includes path linking
all personal control construct variables ......................................................... 185

Figure 5. Structural equation model testing latent factor
structure of conceptual model 1; parameters estimated with
maximum likelihood methods ........................................................................ 186

Figure 6. Conceptual model for predicting the impact of
maternal reporting of needs on subsequent service utilization
and intervention program outcomes in low-income families .......................... 187

Figure 7. Path analysis findings from testing conceptual model 2
with maximum likelihood parameter estimation in participants
randomized to Nurse-CHW Team care intervention ..................................... 188

Figure 8. Path analysis findings from testing conceptual model 2
with maximum likelihood parameter estimation in participants
randomized to SOC intervention ................................................................... 189

xvii

LIST OF APPENDICES

Appendix A. Eligibility criteria for enrollment into the
Michigan Maternal Health Services Study .................................................... 191

Appendix B. Reasons that Michigan Maternal Health Services Study
cases were excluded from the sample for data analysis for
dissertation study .......................................................................................... 192

Appendix C. Measurement model of perceived personal control construct 194

Appendix D. Factor analysis and reliability analysis of 13
Network Orientation Scale items ................................................................... 195

Appendix E. Distribution of expected needs, reported needs,
and discrepancies between reported and expected needs
among dissertation study participants ............................................................ 200

Appendix F. Means, standard deviations, and correlations used
in path analysis of conceptual model 1 ......................................................... 202

Appendix G. Fit indices for path analysis of conceptual model 1
using manifest variables and estimating parameters with
maximum likelihood methods ........................................................................ 204

Appendix H. Fit indices for path analysis of conceptual model 1

using manifest variables and estimating parameters with

generalized least squares methods .............................................................. 205
Appendix I. Fit indices for path analysis of conceptual model 1

that incorporates path from mastery/self-esteem to

mothering ability esteem ............................................................................... 206

Appendix J. Standardized estimates of error terms modeled
in latent factor model of conceptual model 1 ................................................. 207

Appendix K. Standardized estimates of covariation in error terms
modeled in latent factor model of conceptual model 1 .................................. 209

Appendix L. Fit indices for conceptual model 1 latent factor model ................ 211

xviii

LIST OF APPENDICES (Cont)

Appendix M. Conceptual model 2 variable means and
standard deviations for the Nurse-CHW Team care group
before and after missing data imputation ...................................................... 212

Appendix N. Conceptual model 2 variable means and
standard deviations for Standard of Care group
before and after missing data imputation ...................................................... 216

Appendix 0. Fit indices for path analysis of conceptual model 2
using manifest variables and estimating parameter
with maximum likelihood methods ................................................................ 220

Appendix P. Standardized estimates of error terms modeled
in path analysis of conceptual model 2 — Nurse-CHW Team Care
intervention group ......................................................................................... 222

Appendix 0. Standardized estimates of error terms modeled
in path analysis of conceptual model 2 -— SOC intervention group ................ 223

Appendix R. Descriptive information about service provider turnover for
individuals in the Nurse-CHW Team care intervention group (n = 196). .......... 224

xix

REPORTED NEEDS OF LOW-INCOME MOTHERS: IMPACT ON SERVICE
UTILIZATION AND INTERVENTION PROGRAM OUTCOMES1

OVERVIEW

Research suggests that early intervention programs can be very effective
in supporting families and in enhancing developmental outcomes for children.
Indeed, a recent review sponsored by the Institute of Medicine recommended
that researchers shift their focus from asking if early intervention works to asking
what type of intervention works best with whom and under what circumstances?
(Shonkoff & Phillips, 2000). Such questions are important because, to date, the
positive impacts of supportive services only have been shared by a portion, but
not all, of those persons targeted for assistance (Daro & Gelles, 1992; Gomby,
Culross, & Behrrnan, 1999).

One factor that undermines the effectiveness of services is underutilization
(Gomby et al., 1999). Frequently individuals who might significantly benefit from
active participation are missed by these services (Spieker, Solchany, McKenna,
DeKlyen, & Barnard, 2000). Often families with high needs and few personal and
psychological resources are unlikely to effectively initiate use of available
services (Apodaca, Woodruff, Candelaria, Elder, & Zlot, 1997; Huber, Holditch-
Davis, & Brandon, 1993; Minde et al., 1980; Spieker et al., 2000). Even those

who do connect with services will vary considerably in the extent to which they

 

‘ While the title refers to “intervention program outcomes”, it is important to note that the only
outcomes examined in this investigation were quality of parent-child interactions, characteristics
of the home environment, and infant mental development. These were not the only outcomes
expected to change as a result of participation in either of the intervention programs described in
the study.

remain involved consistently or choose to discontinue participation (Clinton,
1992; Faver, Crawford, & Combs-Orme, 1999; Navaie-Waliser et al., 2000). For
instance, across evaluated Healthy Family America programs, 20 — 30% of
families who initially accepted offers for services failed to successfully engage in
active program participation or completely dropped out altogether (Daro &
Harding, 1999). Such underutilization may severely diminish the social and
economic benefits that parent support and early intervention programs could
othenivise realize. In a recent review about mothers who are difficult to engage in
prevention and intervention programs, Spieker and her colleagues (2000) wrote
“...what is needed to assure that the developmental needs of all children are met
is a way to identify families who qualify for participation in a prevention program
but who are unlikely to engage in it because of parental risk factors and then to
provide mental health treatment and a therapeutic relationship in order to bring
them to the point where it would be possible for them to engage in the program”
(p. 204).

In discussions with practitioners about families who are difficult to engage
in service delivery, many providers refer to broad individual differences between
clients in their “readiness” to participate or fully engage in the intervention
(Kitzman, Cole, Yoos, & Olds, 1997). Although readiness is a concept that has
not yet been fully defined, a growing number of studies refer to constructs like
this as explaining differences in individuals’ behaviors. For instance, some
necessary level of personal readiness has been discussed as a component that

affects problem solving competence (Brammer, 1990). Varying stages of

 

cognitive readiness also have been identified as an important influence on the
process of behavioral change in therapeutic treatment of addictive behaviors
(Prochaska, 1995; Prochaska, Johnson, & Lee, 1998; Prochaska & Prochaska,
1999) and in the process leading to enactment of risk-reducing behaviors in the
transmission of HIV/AIDS (Catania, Kegeles, & Coates, 1990). The progress of
individuals along this readiness process continuum even has been associated
with differential probabilities that people will terminate substance abuse treatment
early (Prochaska, 1996).

Qualitative studies lend further support that actively engaging in help-
seeking actions may involve a process of developing increasing readiness to act.
Work from both health and emotional support domains suggests that effective
help—seeking involves not simply a decision, but rather a longer process, that
eventually results in service utilization (Patterson, Douglas, Patterson, & Bradle,
1992; Rawlins, 1991). This process typically begins with the recognition that a
concern or need exists and later shifts to the identification that a particular
provider is an appropriate agent to resolve the need or assist in efforts to deal
with the concern (Patterson et al., 1992). Families identified as not yet ready to
fully participate in home-based interventions might not have identified the target
areas of the intervention as focal concerns or might not interpret the provider as
an appropriate or effective source of assistance.

Understanding how families begin to identify issues as concerns that
necessitate some action may be central to understanding effective engagement

and use of home-based services. Investigating individual differences in self-

described needs is particularly essential for explaining participation in preventive
interventions (Stein, Bauman, & lreys, 1991). By definition, prevention programs
attempt to ameliorate expected negative outcomes before they have fully
developed; that is, before the severity of a negative consequence is fully
apparent. People who are more proactive in anticipating issues in their lives that
might compromise future success and who identify these as areas of need could
be more likely to use support services consistently than families whose
perceptions of needs reflect more of a reaction to experiencing negative
consequences. Individuals who reflect on and articulate their needs more
effectively also may elicit more effective assistance from family-centered support
services than those with more difficulty identifying and disclosing their concerns.
One research team has begun to theorize that individual differences in
reporting needs may be linked to patterns of service utilization. Daro and Gelles
(1992) described three types of families at risk for child maltreatment: consumer
families, dependent families, and resistant families (Daro & Gelles, 1992).
Consumer families generally recognize areas where they might benefit from help
and seek assistance for these concerns from their informal social networks and
community programs. They often are responsive to public awareness efforts and
traditional service delivery mechanisms. In contrast, resistant families are not so
responsive to interventions. Typically, resistant families do not self-identify
personal needs or concerns and they do not seek help from support services.
Moreover, they often exhibit serious functional problems and are not receptive to

intervention efforts, even when service providers successfully contact them. The

third group of families identified by Daro and Gelles deserves particular attention
as a cluster of individuals who do not initially seek services, but may be very
responsive to appropriate intervention. Dependent families tend not to
independently recognize specific needs for assistance and do not effectively
access services that are available for support. This group of families requires
more extensive outreach to locate, but often responds well to a more intensive
service delivery style that relates parenting strategies in concrete ways to their
particular situations. So, the two groups of high-need families most likely to be
missed by service providers share the characteristic of not self-identifying needs
in areas where programs provide assistance.

Understanding the abilities of individuals to recognize and describe their
personal needs and concerns may have important implications for identifying
individuals at high risk for program attrition and for improving service delivery
practices (Navaie-Waliser et al., 2000; Sontag & Schacht, 1993). While the
importance of self-recognition of needs has not been fully investigated, difficulties
with need recognition may be associated with parenting difficulties and poorer
child development outcomes. For instance, one study of families mandated to
receive child protective services found a very high proportion of clients reported a
discrepancy between the needs they perceived that they and their children had
relative to the needs reported by CPS workers (Faver et al., 1999).

The current study investigated individual differences in reporting perceived
needs within a population of low-income pregnant women. Of particular interest

were observed differences between the extent of concern identified by

 

participants and those that might have been expected by providers based on
more objective characteristics of each participant’s self-described situation at the
outset of intervention. I sought to understand what influenced individual
differences in reported needs and what the significance of reported needs was
for service delivery. To this end, the investigation was conducted in two stages.
The background, methods, and results of each stage comprise a chapter in the
dissertation document that follows. In the first stage, analyses investigated
factors expected to influence individual differences in low-income pregnant
women’s reported needs during the prenatal period. Thus, chapter one outlines
a conceptual model predicting reported needs and examines the efficacy of using
this model to differentiate the need reporting behaviors of pregnant women in the
Michigan Maternal Health Services Study. Then, in the second stage I examined
how differences in these reported needs influenced postnatal maternal/infant
support service utilization, and indirectly, individual differences in parenting
interactions and infant development. Chapter two describes the conceptual
model of expected postnatal relationships and investigates how effectively this
model predicted intervention outcomes for a subset of families in the Michigan
Maternal Health Services study. The final chapter of the dissertation summarizes
the overall findings from both stages in the study and discusses implications for

further research and intervention practice.

CHAPTER 1

PREDICTING REPORTED NEEDS IN LOW-INCOME PREGNANT WOMEN

INTRODUCTION

Interventions Targeting Low-Income Preqna_nt Women.

Considerable research has described inequalities in both the physical and
mental health status of women across socioeconomic groups (Adler, Marmot,
McEwen, & Stewart, 1999; Brooks-Gunn, Leventhal, & Duncan, 2000; Kawachi,
Kennedy, Gupta, & Prothrow-Stith, 1999). Women living in poverty are
disproportionately exposed to physical, social, and psychological stressors
(Gorski, 1998; Lynch, Kaplan, & Shema, 1997; McLeod & Nonnemaker, 1999).
Stressors affect both maternal and infant health and well-being through
biological, psychological, and social pathways (Adler et al., 1999; Ennis, Hobfoll,
& Schroder, 2000). Stress elicits biological responses (neuroendocrine, immune,
increased preterm labor, and less optimal birth outcomes among pregnant
women); mental health responses (depression, negative coping); practice of risk
behaviors (smoking, weight gain, drug and alcohol abuse); underutilization of
resources and services; and compromises individual resources to respond to
daily challenges of parenting in a nurturant manner that promotes positive child
development (Baum & Posluszny, 1999; Crnic & Acevedo, 1995; Eisenstadt &
Powell, 1987; McLoyd, 1995; Rhodes & Jason, 1990; Taylor, Repetti, & Seeman,
1997; Turner & Lloyd, 1999). Over time, stress responses produce cumulative

effects that are associated with negative social consequences across multiple

domains, including: interpersonal relationships, nurturant parenting, productive
employment, participation in maternal-child prevention/intervention programs,
and child development (McEwen, 1998; McLoyd, 1990; Seguin, Potvin, St. Denis,
& Loiselle, 1995).

Targeting pregnancy as a time to initiate supportive interventions that
reduce stress and stress responses may offer maximal benefits to improving the
health and development of both mother and child. Pregnancy is a period when
physicians and other service providers often have contact with low-income
women who otherwise underutilize health care and social service systems
(Zayas & Busch-Rossnagel, 1992). Relationship-based interventions may
establish positive alliances that promote effective health care use and provide a
foundation for positive parenting after delivery (Barnard & Morisset, 1995; Emde,
Korfmacher, & Kubicek, 2000; Heinicke et al., 1999). In addition, as women
experience key transitions during pregnancy, many reassess their life
experiences and are receptive to new information and ideas from trusted sources
that shape their health and parenting behaviors (Brazelton, 1992). Helping
pregnant women cope effectively with poverty stressors also may reduce fetal
exposure to neuroendocrine stress responses and minimize maternal risky
behaviors (e.g., smoking, other substance use) associated with poorer birth and
child developmental outcomes (Gonzalez-Calvo, Jackson, Hansford, &
Woodman, 1998; Paarlberg, Vingerhoets, Passchier, Dekker, & Van Geijn, 1995;
Thoits, 1995).

To capitalize on the potential benefits of promoting positive behavioral and
life changes during pregnancy, a number of communities have implemented
intervention programs that target women identified as “at risk”. Many of these
programs have incorporated a client-centered home-visiting approach that
supplements existing obstetric and pediatric services (Grant, Ernst, &
Streissguth, 1999; Margolis et al., 1996; Poland, Giblin, Waller, & Bayer, 1991).
Providing home-based services allows professionals greater familiarity with the
ecology producing stress and permits the opportunity to deliver an individualized
intervention tailored to the needs reported by each client (Graham et al., 1997;
Hardy-Brown, Miller, Dean, Carrasco, & Thompson, 1987; Kitzman, Yoos, Cole,
Korfmacher, & Hanks, 1997). These facets of the intervention are believed to be
critical for promoting positive changes with high-risk individuals (Emde et al.,
2000; Grant et al., 1999; Kitzman, Yoos et al., 1997).

The effectiveness of home-visiting interventions in achieving positive
outcomes for high-risk clients is currently a subject of considerable debate. Over
the last two decades, tested home-visiting models have demonstrated positive
effects in reducing risky behaviors during pregnancy, enhancing maternal
psychosocial state, improving appropriate service utilization, and promoting
nurturant parenting behaviors (Benasich, Brooks-Gunn, & Clewell, 1992; Ciliska
et al., 1996; Daro & Harding, 1999; Heinicke et al., 1999; Marcenko & Spence,
1994; Olds et al., 1997; Olds et al., 1999; Olds & Kitzman, 1990; Tableman,
1999-2000a, 1999-2000d). Yet, these effects have not been consistently

replicated across intervention programs, populations served, or shifts in service

provision staff (Gomby et al., 1999). There are numerous layers of complexity
that contribute to the inconsistent findings. Some factors include: variation in
intervention content and objectives, intervention dosage (both intended and
actual as well as variations in timing of interventions provided), provider type
(including variations in background, training, support, supervision, and success in
the specific therapeutic relationship established with each client), alternative
interventions received by “control” groups, examination of outcomes in an “intent
to treat” versus “treated” group, and participant attrition prior to documentation of
outcomes (Benasich et al., 1992; Berlin, O'Neal, & Brooks-Gunn, 1998;
Korfmacher, 1998; Olds, 1988; Powell & Grantham-McGregor, 1989; Stein et al.,
1991; Tableman, 1999—2000b, 1999-2000c; Tableman & Sorenson, 1999-2000;
Twohy & Reif, 1997). Yet, one factor that has received little attention in the
literature is the concept of need in study sample selection and as an influence on
individuals’ voluntary participation in programs. Discussing different definitions of
need and investigating relationships between need and intervention effectiveness
may promote a richer understanding of the impact of home-visiting on individuals.

Two Approaches to Defining High Need GI‘OUfi.

Both in empirical literature and in service delivery practice, two different
methods have been employed to identify individuals at high risk for negative
outcomes who might benefit from supportive interventions. These methods
underlie what Gifford refers to in the anthropological literature on risk as the
difference between “a technical objective or scientific dimension and socially

experienced or lived dimension” (Gifford, 1986, p. 215). That is, the existence of

 

needs has been established either through the presence or absence of a set of
objective characteristics associated with a high probability of experiencing needs
or through stated individual perceptions that such a need exists in one’s personal
life experience.

In the design of programs or the development of policies, a target
population that meets specific criteria associated with negative outcomes
typically is specified (Flaskerud & Winslow, 1998; Landy, 2000). For instance,
many home-visiting programs have been developed to assist low-income,
pregnant women because they are presumed to have a set of contextual, social,
and health factors that place them at risk for poor pregnancy outcomes as well as
parenting and life course difficulties that might compromise the opportunities for
their children’s future successes (Binsacca, Ellis, Martin, & Petitti, 1987; Weiss &
Jacobs, 1988; Zimmerman, 1999). So, in this example, the target population with
needs is defined by the objective circumstances related to income and
pregnancy rather than by any individual perception or concern about her life
circumstances on the part of the mother. Often, then, the effectiveness of
programs or policies is evaluated based upon changes observed among a
representative sample of individuals possessing the specified need criteria.

While this approach continues to impact service delivery, over the last
twenty years service providers increasingly have appreciated the significance of
the experiential dimensions of client need as critical foundations for treatment
planning. Many practitioners and programs have adopted family-centered care

intervention approaches (Nelson & Allen, 1995; Riessman, 1990). A central

 

tenet of family-centered care is that interventions are individualized to respond to
the specific perceived needs articulated by families rather than providing a similar
treatment model to all families who share a set of objective characteristics
(Dunst, Johanson, & Trivette, 1991; Keen-Payne & Bond, 1997). The family and
service provider form a partnership in which they strive to achieve family-defined
goals (Thomas, Benham, & Guskin, 2000). Since considerable variation exists in
the needs of pregnant women during their transitions to parenthood and
experiences with early parenting (Flanagan, 1998; Halpern, 1993), client— and/or
family— centered care approaches have been integrated into many interventions
delivered during pregnancy and early parenting (McDonough, 2000). This
practice encourages providers to individualize interventions based on clients’
self-reported needs and provides a more targeted intervention that considers
each client’s past experiences, confidence related to caring for children, personal
skills and strengths, competing environmental stressors, and the availability of
social support to meet the client’s needs in other ways (Kitzman, Yoos et al.,
1997)

The increased personal relevance of information shared in family-centered
care interventions may enhance the educational effectiveness of the treatment
messages (Korsch, 1984; Williams & Meredith, 1984). These ideas are
consistent with health educators’ recommendations to tailor messages to
address specific individual concerns of learners in order to enhance retention of
information and behavioral compliance (Korsch, 1984; Vivian & Wilcox, 2000).

Research also has found that individual changes in attitudes and subsequent

behavior are more probable when information is perceived as highly relevant
(Covington, 2000; Vivian & Wilcox, 2000).

Indeed, studies evaluating family-centered interventions have documented
positive outcomes when these approaches are incorporated into practice across
a variety of service domains (Bradley, 1983; Marcenko & Smith, 1992; Weiss &
Jacobs, 1988; Weissbourd & Kagan, 1989). Participating families frequently
report feeling more empowered using services (Trivette, Dunst, & Hamby, 1996)
and display improvements in family functioning and appropriate parenting (as
indicated by reduced out of home placements for children) as compared to
families participating in more traditional intervention models (Scannapieco;
1994).

However, embedded in the use of a family-centered care model is the
assumption that individuals are equally capable and equally willing to perceive
and describe their needs (across all domains) to service providers. Yet, little
research has described individual differences in reporting tendencies or
examined how these differences might shape program outcomes. But, if a client-
centered intervention is tailored to address an individual’s reported needs,
whereas the program effectiveness is evaluated based on the changes in
expected needs observed in more objective criteria, one might expect individual
differences in reporting tendencies to have a critical influence on program

effectiveness.

_C_l§rifving Terminology.

Throughout this document, reference is repeatedly made to three distinct
types of needs: expected (sometimes also referred to as objective) needs,
perceived needs, and reported needs. Most psychological literature does not
clearly distinguish among these three conceptsz. However, within this document,
the terms expected needs, perceived needs, and reported needs will be defined
in the manners described below.

Expected needs reflect some assessment of risk or presumed need by
someone other than the individual believed to be in need. These needs are often
defined by the presence of one or more conditions, characteristics, or
circumstances that have been associated with some negative outcome in similar
groups of individuals. The association between the circumstance and negative
outcome forms the basis for the judgment that the individual does possess a
need to modify the characteristic or circumstance to something associated with
more positive outcomes. Expected needs are established without regard to
whether or not individuals actually believe that they have a need to alter the
given circumstance. Hence, the operational definition of expected needs in this
document is quite different from terminology in the psychological literature related
to expectancies. Whereas literature on expectancies focuses on how individuals’
own expectations influence their subsequent outcomes, the expected needs in

this study are formulated from the expectations of a third party (e.g., service

 

2 One reference to a somewhat similar distinction between the terms needs, wants, and
demands (roughly corresponding to expected needs, perceived needs, and intentions to act on
needs) was noted in the dental literature. See Davis (1982) for more information.

provider or researcher) who believes that the individual’s specific characteristic or
circumstance will impact her outcome.

In contrast to expected needs, perceived needs are defined by the
individual’s subjective belief that a negative condition or circumstance must be
modified to improve well-being or to survive. By definition, all perceived needs
require some subjective individual awareness. Most studies that examine
perceived needs choose to operationalize them by documenting the needs that
individuals report (e.g., Rawlins, 1991). However, it seems at least plausible that
people might perceive specific personal needs, but not disclose them to others.
Thus, the language of this document differentiates between perceived and
reported needs. Reported needs include the needs that an individual discloses
to another person. In most cases, reported needs are perceived needs that are
subsequently disclosed. However, individuals could also report needs that are
inconsistent with their perceptions in efforts to manipulate the behavior of the
listener or for social desirability purposes. Given the potential discrepancies that
might occur, it is important to distinguish between perceived and reported needs.

Reporting Perceived Needs.

Relatively little work has examined influences on individual perceptions of
needs and characteristic differences in the ways these are described to others.
(Chung, 1993; Fever at al., 1999). Most research related to perceived needs
provides descriptive information on reported needs at the group level. Studies
often provide information about which needs certain groups of individuals in a

target community prioritize as particularly important or identify as unmet by

existing service systems. For instance, research has described adolescent,
African-American mothers’ reported unmet needs for parenting assistance
(Barnett, 1987); the self-described needs of women who received services from a
maternity home and adoption agency during their pregnancies (Keen-Payne &
Bond, 1997), the multidimensional needs of families who survived residential
fires (Keane, Brennan, & Pickett, 2000), the concerns of mothers with infants in a
neonatal intensive care unit (Bolton, Chalmers, Cooper, & Wainer, 1993),
women’s experiences and needs following interactions with a metropolitan health
care system (Taylor & Dower, 1997), and the priority needs described by
homeless adults (Acosta & Toro, 2000) and homeless mothers (Kissman, 1999).
Descriptive studies like these offer information about whether or not practitioners
are “on target” when they identify focal concerns that generally occur within the
entire targeted group. But, such work does little either to understand mismatches
at an individual level or to consider which people might have greater difficulties
articulating their needs and how these individual differences might impact the
effectiveness of the interventions that are delivered. However, it is interesting to
note that in several of these descriptive studies, a subgroup of participants is
listed as having “unspecified” needs wherein people identify needing general
help without clearly articulating the types of help desired (Barnett, 1987; Keane et
al., 2000; Penn, Levy, & Penn, 1986). Although little attention is directed toward
people with unspecified needs in the literature, the repeated presence of such a
subgroup suggests that individual differences do exist in people’s abilities to

report their needs effectively.

l6

Another common line of research into needs considers the extent to which
providers prioritize the needs of the target community in similar ways to the
priorities offered by the members of that community. A number of investigations
have shown that service providers and clients often report different priority
rankings and occasionally even domains of individual client needs. Such
discrepancies have been found in many different types of service delivery,
including perceived child needs among families involved with child protective
services (Faver et al., 1999), desired health promotion activities by low-income
seniors (Bertera, 1999), reported needs for information about sexuality from
teenagers (Cairns, Collins, & Hiebert, 1994), perceived service needs of low-
income, urban Black women (Penn et al., 1986), desired vocational training
information by young adults (Cherry & Gear, 1987), needs identified as serious
by homeless individuals (North 8 Smith, 1993), perceptions of high and low
importance components of parenting support services received by high risk
mothers (Pharis & Levin, 1991), and in the needs identified during a community-
wide needs assessment to determine priority areas for economic and social
development in an underprivileged community (Eng & Blanchard, 1990).

Careful analysis of results from studies with provider-client discrepancies
in perceptions of needs reveals a trend such that clients tend to identify more
concrete needs for basic resources as higher priorities, whereas service
providers seem to identify more abstract needs for intrapersonal change or steps
toward self-actualization more readily as priority needs. Yet, it is important to

remember that provider priority rankings are partly a product of clinical judgment

based on their interactions with the individual, family, or community. Providers’
assessments of clients’ needs may be based partly on nonverbal signals or
comments interpreted as inferring needs rather than on those needs overtly
stated by the client. Differences in perceptions of need between providers and
clients also could be driven by different goals for the client as much as by
differences in the perception of need (Cairns et al., 1994). Hence, there are a
variety of reasons why investigations of discrepancies between clients’ reported
needs and providers’ may not be the most effective way to investigate if people
tend to disclose fewer self-described needs than would be expected based on
knowledge of more objective characteristics of the client and her situation.
Careful examination of what individuals actually report as their needs to listeners
who are not making clinical interpretations and who are not personally providing
services to the person may be an important first step to analyzing actual client
differences in reporting patterns.

Conceptual Model for Predictim Reported Needs i_n Low-Income Pregm
Wm-

Figure 1 depicts my hypothesized model for how key factors predict
reported needs among low-income, pregnant women. This model, like family-
centered care practice, necessarily confounds the self-perception of needs with
whatever information the participant is willing to actually disclose. It is possible,
and even likely, that individuals have some concerns or perceive some personal
needs that they do not openly report. Yet, undisclosed needs cannot effectively

drive the nature of services that participants receive in family-centered care.

Undisclosed concerns also are not easily assessed, particularly since individuals
may be at various stages of recognition that a given experience or condition
constitutes a need or concern for them. For these reasons, the focus of the
model is on factors that predict the needs that individuals actually disclose.

At the core of the model is the assumption that expected needs generated
from descriptive characteristics of the client’s situation will have a direct positive
relationship to the client’s reported needs. This represents the path of
assumptions inherent in client-centered care models that target individuals with a
specific set of objective circumstances and provide individualized intervention
based on areas the client identifies as concerns. By individualizing treatment
protocols to the reported client needs, providers expect to positively impact some
objective indicator that formed the basis for that reported concern. Prior work
has found reported perceptions are often based on realistic appraisals of actual
circumstances (Cherry & Gear, 1987). However, evidence suggests that a
number of other factors also may influence reported needs. Predictions about
these factors are outlined below.

The Mot of State on Reported Needs. Considerable research has
investigated factors that contribute to heightened stress and to the expression of
depressive symptoms. Stress and depression are biopsychosocial responses to
events or conditions (Brooks-Gunn et al., 2000; DunkeI-Schetter, Gurung, Lobel,
& Wadhwa, 2001; Kessler, 1997). Events and conditions found to be positively
associated with greater stress and depression include many of the types of

factors embedded in the expected need index for this study (e.g., physical

abuse, unwanted pregnancy, unemployment, food insufficiency, housing
instability) (Barnfather & Ronis, 2000; Brooks-Gunn et al., 2000; Kessler, 1997).
In fact, such strong relationships have been reported between objective factors
and stress that many investigators choose to operationalize stress by using the
presence or absence of negative life events in place of perceived stress (e.g.,
Brooke, Anderson, Bland, Peacock, & Stewart, 1989; Ferketich & Mercer, 1990).
High levels of perceived stress also have been associated with increased
depressive symptoms among low-income pregnant women (Ennis et al., 2000),
mothers of young children (Hall, 1990; Hall, Williams, & Greenberg, 1985), and
with adolescent mothers (Barnet, Joffe, Duggan, Wilson, & Repke, 1996).

While much work examines the causes and consequences of stress and
depression, research has not directly investigated how stress and depression
impact individuals’ self-reporting of their needs. However, a few studies provide
clues about how these states might influence reporting needs. For instance,
increased levels of stress have been found to promote service utilization initially,
but interfere with consistent participation and engagement when observed
midway through the course of intervention (Eisenstadt & Powell, 1987). If
perceived stress acts as a trigger to motivate individuals to participate in
services, it might also increase individual tendencies to reflect about the origins
of the stress and to report existing needs. This hypothesis would be consistent
with coping literature where the identification of an event as a perceived threat is
important in the initiation of problem-solving and coping behavior (Gravida-Payne

& Stoneman, 1997; McNett, 1987; Thoits, 1995).

20

On the other hand, existing evidence would suggest that depressive
symptoms could interfere with a realistic appraisal of and reporting of needs.
Affect influences the amount of attention available for information acquisition,
memory retrieval or storage, and appraisal processes (Fiske & Pavelchak, 1986).
It also has been found to focus these processes on salient environmental factors,
past events, and response options (Bugental & Goodnow, 1998). Specifically,
one unique function of sadness is to slow down cognitive and motor systems
(lzard, 1993). When sadness occurs for short periods of time at low levels, this
mood may be adaptive by promoting slower and more deliberate scrutiny of
oneself and one’s behavior. However, when sadness is less effectively
regulated, maladaptive behavior may result, including problems with emotion-
cognition-action patterns often observed in depression (lzard, 1993). In fact,
distorted, negative views of the self and a sense of hopelessness about one’s '
ability to impact the future are characteristic of individuals with depression
(Rosenhan & Seligman, 1995). So, it is not surprising that individuals with
depressive symptoms are less likely to utilize services effectively (Rhodes,
1993). Given the described impact of depressive symptoms on functioning, it
would also be expected that individuals exhibiting these symptoms might be less
effective in reporting their needs than those without depressive symptoms. The
presence of depressive symptoms may shift the focus of accurate self-
assessment and identification of needs as well as interfere with the initiative

required to seek help from service providers and report needs accurately.

2]

Relationships Between Perceived A_bilitv to Gone and Reported Needs.
The next cluster of factors expected to predict reported needs are individual and
social constructs that would influence the individual’s cognitive appraisal of her
ability to cope with her needs. Whether or not individuals consider that a given
situation constitutes a need may be influenced by their perceptions of their
options and prospects for managing the event, or what Lazarus and his
colleagues referred to as “secondary appraisal” (Lazarus, 1966; Lazarus &
Folkman, 1984). The model in Figure 1 includes a construct of perceived
individual control as well as support from others that might influence appraisal of
need.

Weisz (1982) describes the construct of personal control as consisting of
the dual elements of competence and contingency. To feel a sense of control,
individuals must believe both that they are capable of succeeding in actions they
undertake (competence) and that their actions will have an impact on the
outcome of a given situation (contingency) (Weisz, 1982). More recent work has
elaborated on these notions to view competence (self-efficacy), contingency
(personal mastery, locus of control), and motivation to act embedded within a
contextual frame as components of psychological empowerment (Zimmerman,
1990a; Zimmerman & Rappaport, 1988). While little work has directly addressed
relationships between perceived control or empowerment and reported needs,
research has shown that individuals with higher measures of empowerment

characteristics are more likely to be active participants in community

22

organizations and activities3 (Wandersman & Giamartino, 1980; Zimmerman &
Rappaport, 1988). Case studies have also demonstrated that individuals who
possess empowerment characteristics are able to effectively negotiate an
unsupportive, bureaucratic health care system to receive desired services
(Birenbaum-Carmeli & Carmeli, 1996).

However, this empowerment research does not definitively suggest a
relationship between personal control or empowerment and reported needs.
High levels of personal control could increase reported needs because increased
awareness of one’s situation may accompany the transformation in belief that
one can alter his or her situation (Koch, Lewis, & Quinones, 1998). An example
might include a survivor of domestic abuse who begins to realize and assert her
personal needs throughout the course of treatment. In contrast, high levels of
personal control could be related to fewer reported needs. Individuals with
greater personal control might feel capable of handling issues independently, and
thus no longer label and report them as needs. This prediction is consistent with
extensive research using the Health Belief Model in which individuals with a
perceived threat (i.e., knowledge about susceptibility to an illness combined with
awareness of severity of condition), weigh their health behavior decisions based
on the perceived benefits of their action, (e.g., Can they prevent it or ameliorate it
through their actions? Will their actions have an effect?), the perceived barriers to

action (e.g., Do they have sufficient knowledge or competence to achieve the

 

3 Later work supported the hypothesis that participation in community activities provides
opportunities to enhance perceived control that helps individuals cope with stress (Zimmerman,
1990b). The participation allows people to engage in the process of learned hopefulness by
utilizing their skills successfully. Thus, it contributes to psychological empowerment. However, in
the cross-sectional study supporting this assertion, the reverse path from empowerment to
participation was not tested and therefore cannot be fully dismissed.

23

intended result if they take action?), and contextual cues that trigger the timing of
action (Rosenstock, 1966; Strecher, Champion, & Rosenstock, 1997). In this
model, greater competence and contingency would lead to increased
independent health behavior action. Though reported needs are not specifically
included in the Health Belief Model, it would appear that people with high
competence and contingency would report fewer needs for topics they perceive
can be handled independently (e.g., parenting, smoking cessation, gathering
necessary baby supplies) because they are likely to engage in those health
behaviors without need for assistance. However, individuals with high personal
control would report greater needs for topics where the health behavior action
necessitates a professional’s involvement (e.g., HIV/AIDS testing, use of prenatal
care, initiation of a chemotherapy regimen). Given that the areas covered in the
need index are home-based prevention and intervention issues, most appear to
correspond to the former situation, and thus higher personal control would be
associated with a reduction in reported needs.

In summary, reasonable explanations exist for both positive
(empowerment raises awareness, thereby enhancing perceived and
subsequently reported needs) and negative (empowerment enhances beliefs
about independent capabilities to handle events without external assistance,
thereby reducing reported needs) associations between personal control and
reported needs. However, the extensive research supporting other well-defined
components of the Health Belief Model lends somewhat greater support to

predict a negative association between personal control and reported needs

24

(Becker, 1974; Burak & Meyer, 1997; Maddux & DuCharme, 1997; Tiedje,
Kingry, & Stommel, 1992). In addition, two studies about parenting support
service utilization lend further weight to this hypothesis. In two diverse
populations of mothers, both the nature and number of nurse home visits in a
parenting support program were moderated by levels of mastery (in a rural,
European American sample) and psychological resources indicating competence
(in an urban, African-American sample) (Olds & Korfmacher, 1998). Individuals
with higher levels of control or competence generally received fewer home visits
from nurses“. This lessened intensity of service delivery would be expected with
clients who report fewer needs and seem to be coping well independently.
Previous work has established that personal control may be related to
reported needs through indirect as well as direct pathways. Considerable work
suggests that personal resources such as self-esteem and mastery can reduce
the severity of appraisals about or responses to stressors (Lazarus 8. Folkman,
1984; Pearlin, 1999; Pearlin, Lieberman, Menaghan, & Mullan, 1981). For
instance, Ennis, Hobfoll, and Schroder (2000) found that among European
American women, high levels of mastery buffered the perceived stress
associated with acute loss of material resources. Turner and Noh (1983) also
found that low socio-economic class mothers with high levels of control and high
levels of social support experienced less psychological distress than women with

other levels of support and control. Other studies have probed beyond distress

 

‘ This study did reveal a slight increase in number of home visits among those at the highest
levels of personal resources. However, despite this curvilinear relationship, the overall number of
visits received remained lower for those with high as compared to low psychological resources
(Olds & Korfmacher, 1998).

25

and found that high levels of personal control also reduce behavioral responses
to stressors in the form of depressive symptoms (Hiroto & Seligman, 1975; Miller
& Seligman, 1975; Seligman, 1975). Personal control mediates outcomes of
adverse events by promoting the initiation of and persistence of coping efforts to
resolve problematic circumstances (Ross & Mirowsky, 1989). High levels of
personal control also have been associated with greater receipt of social support
(Eckenrode, 1983). Thus, people with higher levels of individual resources tend
to garner greater social resources in their environments as well. Through these
links reducing stress responses (especially depressive symptoms) and
enhancing social support, personal control was expected to have an indirect as
well as a direct impact on the needs that individuals report.

Although the specific relationships between social support and reported
needs have not been investigated, a negative relationship between social
support and reported need is hypothesized. Higher levels of social support
increase the informal sources available for individuals to manage difficult
situations (Crockenberg, 1988). Like perceived control, the presence of social
support might reduce individual tendencies to define a given situation as a true
need or concern. Therefore, social support might reduce the likelihood that
people would report their needs even if they perceived that unresolved issues in
these areas did exist. Some support for these ideas is present in social support
research showing that the presence of informal support assists people and
promotes healthier outcomes in difficult circumstances. For instance, evidence

suggests that social support is associated with increased health-related quality of

26

life scores (Achat et al., 1998), reduced stress, anxiety, and depression during
pregnancy (Barnet et al., 1996; Dunkel-Schetter, Sagrestano, Feldman, &
Killingsworth, 1996), increased reliance on problem-focused coping strategies
(Gravida-Payne & Stoneman, 1997), and improved quality of parenting (Simons
& Johnson, 1996). Positive outcomes like these would be expected among
individuals who can manage their concerns more effectively than among people
likely to need formal assistance with their concerns. So, less reported need is
expected among individuals with higher levels of social support.

Self-Disclosure of Perceived Needs. In predicting people’s reported
needs, it is important to consider their willingness to disclose the needs they
perceive to the listener. Participants may be fully aware that they possess a
given set of needs, but choose not to reveal that information to the service
providers or research interviewers.

Indeed, evidence suggests that there are individual differences in people’s
willingness to disclose personal information. Factors found to influence comfort
with, likelihood of, and amount of self-disclosure include past experiences with
personal self-disclosure (Vaux, Burda, & Stewart, 1986), cultural views about
self-disclosure (Warda, 2000), level of emotional expressiveness (Kunkle &
Gerrity, 1997), the nature of the information to be shared (Burnard & Morrison,
1992; Jourard, 1971), the relationship of the listener to the individual (Ahluwalia,
Dodds, & Baligh, 1998; Derlega, Metts, Petronio, & Margulis, 1993),
costs/threat/risk associated with sharing versus not sharing the information

(McNett, 1987; Wheeless & Grotz, 1977), the extent to which others have offered

27

similar disclosures (Jourard, 1971), and the surrounding context (Kaniasty,
2000).

Personal disclosure of needs requires identifying areas where the
individual is lacking some necessary knowledge, object, or capacity. By
acknowledging the presence of a deficit, individuals expose themselves to a
certain level of vulnerability (Riessman, 1990). Thus, in the context of reporting
perceived needs in a one-on-one discussion with a professional research
interviewer, three factors particularly related to comfort exposing oneself to this
vulnerability are hypothesized to moderate the relationship between perception of
and reporting the presence of a given needs. These include: comfort with
personal disclosure, negative network orientation, and prior experience with child
protective services.

The general tendency toward comfort expressing personal matters with
others is hypothesized to increase the likelihood of choosing to report needs that
are perceived, thus increasing reported needs. The indicator of disclosure
comfort was selected as a general assessment of individual tendencies to be
reticent or expressive in disclosure. Disclosure comfort represents the
coalescence of many cultural, cognitive, and personality factors that affect
disclosure, rather than one specific influence. Although little psychological
research looks specifically at disclosure of individual needs, a positive
association between disclosure comfort and reported needs would help explain

the findings of a small study of low-income parents. Eisenstadt (1987) found that

 

5 This is the underlying process expected to influence outcomes observed. However, because
perceived and reported needs are confounded in this study, the three factors are hypothesized to
directly affect reported needs.

28

reported needs were only effective predictors of program participation among
those who were highly expressive about their concerns; the participation of more
controlled individuals was better predicted by another, more objective, set of
factors about the individual’s situation (Eisenstadt & Powell, 1987). If highly
expressive individuals were more likely to report their needs, then the
relationship between program participation and expressiveness with these
families is logical. In contrast, if less expressive individuals reported fewer of
their needs, then program planning on this basis would be more difficult. Thus,
with less expressive families, it seems that service providers may have based the
intensity of the intervention on more objective client factors instead of focusing
exclusively on reported needs.

The second factor predicted to influence disclosure tendencies is negative
network orientation. Negative network orientation extends beyond simple
comfort with self-expression to convey “a set of expectations, or beliefs that it is
inadvisable, impossible, useless, or potentially dangerous to draw on network
resources” (Tolsdorf, 1976, p.160). Individuals possessing this set of beliefs
would find reporting perceived needs to others to be, at best, a waste of time, or
at worst, actually a dangerous activity. Evidence suggests that people with more
negative network orientations have smaller actual and perceived social networks
and less expressiveness about personal and private affairs (Vaux et al., 1986).
Thus, one would expect that negative network orientation would negatively
moderate whether perceived needs are actually reported. Hence, people with

negative network orientations would be expected to report fewer needs, both as

29

a direct result of their cognitive expectancies and mediated through a reduced
general comfort with self-disclosure. The expectancies and discomfort with
disclosure associated with negative network orientations also would be expected
to reduce the perceived and reported availability of usable informal social
support. Indeed, past work supports smaller social networks among people with
negative network orientations (McKinlay, 1973; Tolsdorf, 1976).

The final factor expected to negatively moderate whether perceived needs
are reported is past involvement with child protective services (CPS) as a parent.
History of involvement with CPS may impede full reporting of needs because
families often view that a stigma is associated with reporting the presence of
needs and fear that reporting needs might provide justification for removal of their
children from their homes (Ahluwalia et al., 1998; Faver et al., 1999). This
stigma is likely to be particularly pronounced among families who have received
services from CPS providers who have not adopted client-centered service
delivery approaches. Data collection in the current investigation utilized a group
of community health research interviewers that shared similar backgrounds to the
participants and did not carry the professional “nurse” or “social worker” titles that
are frequently associated with CPS removal experiences. Typically individual
rapport was quickly established and nurtured through repeated interviews by the
same worker at each time point. However, even with these efforts to provide
optimal circumstances for disclosure, it was possible and even likely that a past
investigation by Child Protective Services agency representatives in the county

had reduced participant comfort with disclosure and decreased her actual

3O

reporting of needs to the interviewer. Thus, prior involvement with CPS as a
parent was expected to have a direct negative influence on reporting needs as
well as an indirect effect by reducing comfort with self-disclosure.

Summapy. Very little research specifically could guide predictions about
influences on mothers’ reporting of their needs. However, a variety of more
peripherally related literatures supported the development of a model to explain
the complexity of observed disclosure about needs. Reported needs were
hypothesized to be influenced not only by expected needs, but also by factors
that influence general self-disclosure tendencies, the individual’s state, and the
personal and social resources that were available to manage situations without
formal assistance. Testing the usefulness of this model in predicting reported
needs of low-income pregnant women was expected to provide valuable
information about the need identification and the disclosure process. Thus, the
aim of the first stage of the current study was to examine whether an observed
set of data taken from low-income pregnant women fit the hypothesized set of
relationships predicting reported needs as presented in Figure 1. More specific
information about the sample providing the data, the measures, analysis

methods and findings follows.

31

METHODS

Sample: Michigan Maternal Health Services Studv Recruitment Sample
and Primary Studv Desjgg

This dissertation study was undertaken using secondary analysis from the
existing Michigan Maternal Health Services Study data sources. The Maternal
Health Services Study was a multi-site randomized controlled trial conducted to
examine the effectiveness of a home-based intervention program with a broad
community sample of Medicaid eligible pregnant women. Medicaid-eligible
participants meeting inclusion criteria and consenting to participate were
randomly assigned to either an experimental group receiving Nurse-Community
Health Worker Team care intervention or a comparison group given standard of
care nurse home-visiting intervention7.

Treatment Conditions. The broad aims of both interventions were to
improve maternal health, life course development, birth outcomes, and
subsequent parenting and infant development among families with low incomes
at high risk for psychosocial and developmental difficulties. Both treatments
incorporated a home-visiting approach and provided transportation assistance to
increase participation in prenatal care. Service providers in both groups also
were trained to assess individual client needs and use educational interventions
and referrals to community agencies to increase client capacity and improve

outcomes. However, the programs differed in who the service providers were,

 

3 Approval from institutional review boards responsible for protecting the rights of human subjects
was obtained at both Spectrum Health and Michigan State University. The associated approval
numbers were Spectrum Health Research IRB # 2001-115 and MSU UCRIHS # 01-569.

7 The provision of nurse home-visiting is a Medicaid entitlement for all low-income pregnant
women in the state of Michigan. Thus, the use of a no-treatment control group was neither
ethical, nor practically feasible for this study.

32

how the intervention was delivered, and the level of integration between the
service providers and health services in the community. More specific
information about each of these programs is written below and comparative
information is provided in Table 1.

Treatment A: Maternal and Infant Support Services Standard of
Care (SOC). The standard of care treatment in the community was a
maternal/infant support service that received referrals from community
agencies and physician's offices. In Michigan, all Medicaid-eligible
pregnant women are entitled to receive maternal support services (MSS)
from a nurse throughout pregnancy and the first two months after birth.
Families meeting basic psychosocial risk criteria are further entitled to
nurse-provided infant support services (ISS) during the remainder of the
infant's first year of life. Typically, families were connected to the services
through direct (referral to agency) or indirect (via provision of information
about MSSIISS services to the client) referrals from obstetric and pediatric
clinics. Once a provider referral or client request for services was
received, a MSSIISS nurse was assigned to the case based on the client's
geographic region and provider caseloads. The nurse conducted an
assessment visit and provided case management to eligible and
interested clients according to a jointly developed care plan. Throughout
the duration of the relationship, the nurse established contact with the
client, repeatedly assessed the health status and needs of both mother
and child, and distributed bus tokens or cab vouchers to subsidize
transportation to health clinic appointments as needed. The nurse also
answered questions and provided whatever educational interventions
were deemed appropriate around issues of nutrition, parenting, risky
behaviors, health practices, pregnancy, and infant development and care.
Nurses often focused service delivery on health concerns of the client and
her child and offered referrals to other community agencies for assistance
with basic needs or mental health services. Referrals to nutritionists and
social workers within the agency were encouraged. Assistance was
provided during home visits, with reimbursement for up to a maximum of 9
visits prenatally and 9 visits postnatally without special approval. Prior to
the beginning of the trial, an estimated 50% of Medicaid-eligible women in
the community received a mean of 4 actual visits prenatally and 5 visits
postnatally. As the research trial commenced, supervisors began
documenting the number of staff home visit contacts per day to continue
to implement the standard of care community intervention according to its
intended model.

33

Treatment B: Nurse-Community Health Worker Team Care (Nurse -
CHW Team care). This intervention approached treatment in the home
setting from a more holistic, relationship-based perspective. The
intervention was developed from an ecological stress process theoretical
framework in which a professional nurse and a community health worker
(CHW) function as a team with traditional health care providers to increase
environmental and personal resources and reduce stress and stress
responses among low-income pregnant women (for further information
about the conceptual basis of the intervention, see Roman, Lindsay,
Moore, & Barton, in preparation). Highly trained CHWs worked in teams
with nurses to establish rapport with clients by building on a common
framework of understanding through their shared backgrounds. Within the
context of this non-threatening relationship, interventions were introduced
that responded to and anticipated the unique circumstances of the client's
broad situation. Interactions focused on building capacity to identify and
use informal support networks, setting goals, building basic life skills (e.g.,
budgeting, time management), identifying choices and improving healthy
behaviors and reducing risky ones, understanding issues of pregnancy,
supporting positive parenting, bolstering knowledge of infant care and
development, and enhancing appropriate use of resources from the
community to help individuals achieve their goals. Often assistance was
geared toward shoring up basic needs and building capacities as a core
foundation that could be used in future interactions. Issues of
communication and sustaining effective interpersonal relationships often
were addressed in the context of the CHW-client relationship. Nurses
guided the intervention delivered by CHWs and met with the client
periodically to assess, monitor, and provide education and care regarding
maternal and child physical health concerns. Meetings between nurses,
CHWs, social workers, nutritionists, social service workers, and primary
care staff acted to avoid duplication of services and meet client needs in a
more integrated way. This intervention employed core principles of
relationship-based support to empower individuals and promote maternal
and child health and development. The treatment especially sought to
influence women's perceived stress, rates of depressive symptoms,
psychosocial resources (self-esteem, mastery, social support) and life
course development (education, unintended repeat pregnancies, and job
participation). A curriculum designed to address these issues in the
context of pregnancy and parenting provided a foundation for the
intervention, but it was individualized to address each client’s unique
situation and concerns. Secondary intervention benefits were expected in
the domains of parenting and child health and development.

Eligibilig. The overriding goal in selecting eligibility criteria for participants

in the primary study was to consider the effectiveness of using a mixed service

34

provider model versus a nurse home visiting model of maternal support services
within a broad community setting. Thus, a conscious effort was made to
consider the impact of treatment under usual clinical conditions (an effectiveness
study) and broadly targeted populations, rather than under ideal conditions (an
efficacy study). To this end, research participation was not limited strictly to
primiparous adolescent mothers or to mothers of certain ethnic backgrounds.
Rather, it was available to the full breadth of low-income, pregnant women
entitled to maternal support services in Michigan. Eligibility criteria (See
Appendix A) were determined based on the requirements that each woman:
could receive the randomized intervention (i.e., was entitled to it based on
Medicaid eligibility, pregnancy, and residence within Kent County, Michigan, and
did not have an ongoing relationship with standard of care group health nurse
prior to initiation in the study), was influenced by existing state requirements for
work and eligibility for entitlement benefits (i.e., is at least 16 years of age), and
had not been diagnosed or received treatment8 for any pre-existing mental health
conditions in the last two years. In addition, data collection was administered in
either Spanish or English, so individuals were required to speak one of these
languages in order to participate. Notably, eligibility was n_ot influenced by
whether or not individuals had literacy skills, a telephone, reliable transportation,

or stable housing. Indeed, significant efforts were undertaken during the

 

° Individuals were asked if they had received medications or mental health therapy for any mental
health condition in the last 2 years. Examples of conditions specifically mentioned by name in the
eligibility screening included depression, bipolar disorder, schizophrenia, delusional disorders,
multiple personalities, panic attacks, and post traumatic stress syndrome.

35

recruitment process specifically to include more difficult to serve women and to
closely mirror the target population for local health department programs.

Recruitment. Study participants were recruited from four prenatal clinic
sites that provided health care to underserved, Medicaid-eligible women of
diverse ethnicities. Clinic staff supplied the research staff with the contact
information for all individuals who had called or visited the clinic to schedule an
initial appointment for obstetric care. Identification of potential participants was
not based on any professional referrals or even attendance at the scheduled
prenatal appointment, thus reducing a common source of selection bias found in
many studies.

Recruitment into the research and subsequent data collection were
conducted by six community health research interviewers. These individuals
spanned diverse ages, ethnicities, and education levels. Many of them shared
similar past experiences to the study participants and were skillful in rapidly
establishing rapport with individuals and explaining research questions and
concepts in a non-threatening manner. Each individual received extensive
training and ongoing feedback regarding research interviewing, professional
expectations, and data collection quality improvement. The Maternal Health
Services Study team expended considerable effort to enroll and maintain the
hardest-to-reach women in the research, including those without telephones,
transportation, and who repeatedly were not present at scheduled appointments
(for more information, see Roman, Lindsay, Moore, Barton et al., in preparation).

Throughout the duration of the longitudinal study, these interviewers were

36

masked to differences between the two interventions and to knowledge about
which intervention group each client had been randomly assigned to receive.

Between January 1997 and August 1998, 613 pregnant women (mean
11.9 weeks gestation) were enrolled into the research study, completed baseline
interviews, and were randomly allocated to treatment group (for more specific
information about study design and recruitment, see Roman, Lindsay, Moore,
Barton et al., in preparation). Single blind randomization was conducted using a
computer generated randomization schedule placed in sealed opaque envelopes
in blocked groups of four, and stratified on site and level of need for social
support (high or low). Need for social support was estimated using the baseline
interview Support Functions Scale (SFS) score (Dunst & Trivette, 1988). High
need was defined as a score of greater than the median value of 60 measured
during a preliminary study with similar women. Lower need was defined as a
SFS value below this median cutoff.

_S_ai_rpple: Dissertation Subsample.

Eligibilig. Study aims were examined in the subset of Michigan Maternal
Health Services Study participants that delivered a live birth with intent to
maintain custody of the child and who completed the late pregnancy research
interview (typically at 34-38 weeks gestation). One-hundred seventy-five
individuals were excluded from the analysis sample based on these criteria,
resulting in a final sample of 438 individuals (see Appendix B for a breakdown of

cases excluded).

37

Although statistical imputation of missing data was considered to permit
inclusion of all study participants, imputation was not deemed to be conceptually
defensible when either of these eligibility conditions was not met. The imputation
of developmental information about a child that was never born would be
meaningless. The exclusion criteria based on the presence of the 34 week
research interview is a more unusual case. However, a focal interest in this
study was to examine individual differences in need reporting tendencies. The
measurement of these reporting tendencies is described in more depth later in
the methods section, but it required extensive information from both the
enrollment and the 34 week research interview. Imputation of all information
from the 34 week research interview in more than a minor subset of individuals
could substantially confound actual individual differences in reporting tendencies
with error associated with the missing data imputation process. Thus, in this
case, imputation could seriously compromise the meaningful interpretation of
analysis results. Even if the characteristics of the sample were slightly modified
by exclusion of cases, it was determined that meaningful results about reporting
tendencies in more circumscribed sample population would provide a greater
contribution to the literature than ambiguous findings in a more representative
population. That said, the exclusion criteria were not found to alter the
characteristics of the sample population too dramatically.

Sample Characteristics. Table 2 describes the demographic and
psychosocial characteristics of the study participants. Comparisons also are

provided between study participants and those excluded from the analysis

38

subsample. A significant difference between groups appeared in weeks
gestation at enrollment into the study (t (1, 611) = 3.24, p = .001). Individuals
excluded from the dissertation subsample enrolled in the study significantly
earlier in the pregnancy than those who remained in the subsample. However,
this finding of a difference in weeks gestation at study enrollment was anticipated
because of its association with incidence of the spontaneous abortion exclusion
criteria9 ("Abortion," 1993). Unexpectedly, a higher proportion of those excluded
from the subsample were unmarried (x2 (1, 612) = 4.54, p = .033). Further
investigation found this difference primarily resulted from a higher proportion of
divorced, separated, or widowed marital statuses among those participants
excluded from the secondary analyses.

Overall, the sample characteristics were similar to many groups of low-
income women targeted for services during pregnancy. Most participants were
20-24 years of age. The sample was not skewed with an unusually high
proportion of individuals at either end of the child-bearing age spectrum.
Approximately one-tenth of the group was under 18 years (11.0%) and one-tenth
was 30 years or older (9.4%). Most individuals in the sample described
themselves as Caucasian (41.6%), African-American (26.5%), or Hispanic10
(23.1%). Fifteen percent of the group had immigrated to the United States from
another country. Less than 20 percent of the sample was married, however,

nearly half (47.5 %) of the individuals in the sample lived with the father of the

 

9 More than 80% of spontaneous abortions occur in the first 12 weeks of pregnancy. Thus,
individuals excluded from the study due to the occurrence of a spontaneous abortion would
disproportionately be sampled from individuals who enrolled in the study during the first trimester.
‘° Most of the individuals who identified themselves as Hispanic were of Mexican origin.

39

baby at study enrollment. Forty-three percent of the women were employed
either part-time or full-time at the outset of the study. Forty-four percent of the
sample was primiparous; the other participants had delivered anywhere from one
to seven previous live births. Many women in the study had experienced some
form of abuse, with over half reporting physical (50.9%) or emotional (55.1%)
abuse during their lifetimes and over one-fifth surviving sexual abuse (21.1%).
These experiences, as well as ongoing financial struggles may help explain the
high incidence of depressive symptoms (56.6% over cutoff) and perceived
stress11 among this group.

Data Collection.

Participants were interviewed shortly after scheduling an appointment for
obstetric care, again at 34-38 weeks gestation (late pregnancy), and 6 weeks, 6
months, and one year after delivery. Medical records were abstracted following
each child’s birth and developmental assessments were conducted with the
infants at approximately 6 and 12 months of age. Secondary analyses primarily
utilized data from enrollment and late pregnancy to test the first conceptual
model. Specific information about measurement of key constructs follows.

M_epsurement of Vgflam.

Most instrumentation utilized for this study was selected by the principal
investigators for the Michigan Maternal Health Services Study. In the selection

process, investigators particularly attended to: a) the appropriateness of

 

1‘ Scores on the PSS instrument could range from 0 to 56. While standard deviations were
similar, the sample mean among women in this study was 3 points higher than reported in 2
samples of college students (M = 23.18 8 23.67, SD = 7.31 8. 7.79), 1 point higher than
individuals attending smoking cessation classes (M = 25.0, SD = 8.00), and 6 points higher than a
probability sample of women in the United States (M = 20.2, SD = 7.8) in the instrument validation
study (Cohen, Kamarck, & Merrnelstein, 1983; Cohen & Williamson, 1988).

40

instrument length, content, and language for a low-income, low-literacy, and
ethnically diverse population, b) the theoretical basis of instrument development
and consonance of construct measured with conceptual model being
investigated, and c) the instrument’s established reliability, validity, and
usefulness in prior research in the field. Following is a list of constructs tested in
the conceptual model predicting reported needs and the manner in which these
constructs were operationalized.

Perceived Stress. Perceived stress was measured using a 14-item, five
point likert—type scale developed by Cohen, Kamarck, and Mermelstein (1983).
The Perceived Stress Scale (PSS) was created to measure “the degree to which
situations in one’s life are appraised as stressful” with items designed to “tap the
degree to which respondents found their lives unpredictable, uncontrollable, and
overloading” (Cohen et al., 1983). Past studies reported strong correlations for
the PSS with life-event scores, depressive and physical symptomology, utilization
of health services, and smoking reduction maintenance (Cohen et al., 1983).
The PSS has proven to be a more effective predictor 'of stress-related outcomes
than life event scores (Cohen, 1986; Cohen et al., 1983; Cohen & Williamson,
1988). Regression models also have found negative events at time one to be
significantly related to time two physical symptoms, depressive symptoms, and
P38 scores (Linville, 1987). Coefficient alpha reliability of greater than .84 has
been reported across three samples and test-retest reliability after a two-day
period was .85. The PSS has been used in other studies involving pregnant

women (Lowenkron, 1999; Ludman et al., 2000) and individuals with low-

41

incomes (Lobel, Dunkel-Schetter, & Scrimshaw, 1992). Participants in this study
self—reported responses to the PSS upon enrollment to the study and again in
late pregnancy (typically 34-38 weeks gestation).

Depressive Svmptoms. The Center for Epidemiologic Studies —
Depression (CES-D) Scale was used to measure the occurrence of depressive
symptomology in the study sample. The scale was developed to examine the
presence of depressive symptoms in the general population rather than to
confirm a clinical diagnosis (Radloff, 1977). The scale includes 20 items
representing components of depressive symptoms, including: depressed mood,
feelings of guilt and worthlessness, feelings of helplessness and hopelessness,
psychomotor retardation, loss of appetite, and sleep disturbance. Individuals
reported the frequency of occurrence of each symptom in the one week prior to
the interview. These responses were systematically coded on a 4 point (0-3)
Likert-type scale. Coefficient alphas across three general population and one
psychiatric population samples were all greater than .84 (Radloff, 1977).
Coefficient alphas among pregnant women were .83 or higher (Mercer &
Ferketich, 1988). Test-retest correlations were found to be moderate (.48 - .67
from two week to two month time intervals) (Radloff, 1977). High test-retest
reliability was not expected given shifts in depressive symptoms over time in a
general population and scale measurement of symptoms across only a one-week
period. As anticipated, higher test-retest correlations were observed in individuals
with a shorter time lag between tests (Radloff, 1977). Studies have found

increases in CES-D scores when individuals reported experiencing negative life

42

events and reductions in scores following therapeutic treatment for depression
(Radloff, 1977). The CES-D has been widely used as a research tool.
Numerous other studies have established its validity or observed expected
outcomes when using this scale with large scale community studies (Craig & Van
Natta, 1976; Myers & Weissman, 1980; Turner, Lloyd, 8. Roszell, 1999), primary
care populations (Zich, Attkisson, & Greenfield, 1990), pregnant women (Collins,
Dunkel-Schetter, Lobel, & Scrimshaw, 1993; Mercer & Ferketich, 1988;
Zuckerman, Amaro, Bauchner, & Cabral, 1989), women of color (Hickey, Cliver,
Goldenberg, McNeal, & Hoffman, 1995; Melchior, Huba, Brown, & Reback, 1993;
Munoz, Gonzalez, & Starkweather, 1995; Roberts, 1980; Woods, Lentz, Mitchell,
& Oakley, 1994), and low-income populations (Collins et al., 1993; Melchior et
al., 1993). Participants in this study self-reported responses to the CES-D upon
enrollment to the study and again in late pregnancy (typically 34-38 weeks
gestation).

Social Support. The Multidimensional Scale of Perceived Social Support
(MSPSS) was used to measure subjectively assessed social support. The
MSPSS is a 12-item measure of individual perceptions of available social support
from family, friends, and significant others (Dahlem, Zimet, & Walker, 1991). The
scale has been reported to have an alpha coefficient of .88 or higher, test-retest
reliability of .85, and has demonstrated an inverse correlation with depression
scores (r= -.25) (Dahlem et al., 1991; Zimet, Dahlem, Zimet, & Farley, 1988).
The individual subscales for support of family, friends, and significant others also

have demonstrated these characteristics. Studies have replicated the original

43

factor structure of the scale and have shown that the scale scores have little
relationship to established social desirability measures (Kazarian & McCabe,
1991). The MSPSS has established its validity or displayed expected outcomes
among psychiatric outpatients, pregnant women, adolescents, battered women,
incarcerated women, and homeless populations (Barnett, Martinez, & Keyson,
1996; Canty-Mitchell & Zimet, 2000; Cecil, Stanley, Carrion, & Swann, 1995;
Eker & Arkar, 1995; Singer, Bussey, Song, & Lunghofer, 1995; Wu & Serper,
1999; Zimet, Powell, Farley, Werkman, & Berkoff, 1990). The scale was adapted
to a five point Likert-type format for effective use with the low-literacy population
in this sample. This study utilized self-reported ratings of the 4-item family
subscale of the MSPSS recorded at the late pregnancy interview.

Availability of social network. In addition to the measure of perceived
social support from family, two author-designed questions measured the
availability of social support from family and the father of the baby. At the late
pregnancy interview, individuals were asked “Thinking about the whole time from
the enrollment interview to now, how often has the father of your baby been
positively involved in your life?” Responses ranged from “Never” (5) to “Always”
(1) using a Likert-type format. The question was then repeated regarding
positive involvement of family during the time period. Responses to these
questions gauge the frequency with which network individuals were available to
offer positive support to the participant.

Received support from social network. In addition to the measures of

perceived social support from family and availability of family and the father of

44

the baby, one author-designed question measured the receipt of social support
from family and the father of the baby. At the late pregnancy interview,
individuals were asked ‘Who has been important in helping you throughout this
pregnancy?” The question was open-ended and individuals were encouraged to
identify as many people as they wished who had been important sources of
support to them throughout their pregnancy. Answers were coded as two
separate dichotomous items identifying whether (1) or not (0) the participant had
identified the father of the baby/partner as a source of help and if the participant
had identified any family member or relative in this capacity. Responses to these
questions gauge the extent to which participants viewed themselves as actually
receiving significant instrumental support from their spouses or family members
during the time period of the study.

Perceived Personal Control”. The perceived personal control construct
includes three different indicators — mastery, self-esteem, and mothering ability
esteem.

Mastery. Mastery was measured at the enrollment interview using the
seven-item Sense of Mastery Scale developed by Pearlin and his colleagues
(Pearlin et al., 1981; Pearlin & Schooler, 1978). The scale was designed to
measure “the extent to which people see themselves as being in control of the

forces that importantly affect their lives” (Pearlin et al., 1981). Participants rated

 

‘2 Although the elements measured in this construct also underscore key elements in the
construct of empowerment, l have chosen to refrain from using this term because research has
demonstrated that empowerment is contextually embedded (Foster-Fishman, Salem, Chibnall,
Legler, & Yapchai, 1998) and requires a contextual analysis to be fully understood (Zimmerman,
1990a). This analysis focuses more on the individually-oriented conceptions of empowerment as
primarily a personality variable at one point in time (rather than as a developing process, see
Kieffer, 1984) . As such, I have chosen the term personal control to avoid erroneous
extrapolation to broader work on psychological empowerment.

4S

 

items from 1 to 4 on a Likert-type scale. Coefficient alpha reliabilities on this brief
scale have been reported between .70 -.76 in a group of pregnant women
(Mercer & Ferketich, 1988). Both factor analysis and path analysis confirmed
that the items represent an accurate measurement model of sense of mastery
(Pearlin et al., 1981; Pearlin & Schooler, 1978). Test-retest reliability at a four
year interval was .44 (Pearlin et al., 1981). The measure has demonstrated its
sensitivity in identifying individual differences in mastery that are related to other
constructs. It has been used in research with African-American participants in a
home visiting program (Olds & Korfmacher, 1998), pregnant women (Mercer &
Ferketich, 1988), adolescents and young adults (Lewis, Ross, & Mirowsky,
1999), inner city women (Ennis et al., 2000), a broader urban community sample
(Turner et al., 1999), and single parent and two parent families experiencing
employment transitions (Ali & Avison, 1997).

Self-esteem. Self-esteem, or the judgments one makes about one’s own
self-worth, was measured using the Rosenberg Self-Esteem Scale. Although,
the scale was originally developed as a Guttman-type scale, research literature
now commonly sums the four point responses ( 1, strongly disagree to 4, strongly
agree) to yield self-esteem scores ranging from 10 to 40 (Blascovich & Tomaka,
1991; Rosenberg, 1979; Silber & Trippett, 1965). Coefficient alpha reliabilities in
pregnant women were .84 or greater (Mercer & Ferketich, 1988). Path analysis
confirmed that the items represent an accurate measurement model of self-
esteem (Pearlin et al., 1981) and this measurement model is maintained even in

high-risk samples (Wang, Siegal, Falck, 8. Carlson, 2001). At a two-week

46

interval, test-retest reliability was .85 (Silber & Trippett, 1965). Recent studies
indicate that the scale measures what “appears to be a relatively stable trait that
reflects general life satisfaction and affective symptoms rather than objective
functional status” (Torrey, Mueser, McHugo, & Drake, 2000). This is consistent
with an earlier finding of the scale’s test-retest reliability of .43 at a four year
interval (Pearlin et al., 1981). Numerous studies have used the Rosenberg Self-
Esteem Scale and have found relationships between the construct and other
psychosocial characteristics and outcomes. Studies have utilized the Rosenberg
Self-Esteem scale in samples with pregnant women (Hickey et al., 1995; Kemp &
Page, 1987; Mercer & Ferketich, 1988, 1994; Terry, Mayocchi, & Hynes, 1996),
pregnant adolescents (Bogat, Caldwell, Guzman, Galasso, & Davidson, 1998;
Patten, 1981), women of color (Lutenbacher & Hall, 1998; Rini, DunkeI-Schetter,
Wadhwa, & Sandman, 1999; Wasserrnan, Rauh, Brunelli, Garcia Castro, &
Necos, 1990; Woods et al., 1994), mothers receiving home-visiting services
(Vines & Williams-Burgess, 1994), and low-income families (Banyard, 1999;
Burns, Doremus, & Potter, 1990; Lawes, 1992). Study participants self-reported
responses to the Rosenberg Self-Esteem Scale upon enrollment to the research
study.

Mothering ability esteem. Mothering ability esteem was measured at
enrollment into the research study using four items from the Maternal Self-Report
Inventory (MSI) (Shea & Tronick, 1982). The MSI is a 100 item inventory (short
form 26 items) developed to assess maternal self-esteem. The full inventory

includes the following seven subscales: caretaking ability, general ability and

47

preparedness for mothering role, acceptance of baby, expected relationship with
baby, parental acceptance, body image and health, and feelings concerning
pregnancy, labor, and delivery. In other studies, the MSI typically has been used
to assess maternal self-esteem and expectations about mothering during the
perinatal or neonatal period. However, in this study, individuals responded to
these questions during pregnancy. The four items utilized loaded strongly on the
general ability and preparedness for mothering role subscale and were included
in the short version of the MSI to measure this subscale. The items were “I am
confident I will be able to work out any normal problems I might have with my
baby”, “I think that I will be a good mother”, “I feel that I will do a good job taking
care of my baby”, and “I know enough to be able to teach my baby many things
which he/she will have to learn”. The MSI, including these four items, has
demonstrated high correlations with other measures of similar constructs (Shea
& Tronick, 1982). The MSI also displayed expected relationships to infant health
status, and parity and maintained a four week test-retest reliability coefficient of
.85 (Shea & Tronick, 1982). The cronbach alpha for reliability of the MSI short
form items specific to the general mothering ability subscale was .88 (Shea &
Tronick, 1982). The MSI also was effectively used to investigate the parenting
qualities of low-income adolescent mothers of color (East, Matthews, & Felice,
1994).

Perceived personal control construct. LISREL 8.51 was used to
confirm that the measurement model incorporating mastery, self-esteem, and

general feelings about mothering ability measure elements of the same latent

48

construct (termed personal control). The path diagram is displayed in Appendix
C. The model fit the observed data well (352 (df = 8, n = 438) = 13.26; p = .103;
RMSEA = .039; GFI = .99).

Negative Network Orientation. Negative network orientation is “the
perspective that it is inadvisable, useless, or risky to seek help from others”
(Vaux, 1985). Negative network orientation was measured using the Network
Orientation Scale (Vaux, 1985).The scale was designed to measure a person’s
propensity toward utilizing his or her social support network in time of need.
Item-total correlations were found to be consistently positive and high.
Coefficient alphas across five samples ranged from .74 to .88 (Vaux, 1985). The
scale showed excellent test-retest reliability of .87 across a two-week time
interval (Vaux, 1985). Scores on the NOS displayed expected relationships with
measures of perceived social support, social network size, interpersonal trust,
and interpersonal coping (Vaux, 1985; Vaux et al., 1986). Scores also showed
expected differences between individuals who had been abused and those who
had not (Vaux, 1985). The NOS has been used effectively with children over 10
years of age (Belle, Dill, & Burr, 1991), in community samples (Vaux et al., 1986),
and with psychiatric outpatients (Cecil et al., 1995). Although there is currently
some debate about the dimensionality of the factor structure, the measure has
still shown adequate relationships to interpersonal trust instruments, social
support measures, and has predicted depressive symptoms (Forbes & Roger,
1999; Gruen et al., 1994). At enrollment into the study, responses to the 20

items on a four point Likert-type scale were summed to yield a NOS score such

49

that higher values represent more positive network orientations. In this study,
one item with highest face validity for comfort with self-disclosure was removed
from the scale sum and modeled separately (see next section for specific
details).

Given the item removal, prior concerns in the literature regarding scale
factor structure, and relatively limited use of this scale among low-income
pregnant women, factor analysis was conducted with this measure prior to use.
Indeed, the original factor structure was not replicated in this sample. Results
showed some similarity with other recent investigations on this issue, but were
not uniformly consistent (Forbes & Roger, 1999; Gruen et al., 1994). For study
purposes, 13 items loading on two factors were extracted. These were labeled
network mistrust (e.g., “You can never trust people to keep a secret”, “In the past,
I’ve rarely found other people’s opinions helpful when I’ve had a problem”, or “If
you confide in other people, they will take advantage of you”) and “network
usefulness” (e.g., “It really helps when you are angry to tell a friend what
happened”, ‘When a person gets upset they should talk it over with a friend”, or
“In the past, friends have really helped me out when I’ve had a problem”). These
factors consist of 7 items and 6 items respectively. Subsequent reliability
analysis found that each scale had reasonable internal consistency given item
length (coefficient alpha = .76 and .71 respectively), and represent two related
but not identical dimensions of the same higher order construct (13 item
coefficient alpha = .72). Appendix D includes more specific information about the

factor analysis and reliability analysis conducted. Due to the moderate coefficent

50

alpha in the reliability analysis, the network mistrust and network usefulness
scales were both entered into the data model as separate indicators rather than
grouped into a revised general network orientation scale.

Comfort with Self-Disclosure. A single item on the Network Orientation
Scale with high face validity for comfort with self-disclosure was modeled
separately in this study. At enrollment, participants agreed or disagreed on a
four point Likert-type scale to the item “It is easy for me to talk about personal
and private matters”. Higher scores represent greater disclosure comfort.

Prior Involvement with Child Protective Services. Prior involvement with
Child Protective Services was measured by a single self-report item asked upon
enrollment into the study. Individuals were asked “Have you ever been involved
with child prevention/protection services?”. Involvement was coded separately
for reported involvement as the child or involvement as a parent. The
dichotomous variable of reported involvement with child protective services as a
parent was incorporated into this study.

Expected and Reported Need lndices. Two separate indices were created
based on self-report study data collected from participants. Expected and
reported needs indices contained items related to needs about 13 specific areas
that clustered into 11 domains. The need domains included information about
pregnancy and childbirth, information on parenting and caring for children,
educational needs (of respondent), health insurance, employment, housing, food,
personal safety, counseling/mental health, drug use, and pregnancy support. All

conversations about expected and reported need variables were conducted

SI

during the research interviews by trained individuals who were not involved in
providing intervention services to any clients.

Reported needs index. The reported needs index was based upon a
series of questions that research interviewers asked participants at 34-38 weeks
gestation about each area of need since enrollment into the research study.
Individuals were told to reflect on their needs and concerns during the elapsed
time from enrollment until the late pregnancy interview. Then, they were asked,
“Since you enrolled in the study on (date given), did you have a need or concern
about/with/for/related to (insert area of potential concern)?” Individuals received
2 points on the reported need index for each “yes” response and 0 points for
each “no” response to questions”. Table 3 contains the specific questions asked
on the reported needs index.

Expected needs index. The expected needs index was based on
specific descriptive information provided by the participant about her situation
that would suggest a need would likely exist, but did not require her to state that
she possessed that need in her response. So, for instance, participants who
reported never delivering a live birth before were expected to have a need for
information about childbirth. The information used to construct an expected level
of need for each area was primarily taken from data collected at the pre-
randomization baseline interview when participants enrolled into the research

study. The only exceptions were in the domains of employment, housing,

 

'3 Please note that this choice to adopt a dichotomous measure of simple presence or absence of
needs across each domain probably minimized the extent to which individual differences in
interviewer abilities to elicit communication and individual differences in clients’ abilities to
skillfully describe their needs influenced the level of actual reported needs. Future research that
utilizes more open-ended measurement strategies may wish to explore these variables as
predictors of reported needs as well.

52

personal safety, and drug use. In these domains, it was determined that a more
accurate level of expected need during the pregnancy period could be obtained
by coupling information taken at enrollment with information about status
changes that occurred between enrollment and the 34-38 weeks gestation
interview. Individual information from participants was used to categorize the
participant anywhere from 0 (low) to 2 (high) in level of expected need for that
area of concern. Table 3 contains the specific information extracted to generate
the expected needs index scores in this study.

Need index characteristics. The participant distributions of expected
needs and reported needs are depicted in Appendix E. Expected needs were
fairly normally distributed within the sample, with scores ranging from 1 to
21.50“. In general, reported need scores were lower than expected need scores
although the range of values (0 to 22) was quite similar. Data were influenced by
the two point intervals associated with reported needs, but still retained a fairly
normal curvature across scores. Finally, the distribution of difference scores
calculated between reported need and expected need index scores also is
shown in Appendix E. The data depict remarkable variability in reporting
patterns. Difference scores ranged from -14.50 to +11.50. These scores were
fairly normally distributed around a mean of -1 .88 with a standard deviation of
4.66. Thus, the indices created for the study demonstrated good data

characteristics for use in subsequent analyses.

 

1‘ Participant scores on both the expected and reported need indices could range from 0-26.

53

Data Analysis.

Missing Data Imputation. Missing data were imputed in multiple stages.
For enrollment information there was very little missing data. All study
participants had virtually complete enrollment data. Missing data were typically
due to interviewer error (e.g., inadvertently skipping a question) and occurred on
multi-item scales. In each case, less than one percent of the participants had
missing data and less than twenty-five percent of data were missing from each
individual case on the scale. Scores were prorated based on the remaining
information from the scale. If the scale had established subscales, then data
imputation was based on these subscales instead of the whole scale.

At the late pregnancy interview, there also was very little missing data
utilized for analysis. This was primarily due to the established eligibility criteria
requiring completion of the late pregnancy interview for inclusion in the analyses
involving reported needs (see eligibility section of methods for further justification
and explanation). Thus, with regard to conceptual model 1, the only instances
with missing data involved situations where single items were skipped (i.e.,
typically due to interviewer error). When items were skipped on a scale, scores
were prorated using the same process as described earlier with enrollment data.
There were a few cases where a single reported need item was missing from the
interview. In these cases multiple regression was used to develop a predicted
value based on other reported needs and expected needs. Again, overall less
than one percent of data analyzed was missing and each case had more than

seventy-five percent of information completed.

54

Analytical Methods. The hypothesized conceptual model was investigated
using structural equation modeling with LISREL 8.51 software (Joreskog &
Sorbom, 1996). The observed covariance matrix of variables was compared to
what would have been expected given the set of interrelationships depicted in the
hypothesized model. The model was tested using both a structural and a
manifest model. The structural model incorporates the measurement models for
latent constructs into the analysis, whereas the manifest model explores
interrelationships between measured variables rather than latent constructs
(Schumacker & Lomax, 1996).

Data from the two intervention groups were combined to test the first
conceptual model. Three factors were considered before collapsing the groups.
First, the reported needs that comprised the outcome variable in this conceptual
model were recorded by research interviewers that were not part of either
intervention group. Therefore the recorded responses of participants should not
have been impacted by subtle programmatic biases or by differential responses
of participants to the distinct interventions. Secondly, neither program
specifically targeted changing individual tendencies in reporting needs as a focal
goal of the intervention. Thus, while active intervention involvement might alter
client behavior or the amount of help received (e.g., for drug use), the
intervention was considerably less likely to have a substantial impact on
perceptions and reported needs themselves. Finally, only one significant
difference was found between intervention groups on the thirteen reported needs

that comprise the reported need index (see Table 4). Pregnancy education was

55

the only reported need area showing a significant difference between intervention
groups. This significant difference was only observed among 16-19 year olds in
the study and there was no pattern of similar trends with this adolescent group
between intervention groups on other variables. Since no consistent pattern of
significance was observed and one difference was expected based on chance
alone, the effect was viewed not to be strong enough to require separate
examination across intervention groups. Hence data from individuals in both

intervention groups were collapsed for model testing.

56

RESULTS

Mnifest Model.

Structural equation modeling was employed to investigate whether the
observed relationships between measured variables reflected those predicted in
the first conceptual model (see Figure 1). The statistical properties of variables
necessitated some modifications to the hypothesized model during the course of
analysis.

Moael Adaptations. First, involvement with Child Protective Services
(CPS) as a parent was dropped as a variable from the model. Given that nearly
45 percent of the sample was primiparous and over 85 percent of multiparous
individuals did not report contact with CPS, the variable was highly inflated with
zero values. The restricted variance coupled with the inflated zero values posed
considerable challenges for parameter estimation (especially as a single item
indicator in the structural model), and it was dropped from the model.

Careful examination of the data also revealed correlations greater than .60
between perceived stress and depressive symptoms measures. While existing
literature had suggested strong relationships between these variables (Ludman
et al., 2000; Sachs, Hall, Lutenbacher, & Rayens, 1999), the presence of this
characteristic created severe multicollinearity problems. The relationships
observed in one variable were altered when the other variable was added to the
model; This problem persisted even after attempts to center both variables and
to use the categorical rather than continuous measure of depressive symptoms.

Hence, perceived stress and depressive symptoms were collapsed into one

57

variable (stressor responses) by summing the z-scores obtained on the
measures of perceived stress and depressive symptoms collected at enrollment
and at the late pregnancy interview.

Preliminary examination of the measurement model for perceived
personal control suggested that although mothering ability esteem loaded on the
control construct, the relationship was somewhat weaker than that observed for
mastery and self-esteem. Thus, each of these variables was examined
individually prior to collapsing them in the manifest model. Mastery and self-
esteem revealed similar relationships to other variables, with only minor
deviations in strength of relationships observed”. However, mothering ability
esteem demonstrated opposite relationships to reported need. Mastery and self-
esteem were collapsed into a single variable by summing z scores on both from
the enrollment assessment, whereas enrollment mothering ability esteem was
separated and included as a distinct variable in the model.

Model Fit. Generally, the hypothesized model of influences on reported
needs was supported (See Figure 2). Appendix F contains the means, standard
deviations, and correlations that were used in the analysis. The normal theory
least-squares weighted chi-square for the overall model (df = 29, n = 438) was
19.86 (p = .90), suggesting that the model fit the data well. The chi-square
measures the discrepancy between the covariance matrix of relationships
observed in the sample and the estimated matrix from the model (Schumacker &

Lomax, 1996). The Root Mean Square Error of Approximation (RMSEA)

 

‘5 Mastery was more strongly related to both reported needs and social support than self-esteem,
but differences between them were relatively minor.

58

(Steiger, 1990) was 0.0, indicating exact model fit (Brown & Cudeck, 1993).
Other indicators of good model fit also supported adoption of the model
(Goodness of Fit Index (GFI) = .99; Comparative Fit Index (CF I) = 1.0; See
Appendix F for other indicators of fit). These results were found using maximum
likelihood parameter estimation techniques. Similar findings also were observed
when data were re-analyzed using generalized least squares approaches. The
model fit criteria using generalized least squares were nearly identical (normal
theory least-squares weighted chi-square (df = 29, n = 438) = 22.99, p = .78;
RMSEA = 0.0, GFI = .99, CFI = 1.0; See Appendix H for other indicators of fit).
Figure 3 displays the path estimates using generalized least squares
approaches. No path estimates deviated in the direction of relationship or in
statistical signficance status from the maximum likelihood model. All of path
estimates were within .03 of the estimates in the maximum likelihood model.
Observed Effects. The adopted model included statistically significant
hypothesized relationships in the expected directions between expected and
reported need index scores, expected needs and stressor responses,
mastery/seIf-esteem and stressor responses, mastery/self-esteem and social
support. Heightened stressor responses were significantly related to higher
reported needs. This effect was found with both perceived stress and depressive
symptoms collapsed together and with each measure individually, whereas a
positive relationship to reported need was only predicted for perceived stress.
Neither level of social support nor indicators of network orientation displayed

significant relationship to reported needs. However, greater comfort with

59

 

personal disclosure was surprisingly related to less, rather than more, reported
needs. Of disclosure-related indicators, only network use usefulness displayed
the expected significant positive effect on social support.

Contrary to predictions, personal control in the form of mastery and
general self-esteem exhibited a statistically significant positive relationship to
reported needs. However, more specific mothering ability self-esteem showed
the expected (although non-significant) negative relationships to reported needs.
These relationships from general mastery/self-esteem and maternal self-esteem
to reported need were retained even when a path from general mastery/self-
esteem to mothering ability self-esteem was included in the model. The added
path indicated a statistically significant relationship (standardized estimate = .29).
With the inclusion of this path, the overall model fit, although still indicative of a
close fit, did deteriorate (minimum fit function chi-square (df = 27, n = 438) =
38.53; RMSEA = .031; GFI = .98; CFI = .98). The direction and significance of
the paths of interest remained unchanged (general mastery/self-esteem to
reported needs estimate = .13; mothering ability esteem to reported needs
estimate = -.08). See Figure 4 for the complete model of estimates following
inclusion of the personal control path using maximum likelihood estimation
approaches. Indicators of overall model fit for Figure 4 are listed in Appendix I.

Although not included in the conceptual model, maternal age also was
tested to consider whether it was confounding other relationships in the model.
Age was included as an exogenous variable impacting reported need.

Regardless of whether age was modeled as a continuous variable or a

60

 

categorical variable (5 18 or 5 19 both modeled), the standardized path estimate
to reported need was not statistically significant (consistently standardized path
estimate = -.01). Other paths in the model also remained unchanged. So,
maternal age was not included in the final model.

Structural Model.

Structural equation modeling also was used to examine whether the
hypothesized interrelationships between variables were modeled effectively at
the latent construct level as well as the observed variable level.

Model Adaptations. For the structural model, expected and reported need
indices were broken into three indicators based on conceptual factors. Lack of
information about pregnancy/childbirth and lack of information on
parenting/caring for children were combined into a parenting factor. Education,
health insurance, employment, housing, food, and pregnancy support were
collapsed into a basic needs factor. Finally, personal safety (including domestic
abuse), counseling/mental health, and drug use were collapsed into a high risk
life experiences factor. Index scores were summed for each factor and utilized
as indicators in the structural model.

The two factors generated from the network orientation scale (mistrust
network and network use usefulness) as well as the single item from that same
scale representing disclosure comfort were modeled as three indicators of a
higher order latent factor titled “disclosure”.

The investigation of whether the observed data were adequately

described by the conceptual model began with entry of the conceptual model

6]

pattern as shown in Figure 1 (other than the aforementioned modifications to
need indicators and the disclosure construct). However, serious problems were
encountered with the stress and depression constructs. Both factors had only
two indicators and numerous paths to be estimated. This created identification
problems with the data. Various steps were undertaken to minimize problems,
including the development of two parallel forms of each perceived stress and
depression scale to double the number of indicators. However, these efforts
were not sufficient, particularly with the considerable estimation demands on the
reported need construct. Ongoing multicollinearity problems like those observed
in the manifest variable model also were encountered. Thus, I chose to collapse
the perceived stress and depressive symptoms constructs and model them jointly
as stressor responses.

After creating the stressor response variable, the data still did not closely
fit the model. Given that relatively little background was available for predicting
model relationships at the outset of the study, I engaged in exploratory
examination of the data to investigate whether unforeseen relationships between
variables masked an acceptable model fit. Modification indices provided by
LISREL based on empirical relationships in the data were considered for
theoretically substantive relationships that could be included to optimize the
model fit. Throughout this process, care was taken not to violate the
precedence rule. That is, correlations between constructs were never estimated
if a path, either directly or through a series of indirect paths, already modeled the

relationship between the two variables (McDonald & Ringo Ho, 2002).

62

Substantive additions not initially developed in the conceptual model or
the manifest model included a path from expected need to personal control, a
correlation between disclosure and expected need, and a correlation between
disclosure and stressor responses. The negative relationship from expected
needs to personal control would be consistent with the common premise that
early life experiences (e.g., abuse history), developmental phase (e.g.,
experiencing transition to parenthood or multiparous pregnancy), and current life
situation (e.g, access to basic life needs) influence each individual’s current
appraisal of her control over her environment. Likewise, the correlation between
expected needs and disclosure represented a relationship between life
experiences and one’s willingness to trust others with confidential information
and share personal information about oneself with them. Finally the correlation
between disclosure and stress responses would logically model the relationship
between comfort with actual personal disclosure and scores on self-report
instruments identifying circumstances of individual vulnerability (e.g., items like: “I
had crying spells” or “Felt that you were unable to control the important things in
your Iife”) to the interviewer.

Other changes to the model included permitting correlations between
residuals that were not explained in the model itself. These residual correlations
were based on substantive consideration of the recommended modification
indices. The residuals that were permitted to correlate are listed in a table in

Appendix K. Each of the relationships indicated is interpretable.

63

Model Fit. Despite the complexity of this structural model, the
hypothesized model of relationships was generally supported. Figure 5 displays
the final model of relationships that was adopted. The normal theory weighted
least squares chi-square value (df = 224, n = 438) was 547.72, (p = 0.0).
Although the chi-square never dropped below the threshold of significance or the
rule of thumb of a chi-square value less than twice the degrees of freedom, other
fit indicators suggested that this model was an acceptable fit for the data
(RMSEA = .058, GFI = .90, Independence AIC = 3171.3, Model AIC = 699.72).
Acceptable fit (as opposed to close fit) threshold values are more common in
structural than manifest models because the measurement error, as well as the
relationships between variables influences model fit. The values of alternative fit
indices for this model are listed in Appendix M.

Measurement Models. The measurement models embedded within the
larger structural model were effective throughout the model. The measurement
model being tested met the necessary condition for identifiability that each
indicator loaded on only one latent construct. Indeed, no indicators loaded on
more than one construct and each one loaded significantly on the expected latent
factor. The only exception was in the case of the Family Helper indicator that
was expected to load on the social support construct. In this case, the factor
loading for social support was only .02. The Father of Baby (FOB) Helper
variable also yielded a low factor loading (.15), however it retained statistical
significance. These poor measurement model characteristics for the social

support variable are explicable on both empirical and theoretical grounds. First,

64

 

both helper variables are author-designed single item measures with less prior
validation than other social support indicators in the model. Second, whereas the
other three social support items were measured using Likert-type response
scales, the two helper variables were scored from participant response to an
open-ended question. So, individual differences in response in these diverse
measurement methods may have contributed to poorer factor loadings for these
variables than for the other three support indicators. Finally, social support is
widely regarded as multidimensional construct (Dunst & Trivette, 1988). The five
indicators in this model represent 3 different types of social support. The two
helper variables were intended to measure received instrumental support
whereas the other three indicators tapped available and perceived support. The
variability in these types of support as well as in the measurement characteristics
of the variables probably produced observed differences in the social support
factor loadings. Although the family helper indicator did not load significantly on
the social support construct, evidence suggested that this indicator played a
critical role in the cohesiveness of the factor structure, perhaps through residual
correlations across types of support as well as in the factor loadings. Despite the
lack of significant factor loading, the exclusion of this variable from the model
resulted in an unstable social support factor and contributed to a lack of
convergence in the overall model fit. In addition, the indicator did not load on
other factors in the model more effectively than it loaded on the social support
construct. Given the empirical and theoretical considerations underlying the low

factor loading and the significance of the indicator’s inclusion to the factor, the

65

family helper variable was retained in the model despite the non-significant factor
loading.

The measurement characteristics of other variables in the model can be
gleaned, in part, from looking at the error terms for indicators. These are listed in
Appendix J. These error terms ranged considerably”. Nevertheless, each of the
indicators other than family helper significantly loaded on the intended factor and
contributed to explanations about relationships in the model.

Observed Effects. The adopted model included statistically significant
hypothesized relationships in the expected directions on paths17 from expected
need to reported need index factors, expected needs to stressor responses,
personal control to social support, and personal control to mothering ability
esteem. Heightened stressor responses were significantly related to higher
reported needs, as was predicted for perceived stress but not depressive
symptoms. As theorized, a negative relationship was observed in the path from
social support to reported needs, however, the strength of this path did not reach
a statistically significant threshold. Contrary to predictions, but consistent with

the findings in the manifest model, personal control as derived from mastery and

 

‘6 One notable problem occurred in the estimation of errors. The error variance listed for the
latent factor of reported need is -.74. However, variances are, by definition, positive values.
Thus, it is clear that an error occurred somewhere in the LISREL 8.51 estimation process. While
this is indicative of a flaw in the software, the discovery of similar problems has occurred by other
users and is not necessarily suggestive of problems with other parameter estimates in the model.
We are, unfortunately, at the software’s mercy until the next program patch is available!

‘7 Please note that standardized estimates documented on Figure 5 do po_t have a maximum
value of 1. The LISREL standardization process divides parameter estimates by unity standard
deviations, resulting in comparable values, though not scaled to a maximum ratio. Thus,
comparisons between paths should be limited to those conducted with ordinal level data. That is,
larger numbers represent a stronger relationship. However, a standardized estimate of 4 is not
necessarily twice the size of an estimate of 2. One path on each measurement model and the
path from expected to reported needs have also been fixed to a value of one. LISREL routinely
requires this action for scaling purposes. Fixing the path from expected to reported needs was
also critical to support model identifiability of the reported needs construct.

66

self-esteem indicators displayed positive relationships to reported needs.
However, more specific mothering ability self-esteem showed the expected, and
now statistically significant, negative relationship to reported needs predicted for
the entire personal control construct. The disclosure factor yielded a surprising,
statistically significant negative relationship with reported needs similar to that
found in the manifest model. However, the strong negative correlation between
disclosure and expected needs may underlie the negative relationship observed
in the disclosure to reported need path. The expected positive relationship
between disclosure and social support appeared, but did not reach a statistically
significant threshold. Finally, the expected strong, statistically significant,
negative path from personal control to stressor responses observed in the
manifest model was not replicated in the adopted structural model. Instead a
small nonsignificant positive relationship was observed. However, these
contradictory results may be explained by the strong, significant negative path
from expected need to personal control that was modeled in the structural, but

not the manifest model.

67

DISCUSSION

The purpose of this investigation was to understand relationships between
expected and reported needs among low-income pregnant women. Are these
constructs the same? The empirical answer was no. Expected needs do form
the basis for reported needs. But, expected and reported needs proved to be far
from identical constructs. The models tested in chapter 1 provided a number of
insights into what factors underlie differences between expected and reported
needs.

Influences on Reported Need_s.

Expected Needs. Path estimates demonstrated that expected needs
formed the basis for the needs that individuals reported. Expected needs yielded
powerful direct and indirect effects on reported need. Both manifest and
structural models included a statistically significant direct positive path from
expected to reported needs. Thus, as assumed in family-centered care models,
the presence of relatively objective risk criteria developed by researchers did
translate into a higher perception of needs in the personal lives of low-income
pregnant women. People who lacked food, housing, personal safety, and
experience with caring for children were more likely to express having needs in
these areas when asked directly about them. However, the size of the path
estimate suggests that while expected needs were influential predictors of
reported needs, they did not directly account for nearly all of the variance in

reported needs.

68

Stressor Responses. Expected needs also displayed dramatic indirect
effects on reported needs. The presence of higher numbers of expected needs
had a significant positive effect on the stressor responses experienced by study
participants. That is, people with fulfillment of fewer basic needs and more
challenging life experiences (including histories of illicit substance use, physical
or sexual abuse) perceived greater stress in their lives and reported more
depressive symptoms such as sadness, crying, and changes in sleep, appetite,
and social activities. These significantly heightened stressor responses were
positively related to greater reported needs. Thus, when expected needs elicited
a stress reaction individuals reported significantly higher levels of need to
interviewers.

The presence of stressor responses mediating a pathway from expected
to reported needs was consistent with existing research. A number of studies
have documented perceived stress and depressive symptoms as responses to
difficult life events (Brooks-Gunn et al., 2000; Dunkel-Schetter et al., 2001;
Kessler, 1997). Thus, the link found between expected needs and stressor
responses was not unique to this study. However, less prior investigation has
focused on relationships between stressor responses and reported needs. The
significant positive effect that was modeled is consistent with a study by
Eisenstadt (1987) in which increased levels of perceived stress promoted initial
service utilization. Perceived stress may trigger both the self-report of needs and

the active response of seeking services for assistance.

69

The relationship between stressor responses and reported needs also
was consistent with work using the health belief model. Although it is the least-
studied component of the health belief model, cues to action are expected to
trigger preventive health behaviors (Maddux & DuCharme, 1997). External or
internal cues are believed to stimulate the belief-action link when an individual
perceives a threat and when he or she rationalizes that the benefits of health
behavior action outweigh the costs and barriers to act (Strecher et al., 1997).
Stressor responses may serve as cues for individuals that create a sense of
immediacy to resolve the issues underlying those responses. Data collected in
this study were insufficient to test whether stressor responses increased the
perception of threat from inaction or, instead, were critical in instigating behavior
(i.e., reporting need) in individuals who already perceived a threat but would not
othenNise choose to act on that perception. Future research should consider this
distinction further to clarify more precisely how stressor responses influence
reported needs.

An alternative explanation for the stressor response to reported need
effect could include the influence of stressor responses in the process of labeling
oneself as “in need” or not. Health research suggests that most people possess
some measure of illness or disequilibrium in their lives (Antonovsky, 1987). Yet,
considerable variety exists in when people begin to identify themselves as ill and
act on these beliefs. Prior work found that individuals labeled themselves on a
continuum from healthy or ill depending on their condition along four dimensions:

degree of pain, prevention of role and task performance, threat to life, and

70

external recognition that the condition requires care (Antonovsky, 1973;
Gochman, 1997). Since the degree of pain and threat to life are considerably
weakened in most of the maternal support services intervention areas, it may
require a higher degree of urgency along the other two dimensions to create the
failure of functioning and disequilibrium necessary to label oneself as in need of
assistance. The presence of heightened stress responses may have contributed
to reporting a “need” because the stressor responses interfered with routine role
and task performance in daily life.

Finally, the relationship between stressor responses and reported needs
could be attributed to shared methodological bias. Questions on the PSS and
CES-D pertain to feelings or experiences of vulnerability. For instance, each
participant rated the frequency of experiences in which she “felt difficulties were
piling up so high that [she] could not overcome them”, “felt that [she was] unable
to control the important things in [her] life”, “had crying spells”, or “thought [her]
life had been a failure”. Individuals who were uncomfortable revealing these
experiences to interviewers also may have been uncomfortable reporting about
personal needs for pregnancy support, drug use, housing, or domestic violence.
However, in the structural model a correlation between disclosure and stressor
responses was included to model this more general reporting tendency. Even
after the inclusion of the significant positive correlation between disclosure
tendencies and stressor responses, a sizeable significant positive path from
stressor responses to reported needs remained. Given the strength of this path

estimate, it seems unlikely that shared methodological variance could completely

7]

account for the relationship. This is consistent with an older study by Yamamoto
and Kinney (1976) in which lie scale scores were significantly correlated with
manifest anxiety scale18 scores (r = .43, p < .01), but still left sufficient
unexplained variance for life event scores and adaptive potential scores also to
provide significant, unique contributions in the regression analysis predicting
anxiety. Future work may wish to expand the model to include a social
desirability component and/or a lie scale to more completely discount the
possibility of methodological bias contributing to increased reported needs
among individuals with higher levels of stressor responses.

Depressive symptoms. Both perceived stress and depressive
symptoms displayed similar relationships within the model when each was
included individually. Like perceived stress, higher levels of depressive
symptoms were significantly positively related to reported needs. So, the
hypothesized positive relationship between perceived stress and reported needs
was supported whereas the expected negative relationship between depressive
symptoms and reported needs was not. The depressive symptom-related
interference in the cognitive processes of appraisal described by Fiske and
Pavelchak (1986) and others may have been attenuated by the use of an
interviewer engaging the participant in conversation in the research methodology.
People who successfully completed the one-hour interview maintained a
reasonable level of social interaction throughout the session. So, even

individuals who reported high levels of depressive symptoms in the week prior to

 

'8 Although not included in the stressor response variable used in this study, anxiety is often
coupled with perceived stress and depressive symptoms as a response to difficult life events
(Nations, Camino, & Walker, 1988).

72

the interview probably were not exhibiting their highest levels of symptoms during
the actual interview session when they reported their needs. Finally, the data in
the analyses undertaken did not differentiate whether participants with high
depressive symptoms displayed inaccurate self-assessments by over-, rather
than under-, estimating their needs. Future research will need to explore if over-
reporting occurs among individuals with depressive symptoms or if people with
depressive symptoms continue to make accurate appraisals of need in their self-
reports. Based on the data provided through investigation of conceptual model
one, it appeared that both depressive symptoms and perceived stress operated
similarly to increase reported need.

Perceifl Personal Control. Contrary to predictions, a direct positive
relationship between perceived personal control and reported need was found.
Individuals who reported higher levels of mastery and self-esteem also reported
more needs to the interviewer. This finding was consistent with work suggesting
that empowering individuals also raises their awareness about needs and issues
in their lives. For instance, in a qualitative study investigating resistance among
mothers in a homeless shelter, one mother said (Koch et al., 1998):

“I never thought being in the shelter was going to lead to what it led to.

Finding out things that you ignore, and you hide....The shelter was

positive as far as getting the support and the help and the awareness of

what is actually going on in your life, that I kept shutting out.” (76)

In this example, as the shelter intervention expanded feelings of personal
confidence and worth, individuals were increasingly able to label and confront the

difficult truths in their lives. Another study validated this assertion as well. It

found that abused women seeking help from a shelter reported higher levels of

73

abuse and learned helplessness as compared to abused women who had not
sought help and to women who had not experienced abuse (Wilson, Vercella,
Brems, Benning, & Renfro, 1992).

However, the positive relationship between control and reported need is
somewhat inconsistent with existing research finding that individuals with greater
personal control generally received fewer visits from service providers in a client-
centered home visiting program (Olds & Korfmacher, 1998). Whereas individuals
reporting more needs would typically be expected to receive more services, the
individuals with high levels of control actually completed fewer home visits. This
inconsistency may be explained, in part, by the relationships between control and
other variables in the model.

As hypothesized, a strong, statistically significant negative relationship
was modeled between manifest control and stressor response variables.
Individuals with higher levels of control exhibited fewer stressor responses. So,
less frequent home visits to individuals with high control may have been
necessary despite the high level of reported needs because service providers did
not have. to closely monitor and intervene to reduce the depressive symptoms
and perceived stress observed. However, closer examination of the data in the
structural model suggested that the negative relationship between personal
control and stressor responses was a spurious one, generated by links with
expected needs. The presence of higher expected needs was related both to
increased stressor responses and to reduced perceived personal control. These

dual relationships suggest that the heightened frequency of home visits reported

74

by Olds and Korfmacher (1998) actually reflected intervention sensitivity to the
higher expected needs that existed among those low in control.

The finding that both stressor responses and personal control were
influenced by expected needs is consistent with Link and Phelan’s (1995)
assertion that social factors such as socioeconomic status are “fundamental
causes” of disease and should be targets for intervention”. It is also consistent
with work by Turner and his colleagues (1999) documenting that the social
distributions for mastery and self-esteem complement those distributions for
depressive symptoms and depressive disorders. However, whereas Turner et al.
used a series of regressions to assert that mastery and self-esteem largely
mediated the connection between socioeconomic status and depressive
symptoms, the findings in the present study do not concur (Turner et al., 1999).
Using simultaneous estimation within structural equation modeling, l instead
found no significant relationship between personal control and stressor
responses (including depressive symptoms) after the influences of expected
need on each factor were taken into account. It is important to consider that this
study differed from Turner’s in operational definitions as well as in analytical
methods employed. Although all participants in the current study were classified
as low socioeconomic status, considerable variation was observed in

participants’ expected need scores. Thus, using the more refined measurement

 

‘9 Link and Phelan posit that interventions must target these “fundamental causes” of disease
rather than any mediating conditions because “in the context of a dynamic system in which risk
factors, knowledge of risk factors, treatments, and patterns of disease are changing, the
association between a fundamental social cause and disease will endure because the resources
it entails are transportable to new situations. If one genuinely wants to alter the effects of a
fundamental cause, one must address the fundamental cause itself” (Link & Phelan, 1995, p. 88).

75

of need rather than socioeconomic status may account for differences in study
outcomes.

Mothering ability esteem. To this point, my discussion about the
personal control construct has focused on control as defined by mastery and self-
esteem. Yet, four indicators measuring general feelings about
mothering/matemal ability self-esteem also measured elements of personal
control. Contrary to expectations however, these specific esteem appraisals
displayed quite different relationships to reported needs than did the general
measures of personal control. Whereas general personal control surprisingly
exhibited a strong positive influence on reported needs, specific mothering ability
esteem displayed the hypothesized negative relationship to reported needs. This
relationship was weaker than that observed between the general control
construct and reported needs, however, it reached statistical significance when it
was modeled at the latent variable level. The finding that individuals with less
mothering ability esteem reported greater needs is consistent with prior home-
visiting research in which the most beneficial gains from participation
(presumably stimulated by intervention in areas of unmet needs important to the
client during the intervention period) were documented with primiparous
adolescent mothers (Olds et al., 1999).

Personal control and appraisal processes. The findings with regard to
the opposite influences of general and more specific constructs of personal
control on reported needs raised interesting questions about the self-appraisal

process. Do individuals sequentially perceive personal needs, report them,

76

 

appraise their prospects for managing the event effectively with various actions,
and then implement selected health behavior activities? Or, is the process of
appraising one’s prospects for managing the event intertwined with whether the
event is initially perceived to be, and thus, reported as an existing need? The
current study was not designed to test these alternative possibilities. However,
embedded in the findings was some support for the latter of these process
models.

The results of the current study supported the assertion that people do
use general assessments of personal competence and contingency to appraise
their prospects for changing a given condition prior to labeling themselves as
having a need and reporting it. However, appraisals of personal control did not
reduce one’s tendency to describe a situation as a need (because presumably
one could handle it independently). Instead, strong appraisals of personal
control actually augmented one’s labeling a given condition as a personal need.
A potential explanation could be that individuals attributed a personal need to
exist only when the given condition was dissonant from the condition expected
based on their general views of their competence and their beliefs that their
individual actions had contributed to that condition. For example, study
participants without high school diplomas might only have perceived themselves
as having needs for further education if they felt that they possessed the
capabilities to continue pursuing education successfully (yet had not done so)
and believed that their actions to access further education would make a

difference in changing their future educational status into high school graduates.

77

Individuals might tend not to define education as a need if they believed either
that they could not successfully complete the required work even if the
opportunity arose or that external factors (e.g., cost, logistics of getting to
classes, attitudes of a boyfriend/spouse toward attempts to further education),
rather than their own actions, primarily influenced their current and future
educational status. While this potential explanation is speculative, the current
study results did suggest that a foundation of personal control was necessary to
report a need. Thus, reporting a need seemed to involve defining that a given
condition not only could be changed, but also constituted a personal “need” that
the individual would like to change within her individual situation at some point in
the future. That is, reporting a personal need had a considerable amount in
common with setting a personal goal to change the condition or circumstance”.
The seif-report of a need related to a specific issue was, in effect, the first step in
commitment to change the condition.

The view that reporting a need is an incremental step toward behavioral
action is consistent with the importance of problem perception in existing
theoretical models of risk reduction behavior change. In the AIDS Risk
Reduction Model (AARM), problem perception indicated by knowledge of risk,
personal susceptibility, and perceived undesirability of the consequence

(HIV/AIDS transmission) is the first step in enactment of behavior changes

 

2° This commonality between reporting needs and goal-setting has not previously been
addressed in the literature. In fact, it was not obvious at the outset that one would expect these
two concepts to take such similar forms. For instance, one could have found that people try to
appraise whether or not they had a need in specific areas by taking the perspective of others
(e.g., her mother, the father of the baby) and reporting their perceptions of her needs or by a
more objective comparison of their competence or fulfillment of needs in each area relative to
other people they know.

78

(Catania et al., 1990). Likewise, in Prochaska’s theory of change, the critical
difference between the precontemplation and contemplation stages of change is
not the individuals’ readiness to commit to behavior change, but rather his/her
awareness that a problem does exist (Prochaska, DiClemente, & Norcross,
1992). While both of these theories acknowledge that identifying the problem or
the presence of a need for change is an important step in risk reduction, neither
theory has discussed personal control as a critical element influencing labeling of
the problem. In contrast, researchers studying social cognitive theory have
reported that efficacy beliefs influence goal-setting practices, expected outcomes
of action, behaviors undertaken, and the extent of continued persistence toward
goal achievement when barriers were encountered (Bandura, 1986). This
research also has found that people act to enhance their self-evaluations when
behaviors are compared to personal standards (Bandura, 2001; Higgins, 1987).
Hence, social cognitive theorists might explain the current study results as
indicating that people who appraise themselves as not having the capability or
the agency to evoke change would be motivated to avoid reporting a need (or
goal) to change. Using this social-cognitive interpretation, the current study
findings then were consistent with a large body of work on learned helplessness
in which individuals and animals experiencing repeated failure in altering an
unpleasant situation eventually cease to initiate self-protective behaviors (Maier
& Seligman, 1976; Seligman, 1968, 1974; Seligman & Meier, 1967).

However, data in the current study also suggested that two appraisal

processes may operate simultaneously to influence reported needs. While

79

 

general indices of personal control (overall mastery and self-esteem) displayed a
direct positive effect on reported needs, more specific content-related areas of
self-esteem exhibited the opposite relationship to reported needs. So, individuals
expressing greater mothering ability esteem reported significantly fewer needs
than those identifying ambivalence or low mothering ability esteem. Thus,
individuals appeared to simultaneously evaluate their prospects for managing the
event (through general control constructs) and their perceived existing level of
competence in the specific need area. So, skills and deficits related to parenting
were considered prior to individuals’ reports of whether or not they each
possessed a need related to parenting. Participants who were more confident
about their parenting abilities and caretaking skills were less likely to report a
need to the interviewer. This finding is consistent with work conducted with the
health belief model (Strecher et al., 1997). Individuals with higher perceived
competence would be expected to see less sizeable benefits from reporting and
acting on a need, and thus be less inclined to do so than people who could gain
more from this activity.

Finally, one might be tempted to examine the relative size of path
estimates from general and specific personal control constructs to reported need
and infer the relative importance of the two processes. However, the influence of
mothering ability esteem on reported need may have been diminished because
of the way that the constructs were measured. The broader self-esteem
literature has noted that specific indices of self-esteem can be powerful

predictors of attitudes and behaviors for certain areas and they explain unique

80

 

variance beyond that measured in more general self-esteem indicators (Hoge,
Smit, & Hanson, 1990; Rosenberg, Schooler, Schoenbach, & Rosenberg, 1995).
However, the significance of these specific indices of esteem tends to be limited
primarily to the domain addressed in the index. Likewise, the specific mothering
ability esteem construct was likely to show the greatest influence with reported
needs specific to parenting abilities. It would be expected to have less influence
(if any) on reported needs in other areas such as housing, personal safety, and
drug use. However, the reported needs construct was a combination of various
domains of need. Whereas general control indicators would be expected to
display relationships across domains, the more specific esteem indicators would
not. Thus, readers should cautiously interpret the relative size of these two path
estimates until future work can examine shifts in these patterns when reported
needs are restricted to the focal domain measured in the specific esteem
construct.

gagnitive Agility. The current study did not assess relationships between
verbal or general intelligence and reported needs. Thus, little is known about the
influence of cognitive ability on reported needs. No support was found for any
significant influence of developmental differences in cognition as indexed by age
on reported needs. However, eligibility criteria restricted the age of participants
to at least sixteen years of age, thereby reducing the variability in cognitive
patterns that might otherwise have been observed. The potential influence of
cognitive ability on need-reporting behavior also may have been minimized

through the methodology by which individuals reported needs. Instead of open-

81

ended questions, interviewers named specific areas of potential needs and
asked study participants to simply acknowledge whether or not they had
experienced a need or concern with each issue. Future research should explore
the influences of cognitive ability, verbal fluency, and literacy on reporting of
needs in various interviewing approaches.

Social Support. Results found a significant, positive path estimate
between personal control and social support. Individuals who reported greater
feelings of both competence and contingency also were more likely to report
access to support from family and friends and actual receipt of instrumental
support during the pregnancy period. This finding was consistent with previous
research and theoretical discussions in the social support literature (Eckenrode,
1983; Sheppard, 1993). Individuals with an internal locus of control have
reported more supportive contacts with people in their social networks following
stressful experiences (Eckenrode, 1983). Thus, results suggested that when
higher levels of expected needs undermined personal control, they also
produced the indirect effect of reducing the social support available to the
individual. So, individuals with greater expected needs experienced higher levels
of stressor responses and had fewer individual (i.e., personal control) and social
(i.e., social support) resources with which to cope.

Despite a lack of research investigating links between social support and
reported needs, higher levels of social support were hypothesized to reduce
reported needs. The presence of social resources to assist with difficult

circumstances was expected to reduce individual tendencies to define a situation

82

as a need. The findings in this dissertation did not lend support to this
hypothesis. Although both manifest and structural models generated negative
path estimates between social support and reported need, these estimates were
consistently small and non—significant. Thus, it may be that individual appraisals
of need were relatively unaffected by the presence or absence of social
resources to meet that need. Or, it may have been that the true relationship
between social support and reported need was masked by inadequacies in
measurement in the current study. The latent social support construct included
available, perceived, and received social support from two different sources
(family and father of the baby). Future research may look at each of these
dimensions and sources individually to consider if specific types or sources of
support have an important influence on reported needs. In addition,
measurement of social support in specific situations relevant to the needs (e.g.,
with parenting issues, providing housing or food, helping you deal with a
substance use issue) rather than global indices may produce stronger
relationships to need. In the same way that specific indicators of mothering
ability esteem indicated different relationships to need than general indicators of
overall esteem, more specific indicators of social support might yield interesting
results. In addition, measurement of social support in more specific situations
may inform researchers about whether social resources are critical to the need
appraisal process in widely varied situations. For example, it may be that the
presence of support from the social network with parenting may overcome the

attribution of a personal parenting need whereas even the presence of social

83

support related to drug use issues would not because of a differential perception
that professional involvement is needed to handle that issue.

Disclosure. The final set of relationships tested in the model involved the
disclosure construct. Surprisingly, in both the manifest and structural models,
disclosure yielded a significant negative relationship on reported needs. That is,
people who agreed that “when a person gets upset they should talk it over with a
friend” and “it is easy for me to talk about personal and private matters” actually
reported fewer personal needs to the interviewer than people with lower scores
on disclosure items. Initial explanations for this counterintuitive finding focused
on the relationship between disclosure and social support. Participants who
expressed greater comfort with disclosure might rely on informal support
networks more than professionals, and thus, disclose less to the research
interviewer. This explanation was neither completely supported nor discredited.
Path estimates for the direct path from disclosure to social support were
consistently positive, but they were small. In the structural model, the
relationship between the two latent constructs was nonsignificant. However, the
power of this path may have been minimized by the inclusion of the mistrust
indicator with the talk usefulness indicator. Although both of these indicators
loaded significantly on the disclosure construct, they appeared to have different
relationships to social support. When each indicator was modeled individually in
the manifest model, mistrust was unrelated to social support whereas talk
usefulness was significantly associated with higher levels of support. These

differences are reasonable given that individuals vary their disclosure levels

84

depending on their relationships with the listener (Burnard & Morrison, 1992;
Derlega et al., 1993; Jourard, 1971). Thus, the usefulness of talking problems
over with friends (as measured in the talk useful indicator) would be very relevant
to eliciting social support from people inside the social network. In contrast,
whether or not individuals displayed general mistrust of “other people” may be
considerably less related to their actual disclosure levels among friends and
family members“.

An alternative explanation for the counterintuitive negative path from
disclosure to reported need involves links to expected need. In the structural
model, a moderate, statistically significant, negative correlation between
expected need and disclosure was found. This correlation implies that a
common underlying experience (e.g., insecure attachment histories) contributed
both to higher expected needs and to less trusting disclosure patterns. If this
was the case, then the heightened reported needs among those low in disclosure
might have reflected greater needs among individuals sharing this underlying
experience rather than a main effect of disclosure on reported needs. Existing
research literature supports the possibility that a pathway of this type might exist.
Research on disclosure established long ago that people’s comfort with self-
disclosure is influenced by their past experiences, including the previous
responses (e.g., lack of acceptance, ridicule, lack of confidentiality) of individuals

to them after sharing occurred (Burnard & Morrison, 1992; Derlega et al., 1993;

 

2‘ Instructions read prior to answering the scale asked individuals to consider “what [they] think
about talking to people in general”. So, unless a question specifically referenced a “friend”,
individuals were primed to expect that “other people” indicated individuals from the larger
population in general rather than common social network associates.

85

Vaux et al., 1986). In addition, a qualitative study investigating preventive
intervention effects with 47 pregnant mothers at exceedingly high risk to abuse or
neglect their children found the following co-occurring characteristics within their
sample: difficult attachment histories”; prior abuse or neglect of a child of their
own; considerably limited life skills and high needs for basic necessities (food,
housing, clothing, etc); and extremely guarded and nonverbal behavior with
strangers (Pharis & Levin, 1991). Other studies focused on survivors of domestic
abuse also have documented links between components in this pathway.
Findings have included the following: links between difficult relationship and
attachment histories over time with subsequent cycles of repeated coupling and
de-coupling with abusers (Landenburger, 1998), associations of involvement in
abusive relationships with lack of fulfillment of basic life needs and a paucity of
individual life skills and personal resources (Lein, Jacquet, Lewis, Cole, &
Williams, 2001), and limitations imposed by the abuser on the survivor’s social
contact opportunities, resulting in social isolation23 and self-blame (Barnett et al.,
1996; Kurz, 1998). Taken together, this work suggests that disclosure probably

mediates a more significant causal agent rather than drives a reduction in

 

22 At the outset of the study the authors describe individuals as “extremely high-risk mothers,
many of whom are products of chaotic homes, their lives replete with evidence of their lack of
previous success in human relationships - women in whom the capacity for relationships might
well seem to have been destroyed...” (Pharis & Levin, 1991). More than 50 percent of the
women had experienced 9 or more events on an 18-item index of misfortune. Events on the
index were highly difficult life experiences. Examples included: previous experience of physical
or sexual abuse, psychiatric hospitalization, early childhood loss of significant others, and
expulsion from school. Many of the factors on the list include components associated with
considerable disruption in the parent or caregiver-child relationship.

One study quoted a woman as saying “I was not allowed to go out. I wasn’t really allowed to
talk on the phone. . . I wasn’t allowed to have a job. I wasn't allowed to have friends." (Kurz,
1998, p. 205). This level of control exerts extreme limits on the amount of disclosure that can
occur and, over time, erodes the individual’s interest in and willingness to even attempt
disclosure.

86

reported needs. Further research is needed to investigate which factors influence
disclosure, how individual differences in disclosure not attributable to those
underlying factors impact reported needs, and which reported need areas are
most effected.

Summapy.

Testing a conceptual model examining the interrelationships between
expected and reported needs among low-income pregnant women has yielded
interesting findings bridging numerous existing literatures on human behavior.
Findings from this investigation have supported the common assumption of
numerous research and family-centered care intervention models that reported
needs do reflect expected needs defined by more objective criteria. Expected
needs underlie reported needs, yielding their influence through both direct and
indirect pathways.

Reported needs also appeared to incorporate some level of readiness
within participants to address events and conditions contributing to need areas.
People were more likely to report needs when they experienced a stressor
response that perhaps heightened their perception of threat associated with
inaction or acted as a cue to focus attention toward meeting needs and/or to
trigger action. They also were prone to reporting needs when they viewed
themselves as having general competence to take efficacious action at some
point related to that issue and when benefits would be greatest due to low
existing competence in the specific issue area characteristics. Finally reported

needs were greatest among those with the least open disclosure tendencies.

87

Whether this effect was a product of differential disclosure patterns to informal
support network members as opposed to professional sources or related to
influences of life experiences and expected needs on disclosure tendencies will
need to be explored in subsequent research.

Understanding the factors that coalesce to alter an individual’s willingness
to report a need has important implications for individualizing intervention
approaches to clients with different levels of reported needs. These implications
are addressed further in Chapter 3. Chapter 2 describes the investigation of a
critical follow-up question. That is, how do expected and reported needs

influence program participation and more distal outcomes observed in families?

88

CHAPTER 2:

PREDICTING THE INFLUENCE OF REPORTING TENDENCIES ON
SERVICE UTILIZATION AND INTERVENTION PROGRAM OUTCOMES

INTRODUCTION

The previous chapter described the investigation of and findings about
factors that influenced reported need behaviors among low-income pregnant
women. These results provide critical information about relationships between
expected and reported needs and offer some insights into potential processes
that influence need perception and disclosure. However, given the paucity of
research on individual differences in reporting needs and the documented
success of family-centered care practice without attention to this issue, skeptics
might be tempted to ask what the significance of these individual differences is
for service delivery and family outcomes. ls variation in reporting needs
predictive of service utilization? How do expected and reported needs relate to
subsequent parenting and child development outcomes?

Conceptual Model for Predicting the Impact of Maternal Reporting on
Service Utilization and Intervention Program Outcomes.

Figure 6 depicts my hypothesized model for predicting parenting and child
developmental outcomes using information about maternal reporting behavior. It
examines how participant need relates to use of formal support services and
indirectly to parenting interactions and the mental development of the child. The

conceptual basis for the interrelations depicted is described in more detail below.

89

 

Relationfljps Between Needs and Service UtLization. Family-centered
early intervention programs are designed to provide varying levels of support
intensity that is commensurate with each family’s specific needs. Hence, the first
section of the model examined how needs were related to service use. One
would expect that programs devote greater intensity of intervention to the
neediest families. In fact, traditional policy initiatives often target high-risk
populations or provide additional funding for more intense service delivery when
families meet criteria suggesting especially high levels of expected needs
(Zimmerman, 1999). Also, previous studies have established that socioeconomic
status is an important predictor of increased intensity of home visiting support
service use among both rural Caucasian women and urban African-American
women (Olds & Korfmacher, 1998). Although much broader, socioeconomic
status is a proxy variable for hardship and expected needs within a population.
Thus, expected needs were predicted to directly increase maternal and infant
support service use intensity among families. However, expected needs also
were likely to have an indirect effect by positively influencing reported needs. In
the family-centered care framework, interventions should be shaped by the
reported needs, goals, and concerns of the clients. So, the reported needs also
should directly influence the intensity of services received by families. Indeed,
several recent studies found that individuals who reported having needs for
support consistent with program objectives were more likely to maintain longer

active participation in a preventive intervention programs than those who did not

90

report the presence of similar needs (Gross, Julion, & F099, 2001; lreys, DeVer,
& Chernoff, 2001; Navaie-Waliser et al., 2000).

Maternal Availapility and Service Utiligtion. Two additional factors were
hypothesized to influence maternal support service use intensity by influencing
the availability of the mother to receive the intended services. The first was the
presence of the mother in the county where the services were delivered. In
Michigan, eligibility to receive maternal and infant support services is based on
county residence. Thus, when an individual moved out of the county she may or
may not have been connected with similar services in the next county or state of
residence, but she did not continue to receive the intervention services under
study in this research. Because of this county-based service delivery method a
positive relationship between number of postnatal data collection points when the
client resided in Kent County and intensity of service utilization was expected.

Postnatal employment also was hypothesized to influence intensity of
service utilization among low-income pregnant women in the sample. Study
participants were followed from 1997 - 2000. This was a time period
immediately following radical changes in national and state social welfare policies
(Greenberg et al., 2002; Larner, Terman, & Behrrnan, 1997). Michigan
implemented new requirements that mothers receiving Temporary Assistance to
Needy Families (TANF) must work 20 hours per week beginning twelve weeks
after the birth of an infant or else their stipends would be reduced or discontinued
altogether. Thus, numerous study participants, including both those who had

been and had not been active in the workforce prior to delivery, became

91

employed postnatally. Finding and maintaining employment, transportation to
and from employment, and reliable child care were considerable challenges for
many of the participants in the first six months after delivery. The pressing time
demands following employment and conflicting schedules associated with
service provider availability (more limited evening availability) were reported by
service providers to interfere with client participation in the services being
studied“ as well as in other programs (Gross et al., 2001; Kitzman, Cole et al.,
1997). For these reasons, employment during the postnatal period was expected
to reduce the intensity of intervention services actually received by families.
Perceived Program Haljmfiness and Service Utilization. The final factor
hypothesized to influence the intensity of service use was the participant’s
perception of the helpfulness of the program. Whereas expected and reported
needs were likely to impact the intensity with which providers attempt to visit
families, the mother’s perception of program helpfulness may contribute to her
engagement in the program and to the successful completion of attempted visits
(Berlin et al., 1998). Ratings of program helpfulness may, in part, reflect a
client’s assessment of the extent to which she and the service provider(s) share
a mutual understanding about the perceived problem and goals that they hope to
jointly achieve as well as the usefulness of intervention activities in promoting
changes. These more specific characteristics that may be embedded in program

characteristics have been linked to active program involvement in other studies

 

2‘ This information is based on preliminary review of qualitative interviews with MHSS service
providers in both intervention programs. Complete analysis and dissemination of results is
expected in future months. The relationship between increased employment and reduced
participation in services also has been corroborated anecdotally in conversations with individuals
engaged in research with another home visiting program in Denver, Colorado.

92

(Epperson, Bushway, & Warrnan, 1983; Gross et al., 2001). Regardless of
whether perceived program helpfulness was an indicator of these specific
characteristic or a more general appraisal of overall satisfaction, stronger
perceptions of the program as helpful were expected to motivate individuals to
overcome potential service use barriers and to more effectively make and keep
appointments with providers.

The lrmact of State on Service Use and Parenting Interactions. The
mother’s state also has been identified as an important factor in predicting
behavior. As discussed in the previous chapter, perceived stress has been
closely linked to objective need characteristics typical of expected need
components and positively related to the expression of depressive symptoms.
The challenges imposed by high levels of perceived stress also have been
identified as negatively affecting service use intensity during the middle phases
of program participation (Eisenstadt & Powell, 1987; Gravida-Payne &
Stoneman, 1997; Gross et al., 2001; Josten, Mullett, Savik, Campbell, & Vincent,
1995). Whereas initiating contact with a new program might be seen as a positive
response to overcoming the stress at the outset of program use, the perception
may be very different midway through the intervention. At that point, the
intervention may be perceived as partially contributing to the individual’s stress
level, taxing the individual’s stress-limited problem-focused coping capabilities
(Gravida-Payne & Stoneman, 1997). In fact, qualitative interviews with nurses in
a home-visiting program with low-income parents validated this idea (Kitzman,

Cole et al., 1997). They reported a reduction in service intensity during periods

93

when clients had been emotionally stressed by broaching a difficult intervention
concern together or when participants were physically taxed by the added
stressors of employment coupled with responsibilities of caring for a young infant
(Kitzman, Cole et al., 1997).

Perceived stress also was expected to both indirectly and directly reduce
the quality of parenting interactions between mothers and their infants. Evidence
suggests that the relationship between high levels of perceived stress and
negative parenting interactions is significantly mediated through an increase in
depressive symptoms among women with heightened levels of stress (Simons &
Johnson, 1996; Simons, Whitbeck, Melby, & Wu, 1994). A large body of work
relates maternal depression to less positive parenting interactions, including both
withdrawn and intrusive interactional styles (Campbell, Cohn, & Meyers, 1995;
Field, 1998; Planos, Zayas, & Busch-Rossnagel, 1997). Studies have noted that
depression results in reduced abilities to modify speech and behavior to the cues
of infants and young children in a structured teaching task (Bettes, 1988;
Goldsmith & Rogoff, 1995). Thus, perceived stress was hypothesized to increase
the depressive symptoms expected to be directly, negatively related to high
quality parent-infant interactions in a structured teaching task.

Yet, other research supports the presence of a direct, negative
relationship between stress and positive parent-child interactions (McKelvey,
Fitzgerald, Schiffman, & von Eye, 2002). This direct relationship even has been
replicated in research that has incorporated depressive symptoms as a mediator

of interactions (Sachs et al., 1999). Other researchers might attribute this direct

94

relationship to unstudied mediators like authoritarian child-rearing beliefs
(Conger, McCarty, Yang, Lahey, & Kropp, 1984), ease of anger expression
(Rodriguez & Green, 1997), or negative changes in interpersonal relationships
within the broader family unit (Crnic & Acevedo, 1995). However, in each case
when these factors were not measured, stress was expected to produce some
negative impact unrelated to depressive symptoms that would generate a direct
negative relationship between stress and parenting interactions when conceptual
model 2 was tested.

Predicting the Quality of Parent-Chiltflnteragions: Infidence_§_o_f
Intervention, Child Cues, and the Mothers’ Self-Reporting of Nefl Research
on the quality of parent-child interactions has indicated the importance of parents’
abilities to notice and interpret each child’s cues and to adapt their behavior to
meet the physical, social, emotional, and cognitive needs of the child within the
shared context of the interaction (Sumner & Spietz, 1994). Earlier, I described
how the presence of depressive symptoms may interfere with cognition-emotion-
action patterns that influence maternal behavior and interpersonal interactions.
However, other factors also may influence the extent to which high-quality
parent-child interactions are displayed.

The clarity of the child’s cues and his/her responses to parent behaviors
are important predictors of positive parent-child interactions. The skill and clarity
with which infants send cues to their caregivers influence how easily individuals
can read and respond to those messages (Sumner & Spietz, 1994). Caregivers

tend to have more difficulty maintaining positive interaction styles with infants

95

who display ambiguous cues or who provide few subtle cues about their
changing needs prior to displaying potent responses (Minde, 2000). Studies
have found that children with medical complications following birth, serious health
conditions, or who were born prematurely are more likely to display unclear or
disorganized cues in interactions (Parke & Tinsley, 1983). Research also has
shown that caregivers experience greater frustration during interactions with
these high-risk infants (Beckwith, 1984). Medically high-risk and preterm infants
also are documented to experience higher rates of child maltreatment, perhaps
partly in response to difficult interaction sequences (Kotch et al., 1995; Sachs et
al., 1999). This evidence supports the prediction of a direct positive association
from child health to the clarity of cues displayed during caregiver-child
interactions and a direct positive effect of clarity of cues on the overall quality of
parent-child interactions.

Another factor hypothesized to show a direct positive relationship to the
quality of parent-child interactions is the degree of matching between maternal
reported and expected needs. An important element in high quality parent-child
interactions involves the parent’s accurate perception of both the child’s needs
and the task demands and the parent’s ability to tailor his/her behavior to provide
needed assistance (Rogoff, Ellis, & Gardner, 1984). Although previous research
has not investigated the specific relationship between parental reports of high
proportions of expected personal needs to their subsequent interactions with
their children, the hypothesis of a positive relationship extends from the belief

that mothers who can more effectively define and report their own needs may be

96

better at perceiving and responding to their children’s needs as well. Research
from the area of emotion regulation lends some support to this notion. Several
studies have found that individuals who have difficulties labeling and discussing
their own emotions show similar difficulties in labeling, discussing, and
responding to the emotions of their children (Dunn & Brown, 1994; Gottman,
Katz, & Hooven, 1997; Parke & Buriel, 1998). The action patterns established in
their own behavior also were employed in parenting behaviors. Another line of
research that examined families with dysfunctional interaction patterns also has
noted a frequent breakdown in parents’ appropriate perceptions of their children’s
needs. For example, after a general lack of cooperation, the second most
frequently cited barrier interfering with children in CPS caseloads receiving
needed services was “family’s misunderstanding of child’s needs” (Trupin,
Tarico, Low, Jemelka, & McClellan, 1993).

A number of interventions have been developed to help parents become
more adept at interpreting and responding to children’s cues (Barnard et al.,
1987; Kang et al., 1995). Although variation exists in the processes used to help
families develop more positive parent-child interactions, many interventions,
including both of the interventions in the Michigan Maternal Health Services
Study, expect that program involvement will promote more positive parent-child
interactions (Roman et al., 2001; Zeanah & McDonough, 1989). The most
positive outcomes particularly are expected among individuals who remain
actively engaged in the program long enough for the focal intervention

components to be delivered. So, as shown in Figure 6, service use intensity was

97

expected to have a direct positive effect on the quality of parenting interactions
of participants.

Predictors of Mental Develoament in Children. Over the last several
decades, studies have demonstrated links between the quality of parenting and
child development (Baumrind, 1993; Maccoby & Martin, 1983; Maccoby, 1992).
Taken together, research suggests that parenting styles characterized by
warmth; clear, responsive communication; inductive reasoning; and appropriate
monitoring; foster improved cognitive functioning, developmentally appropriate
social skills, and psychological adjustment of children (Simons & Johnson, 1996).
The Nursing Child Assessment Teaching Scale (NCATS) has been widely
validated and utilized as an indicator of the responsive parent-child interaction
component of positive parenting styles (Barnard & Kelly, 1990; Sumner & Spietz,
1994). Evidence suggests that high quality parenting interactions as measured
by the NCATS are positively related to home environment ratings on the HOME
scale (Sumner & Spietz, 1994), mental development index scores on the Bayley
Scales of Infant Development (BSID) (Barnard & Eyres, 1979; Sumner & Spietz,
1994), and indices of IQ at older ages (Bee et al., 1982). Given the findings of
prior research, the quality of maternal parenting interactions was expected to be
directly and positively related to the quality of the home environment and to the
child’s mental development. An indirect relationship was also hypothesized as a
result of the positive association between responsive parent-child interactions

and the presence of a high quality home environment.

98

Provision of nurturing home environments also was predicted to have a
direct, positive impact on children’s mental development. Especially during
infancy, the child’s home environment forms the primary context where young
children interact with people and objects, Ieam to solve problems and develop
understandings of symbols and concepts. Home environments containing high
levels of responsive and non-restrictive parent involvement, developmentally
appropriate play materials, organization, and variety have been associated with
improved cognitive development in children (Bradley, 1993, 1994; Bradley 8
Caldwell, 1988).

In addition to parenting interactions and the contextual environment,
evidence suggests that children’s mental development is also influenced by
several key child characteristics. lnfants’ health was expected to positively
impact their mental development. Studies have demonstrated that preterm
infants and children with medical complications after birth tend to score lower on
standardized measures of infant mental development (Field et al., 1978;
Widerstrom & Nickel, 1997), although many preterm infants reared in highly
supportive environments “catch up” to their full-term peers by school age
(McCarton, Brooks-Gunn, Wallace, Bauer, & al, 1997). Research also has found
that infants who experienced persistent, reoccurring middle ear infections often
display less persistence at tasks and show poorer performance in language skills
embedded in mental development assessments (Feagans & Proctor, 1994;

Feagans, Sanyal, Henderson, Collier, & Appelbaum, 1987; Teele et al., 1990).

99

So, it is important to consider the health history of infants when trying to predict
their mental development outcomes.

In addition to health characteristics, the child’s behavior during the
assessment situation also may impact mental development outcomes.
Children’s abilities to maintain orientation and engagement throughout tasks and
to regulate their emotions effectively are particularly likely to influence observed
behavior in a standardized infant assessment situation (Sattler, 1992). Indeed,
some researchers have identified that the behavioral dimensions assessed on
the Bayley Scales of Infant Development (BSID) behavior rating scale account
for the largest proportion of individual variation in mental scores (Matheny, 1980)
and that optimal behavioral functioning during the evaluation is characteristically
associated with higher cognitive performance of children in the assessment
(Matheny, Dolan, & Wilson, 1974; Roth, Eisenberg, & Sell, 1984). So, more
effective orientation, engagement, emotion regulation, and motor quality during
the test was hypothesized to be directly, positively related to mental development
scores on the BSID.

Intervention Grodp Differences.

The conceptual model described in Figure 6 was tested separately for
each intervention group in the Michigan Maternal Health Services Study. No
between group differences were expected in relationships between parenting
interactions, home environments, and child health and development. However,

intervention group differences were probable in predictions of number of

100

intervention visits and perhaps in the relation of service use to parent-child
interaction measures.

Programmatic differences were hypothesized in the path from perceived
helpfulness of services to intervention intensity. The Nurse-CHW Team care
intervention included a relationship-oriented intervention approach that built on
the similarity of the CHW to the empower the participant, help her establish and
maintain healthy, supportive relationships with others in her environment
(including, but not limited to, her infant), and provide timely client-centered
support, information, modeling, and referral on intervention content issues in the
context of this relationship. Given the careful attention devoted to relationship
issues and client needs in the Nurse-CHW Team care model, I hypothesized that
assignment to the Nurse-CHW Team care intervention would contribute to higher
levels of perceived helpfulness of services. In addition, there was considerable
persistence of the Nurse-CHW Team care group in finding and repeatedly
attempting to meet with individuals who missed appointments. Hence, in addition
to level differences in perceived helpfulness of and actual receipt of services, I
hypothesized a stronger direct positive relationship between perceived
helpfulness of services and service use intensity in the Nurse-CHW Team care
group than in the SOC group.

Both treatment groups were expected to have comparable influences of
county residence, employment, and maternal state on service use intensity. If
any intervention group differences were predicted in these areas it would occur in

the incidence of employment and the levels of state conditions as outcomes of

101

prenatal intervention rather than a differential process impacting service use
intensity.

Since both interventions espoused a client-centered case management
approach embedded in the client’s ecological environment, I did not hypothesize
treatment group differences in relationships between expected and reported
needs and service use intensity. Differences that might exist were likely to be
primarily a function of individual service provider variation in implementation of
the intended intervention as opposed to conceptual formulation of intervention
plans between the programs.

Finally, a critical link in conceptual model 2 (see Figure 6) involves the
relationship between increased intensity of intervention and improved mother-
infant interaction. The relationship-based focus of the Nurse-CHW Team care
intervention was expected to promote greater sensitivity and higher quality
interactions in Nurse-CHW Team care mothers than in SOC mothers within the
parent-infant relationship. This effect was expected to be most pronounced
among those engaged more intensively in the ongoing intervention relationship
(i.e., those with higher levels of service use). However, the Nurse-CHW Team
care plan identified maternal outcomes as primary focal points of intervention
with the belief that these changes across maternal domains would support
improved parenting and child development. In contrast, SOC identified the
mother as the primary client through pregnancy and in the first two months after
birth. But, by eight weeks after delivery, the infant had to be identified as the

primary client in order to continue delivering services. While this is a subtle

102

distinction and there was considerable overlap in the nature of activities in both
programs, the slightly greater focus on infant outcomes in the SOC group might
minimize expected intervention group differences actually observed. Taken
together, participants randomized to the Nurse-CHW Team care group were
expected to display a small, but somewhat stronger positive relationship between
service use intensity and mother-child interaction than was observed among
individuals in the SOC group.

Summag.

Despite the possibility that increased understanding about patterns of
reporting needs might yield important information about service utilization and
intervention outcomes, existing work examining the influences on and the
implications of reported need behavior remains in its infancy. The model in
Figure 6 provides hypothesized interrelationships for understanding how needs
and reporting styles impact service utilization, parenting interactions and
subsequent mental development outcomes of children. Maternal state, child
characteristics, and the contextual features of the home environment also have
been identified as key factors to consider in predicting child-related outcomes.
Examining the extent to which the observed data fit the predicted
interrelationships in this model will increase knowledge about if and how different
tendencies in reporting needs ultimately influence program participation and

subsequent developmental outcomes in low-income children.

103

METHODS

Sam Ie.

The sample for testing the model of conceptual relationships identified in
Figure 6 included a subsample of 438 participants of the Michigan Maternal
Health Services Study. Information about recruitment, eligibility, exclusion
criteria for secondary analyses, and descriptive characteristics of the sample
were described in Chapter 1.

Data Collection.

Data were collected from study participants at enrollment into the study
(prior to treatment group randomization), again at 34-38 weeks gestation, at 6-12
weeks after delivery, and again 6-7 and 12-14 months after delivery. At the final
two data collection points, assessments of infant development, mother-child
interaction, and the characteristics of the home environment were conducted.
Medical records were abstracted for infant health characteristics at birth and
program records were abstracted for contact with program providers. Data
collected at 6-12 weeks and 6-7 months after delivery form the primary sources
of information utilized to test the relationships between needs, service utilization,
and intervention program outcomes as shown in conceptual model 2 (see Figure
6) how reported needs influence intervention outcomes.

Measdrement of Variables.

Expected and Refined Need lndices. The expected and reported need

indices utilized to test the conceptual model in this chapter were identical to that

104

developed and described in detail in Chapter 1. The actual index items are listed
in Table 2.

wee of Matmgetween E_xpect<i1 anflmrted Needs. Two different
measures relating expected and reported needs were developed. Difference
scores were calculated (e.g., reported needs — expected needs) to assess the
distance of reported needs from the expected index score ascribed to them.
Probablility scores were also created by dividing reported needs by expected
needs to determine the proportion of their expected needs that individuals
reported.

Perceived Stress. Perceived stress was measured using the 14-item,
Perceived Stress Scale (PSS) developed by Cohen, Kamarck, and Mermelstein
(Cohen et al., 1983). Specific reliability and validity information about this
instrument was provided in Chapter 1. Study participants were interviewed about
their responses to scale items at 6-12 weeks and 6-7 months after delivery.
Scores at the two time points were summed to yield a postnatal perceived stress
score.

Qagessive Symptoms. The Center for Epidemiologic Studies —
Depression (CES-D) Scale was used to measure the occurrence of depressive
symptomology in the study sample (Radloff, 1977). Further information about
scale development, validation and reliability was described in Chapter 1.
Individuals were interviewed about their responses to scale items at 6-12 weeks
and 6—7 months after delivery. Scores at the two time points were summed to

yield a postnatal depressive symptoms score.

105

Perceived Helpfulness of §ewices. An author-designed question was
used to measure individual perceptions about the helpfulness of services. Study
participants were asked to think about the period since they last had been
interviewed for the study. Then they were asked the following question: “Using a
scale from 1 to 10 where 10 is extremely helpful and 1 is not at all helpful, how
helpful do you think the maternal/infant support services have been for you?”
Scores reported at 34-38 weeks gestation and at 6-7 months postnatally were
summed to yield a overall rating of service helpfulness throughout the course of
the intervention.

Postnatal Employment Statda. At 6-12 weeks and again at 6-7 months,
individuals were asked whether or not they were currently employed.
Regardless of the number of hours or type of work reported, individuals who
reported employment outside of the home at either or both time points were
coded as 1 (participating in postnatal employment). Those who were not
employed outside the home at both time periods were coded as 0 on the
dichotomous postnatal employment status variable.

Postnatal Residence in Kent County. Research interview records were
abstracted to determine whether each study individual lived in Kent county
(regardless of whether or not research data were successfully collected) at each
of the three postnatal data collection points. Individuals were given one point for
each data collection wave at which they resided within the county. All 438
individuals resided within Kent County at both enrollment and 34-38 weeks

gestation, so prenatal variation did not exist in this variable.

106

MSSIISS Postnatal Service Use Intensity. Service use was measured by
the number of support service provider (either nurse or community health worker)
visits between delivery of the infant and one year later”. Both programs
anticipated engaging clients in a greater number of home visits in the first 6
months after delivery. Clinical-paths showed many fewer visits in the second half
of the year. Conversations with service providers revealed that visits
substantially scaled back after six months both as a function of clinical path
recommendations and due to greater frequency of missed appointments
associated with employment conflicts, increased transience, or reduced
commitment to the program.

Since there were several cases with extreme values for service use visits,
an alternative measure of service use also was constructed. Rather than using
the actual number of visits received, individuals were ranked ordinally in amount
of postnatal service use. So, for example, if a set of individuals received 0, 0, 1,
5, and 12 visits, they were ranked 0, 0, 1, 2, and 3 respectively.

_Ch_ild’s Mental Development. Infant mental development was measured
by using the mental development subscale of the Bayley Scales of Infant
Development (BSID). The BSID is the most widely used assessment tool for infant

development (Clark, Paulson, & Conlin, 1993). The instrument includes items that

 

25 I had initially planned to use a measure of visit contacts from birth to six months, however, at
this time that data is neither available nor retrievable without the assistance of a colleague’s
creation of a program to extract the data from its current format. This program will not be
available until Fall 2002. Modification of the model to look at 12 rather than 6 month outcomes
was not possible because final developmental and parent-child interactions data will not be fully
cleaned until Mid-Fall 2002. Thus I opted to use overall postnatal service use as a best estimate
for the intensity of service use individuals received during the first half of that time period. As
soon as the actual contact data becomes available, I will rerun the model with the new variable
and explore any differences that arise.

107

assess memory, habituation, problem solving, generalization, classification,
vocalizations, language, and social skills. Scale administrators assess these skills
by providing infants with situations and tasks that capture their interest and produce
an observable set of behavioral responses. The BSID mental development
subscale has shown high reliability at six months (.92) and at 12 months (.88)
(Bayley, 1993). The BSID’s restandardization based on the 1988 US. Census data
has strengthened its accurate representation of every demographic group and
makes it an excellent measure of infant development (Bayley, 1993). Mental
development was assessed typically around 6-7 months of age and converted into
mental development index scores. Infants born more than two weeks prematurely
were assessed at corrected age.

Child’s Test Behavior. The child’s behavior during the developmental
assessment situation was measured using the total score on the Behavior Rating
Scale (BRS) of the Bayley Scales of Infant Development (BSID). The BRS was
designed to assess qualitative aspects of the child’s test-taking behavior, including
orientation/engagement (toward the tasks, examiner, and caregiver), emotional
regulation, and quality of movement (Bayley, 1993). The total score incorporates
the child’s overall behavior in these three areas during the test situation. The
Behavior Rating Scale has shown high reliability at six months across each of the
subscales (6 month - .87, .75, .83; 12 month - .84, .86, .79) and the total score as
well (6 month - .88; 12 month - .90) (Bayley, 1993).

Q_uajty of Parenting Interactions. The quality of parenting interactions was

measured using the caregiver interaction score on the Nursing-Child Assessment

108

Teaching Scale (NCATS) at 6-7 months after delivery. The NCATS was
developed by Katherine Barnard and her colleagues to assess the behaviors,
content, and responsiveness observed in maternal and child contributions to a
structured social interaction task (Sumner & Spietz, 1994). The caregiver
interaction score includes 50 binary items grouped into four conceptually-derived
subscales assessing caregiver sensitivity to cues, response to distress, social-
emotional growth fostering, and cognitive growth fostering. The NCATS is
widely-used as a research and clinical tool. Internal consistency has been
reported as .87 in mixed groups and in subgroups of participants of color
(Sumner & Spietz, 1994). Test-retest reliability on caregiver scores (total from
four subscales) at 1, 4, 8, and 12 months of age was .85 (Sumner & Spietz,
1994). The instrument has differentiated between groups of high- and low- risk
mothers and was related to subsequent child cognitive outcomes and parent-
child interaction quality (Barnard & Morisset, 1995; Bee et al., 1982; Booth,
Barnard, Mitchell, & Spieker, 1987; Farran, Clark, & Ray, 1990). Administrators
of the NCATS all received training and national certification (indicating a reliability
of greater than .85 on standardized videotapes). Inter-rater reliability in the home
setting by research interviewers was repeatedly assessed throughout the study
duration and consistently exceeded .85 in each case.

Clarityof Child’s Cues. The child’s contribution to the mother-child
interaction was measured by a subscale on the NCATS measure titled “Clarity of
Child’s Cues”. The subscale was designed to assess the extent to which the

child provided readily observable nonverbal and/or verbal reactions to events

109

occurring during the teaching interaction (Sumner 8 Spietz, 1994). Internal
consistency reliability on this ten item scale has been low (Cronbach’s alpha =
.50) (Sumner 8 Spietz, 1994). This finding is probably attributable to the wide
range of cues the scale measures . For example, “child is awake”, "child smiles
or laughs during episode”, “child displays potent disengagement cues during the
interaction”, and "child makes clearly recognizable arm movements during the
teaching” all serve as cues about the child’s interests and experiences to the
mother. But, infants who exhibit one behavior will not necessarily exhibit all of
these behaviors during brief teaching situations with considerable variety in
events and maternal interactions. Thus, one would expect lower internal
consistency in this measure than in more commonly utilized attitude scales
where responses would cluster together more tightly. Despite considerable
changes in infant development between 1, 4, 8, and 12 months of age, test-retest
reliability for the infant scale (including the clarity of cues subscale) was .55
(Sumner 8 Spietz, 1994). Evidence suggests that the clarity of cues subscale is
effective at identifying expected differences between preterm and full-terrn infant
cues and changes in interactive behavior among twins depending on the
presence of the other twin in the interaction (Barnard, Eyres, Lobo, 8 Snyder,
1983; Sumner 8 Spietz, 1994). Inter-rater reliability in the home setting by
research interviewers was repeatedly assessed throughout the study duration
and consistently exceeded .85 in each case.

Qpality of Home Environment. The quality of the home environment was

assessed at 6-7 months using the Infant/Toddler Home Observation for

110

Measurement of the Environment Scale (HOME) (Bradley 8 Caldwell, 1978).
The HOME is a 45 item semi-structured interview and observational tool that
assesses 5 subscales describing the characteristics of the home environment in
which an infant is being nurtured. Subscales include emotional and verbal
responsivity, acceptance, organization of the physical environment, provision of
appropriate learning materials, parental involvement with the child, and
opportunities for variety in daily stimulation. The HOME has demonstrated
acceptable levels of reliability and has a long established record of validity
through associations with child cognitive and language development, positive
parenting behavior, and positive maternal-child interactions in research (Bradley,
1994; Bradley et al., 1989; Bradley, ConNyn, McAdoo, 8 Garcia Coll, 2001 b;
Coon, Fulker, DeFries, 8 Plomin, 1990; Sumner 8 Spietz, 1994). Over time the
HOME has showed consistent associations with both parental educational and
socioeconomic status (Bradley, Conivyn, McAdoo, 8 Garcia Coll, 2001a; Lotas,
Penticuff, Medoff-Cooper, Brooten, 8 Brown, 1992; Sumner 8 Spietz, 1994). It
also has demonstrated reliability in culturally diverse samples (Bradley et al.,
2001a; Lozoff, 1995; Seidman et al., 1994). Inter-rater reliability in the home
setting by research interviewers was repeatedly assessed throughout the study
duration and consistently exceeded .90 in each case.

Child’s Health. The child’s health was measured through the creation of a
six item index representing the presence or absence of conditions associated
with greater risk for poorer health and developmental outcomes. Information was

drawn from the medical record abstraction unless otherwise noted. Cases were

111

assigned one point for the presence of each of the following conditions: 1) < 259
days gestation at birth based on report of last menstrual period (LMP) or
ultrasound if results from an ultrasound conducted prior to 20 weeks gestation
were present in the medical record and the estimated due date was at least two
weeks discrepant from the due date determined by LMP, 2) < 2500 grams at
birth, 3) infant’s discharge from hospital was three or more days after delivery, 4)
infant was admitted to the neonatal intensive care unit for one or more days
during hospital stay, 5) child was diagnosed with a major medical condition
associated with poorer developmental outcomes (e.g., Hirschsprungs disease,
hole in baby’s heart, cystic fibrosis), 6) child was diagnosed with a medical
condition placing the child at risk, but typically associated with less serious
developmental consequences for the child (e.g., asthma, more than three ear
infections diagnosed between birth and six months). Scores ranged from zero to
six with six representing more serious health problems.

Whereas the aforementioned infant health index assessed congenital,
neonatal, or serious long term health issues, a second, two item, health index
was constructed to identify the infant’s more proximal health status. Children
received one point if mothers reported that the infant had been sick with a cold or
upper respiratory infection in the week just prior to the developmental
assessment. Cases were assigned an additional point based on maternal report
of moderate to poor child health from birth to six months. Thus, scores of more

proximal child health ranged from zero to two.

112

Data Analysis.

Missing Data Imitation. Missing data were imputed using PRELIS 2.
Out of the 21,900 data points used in analyzing the second conceptual model,
only 827 (3.78%) missing values were imputed. Even after the imputation
process, every variable contained more than 85% of original non-missing values.
Imputation was conducted separately for each intervention group. Data were
imputed using the hot deck method (Ford, 1983; Schoier, 1999). In the hot deck
method, the analyst defines a set of variables that describe key demographic or
psychosocial characteristics that are highly indicative of the individual’s situation
and significantly related to the outcome variable with missing values. During hot
deck imputation, existing cases with complete data are examined to find a
matching response pattern on the identified variables for the case with the
missing value(s). Once a unique match is found, the existing values from the
complete case are imputed to replace the missing value(s) in the twin case. The
use of these match characteristics minimizes the effects of bias if cases with
missing values tend to be different in important characteristics from the sample
mean or from overall group characteristics. The hot deck method has the benefit
of using relationships within the actual collected data to make estimates for
missing values. In addition, the imputed score also has some error variance
imputed into it since it uses an actual score already found within the data set
(Roth 8 Switzer, 1999). Appendices M and N list the variable means and
standard deviations for each intervention group before and after missing data

imputation. Careful analysis of this information as well as the correlations

113

between variables demonstrates that missing data imputation did not significantly
alter the characteristics of the data utilized in this study.

Analflical Methods.

To investigate the pattern of relationships identified in the second
conceptual model (See Figure 6), data were separated by intervention group
assignment. Baseline characteristics of individuals in each of these intervention
groups are shown in Table 4. No significant differences were found between
groups at enrollment. So, any differences between model characteristics were
expected to result either from differences in program characteristics or from
variation in client characteristics as a result of participation in the intervention.

The hypothesized conceptual model was investigated using path analysis.
Specifically, the observed covariance matrices in each of the intervention groups
were compared to what would have been expected given the set of
interrelationships depicted in the hypothesized model. The estimated paths and
fit characteristics for each intervention were compared against a model with
invariant gamma and beta matrices (same structure of relationships and same
path estimate values) and later against a model where the gamma and beta
matrices displayed the same pattern of relationships and the same starting
values but were free to vary in final path estimate values. The intervention group
comparison necessitated use of a manifest model since the sample size of each

group did not contain sufficient power to analyze a structural model.

114

RESULTS

Manifest Model.

Model adaptations. The only major model adaptation conducted to
optimize fit was the removal of the degree of match between expected and
reported needs variable from the model. An oversight in the operationalization of
this variable was that both the difference scores and the probability scores
constructed to measure this phenomenon were completely linearly dependent on
expected and reported needs variables. Thus the inclusion of all three related
variables in the model created substantive problems in the parameter estimation
process. In future studies, I plan to investigate differences in variable
relationships across groups with different patterns of reported and expected
needs. 80, even with removal of the expected-reported need match variable
from this model, I will eventually be able to address this question through an
alternative methodology.

Model fit. Overall the hypothesized model fit the observed data from both
intervention groups well (See Figures 7 and 8 and Appendix 0). Both groups
shared the same pattern of relationships between variables and utilized the same
starting values. Appendices P and Q list the error terms for the observed
variables. The normal theory weighted least squares chi-square (df= 156, ng1 =
217 n92 = 221) was 117.73 (p = .99), suggesting that the model fit the data very
well. Other fit indices further supported strong overall model fit (RMSEA = 0.0,
CFI = 1.0). Examination on individual group fit suggests that the two groups

provided fairly similar contributions to the overall chi-square (91 = 69.41

115

(56.57%), 92 = 53.28 (43.43%)) and the goodness of fit index supported a model
with close fit in both groups (91 GFI = .96; 92 GFI = .97). Use of the invariant.
constraint on beta and gamma estimates between groups also generated a
model with acceptable overall fit (x2 (df= 175, ng1 = 217, ng2 = 221) = 157.25;
RMSEA = 0.0; CFI = 1.0) and within each group (91 )8 contribution = 94.17
(58.90%), GFI = .94; 92 x2 contribution = 65.70 (41.10%), GFI = .96). However,
selection of the more parsimonious model of invariant estimates resulted in the
adoption of a significant path estimate in each group based on the overall model
that was not present when estimates were modeled separately for each group.
Observed effects. The adopted model included statistically significant
hypothesized path estimates for both groups in the expected directions between
the following variables: expected needs and perceived stress, expected needs
and reported needs, reported needs and service use, postnatal employment and
service use, duration of residence in Kent County and service use, maternal
teaching interactions and home environment ratings, and infant behavioral
ratings on the test day and infant mental development scores. Expected
influences of perceived service helpfulness on enhanced service use only
reached statistical significance in the Nurse-CHW Team Care group. The
expected direct, positive effect of clarity of infant cues on maternal teaching
interactions only reached statistical significance in the standard of care group.
No paths were found to have statistically significant relationships in
directions contrary to predictions. However, surprisingly, a number of expected

relationships were not modeled in this sample that deserve further attention. Of

116

particular interest was the finding that expected needs did not influence intensity
of service use except indirectly by increasing reported needs. Also surprising
was the lack of influence of home environment ratings or of maternal teaching
interaction characteristics on infant mental development scores. The
hypothesized relationship between poor infant health and both the infant’s
expression of cues during a teaching interaction and his/her performance on the
mental development assessment also was not supported. These paths yielded
very small estimates of relationships between health index scores and these
outcome variables in both intervention groups. In addition to nonsignificant
effects observed in child and parenting outcome variables, no statistically
significant relationships were found between perceived stress or depressive
symptoms and maternal teaching interaction ratings. While not statistically
significant in either group, path estimates for both groups did suggest a negative
relationship between reported depressive symptoms and observed caregiver
scores in an infant teaching interaction setting. Finally, nonsignificant
relationships were found in both groups between service use and maternal
teaching interactions. Heightened service use did not necessarily lead to better
outcomes for study participants.

Quaratory Analyses.

After careful examination of the model, several alternative explanations for
the pattern of relationships observed were considered. In some cases, data
were available which permitted exploratory examination of the influences of these

alternative distributions of variables or potentially confounding variables on the

117

outcomes of interest. When possible, these variables were included in the model
individually to examine their effects.

Service use. The variable for ordinal service use was replaced with the
actual number of postnatal visits recorded for each case. This exchange had no
meaningful effect on path estimate or even the overall chi-square. Exploratory
analyses subsequently were conducted to explore whether maternal teaching
interactions were influenced by effective service use modeled as a u-shaped
distribution with highest and lowest intensity clients expected to have poorer
outcomes than more moderate users. Data on number of postnatal visits were
centered using mean and median values for each intervention group. Then, the
absolute value of the centered variable was utilized. This resulted in both those
with high and low numbers of visits relative to the mean and median values
possessing high values on the new variables and individuals nearer the mean
and median possessing low values. Thus, a negative path estimate would have
been consistent with prior hypotheses because lower values on these new
variables (i.e., moderate levels of service use) would be expected to improve
maternal interaction outcomes. Entry of these new variables in the model
replacing intensity of postnatal intervention visits did not alter the nonsignificant
relationships in both groups between service use and maternal teaching
interactions. Standardized path estimates from the revised service use variable
to maternal teaching interactions were small as well (Nurse-CHW Team + .06
using both mean and median-centered variables; SOC -.03 using mean

centered, -.09 using median centered).

118

Infant health. The two-item proximal health variable was inserted in the
model to predict infant cues and mental development. Inclusion of this variable
either in addition to or in place of the six item health index yielded no significant
influences on the expected outcome variables. Modifications also explored the
influence of the six item health index on infant behavior in the test situation (i.e.
poor health may undermine infant attention, motor quality, and effective
engagement in tasks). However, no statistically significant influence was
observed in either treatment group.

Infant sex. Infant sex also was entered into the model to explore any
potential moderating role in the influence of six item infant health on infant cues,
the influence of infant cues on maternal teaching interactions, or the impact of
any of the predictor variables on infant mental development. However the
inclusion of infant sex in the model did not alter the significance threshold or
meaningfully alter the directions or sizes of any pathways considered.

Maternal education. Finally, the model was also re-run with the addition
of maternal education (years completed in school) included as a predictor of
postnatal job status, NCATS scores, HOME scores, and MDI scores. The
inclusion of maternal education resulted in a model with good fit characteristics
(Chi-Square (df=176, n=438) = 151.18, p = .91; RMSEA = 0.0; CFI = 1.0; Nurse-
CHW Team Care GFI = .95, SOC GFI = .96). Statistically significant paths were
found in both intervention groups indicating influence of maternal education on
postnatal employment (Nurse-CHW Team care = .22; SOC = .31) and HOME

scores (Nurse-CHW Team care = .26; SOC = .22). Significant paths between

119

education and NCATS scores were found only in Nurse-CHW Team care group
(Nurse-CHW Team care = .19; SOC = .12). The presence of significant paths
from education to NCATS and from education to HOME weakened relationships
between NCATS and HOME in both treatment groups. Whereas the value of this
path dropped below the level of statistical significance in Nurse-CHW Team care
group, it remained above this threshold in the SOC group (Nurse-CHW Team
care = .13; SOC = .18). No statistically significant relationship was modeled
between maternal education and infant MDI scores in either group (Nurse-CHW
Team care = .01; SOC = -.01). The inclusion of maternal education in the
model also did not alter the significance of paths from NCATS to MDI scores or

HOME to MDI scores.

120

DISCUSSION

The second phase of this investigation was designed to examine how
expected and reported needs influenced service utilization and maternal and
child outcomes of program participation. Were the number of intervention visits
completed related to more objective criteria about an individual’s circumstance or
more closely tied to her perceptions of her circumstances? What were the
implications of differential intensity of service use on selected maternal and child
outcomes? The findings may contribute to ongoing debates about influences on
program participation in home visiting interventions and challenge the field to
more carefully define which individuals benefit from home-based interventions.

lnflggnces on Service Utiliaation.

Qpected Needs. Of particular interest was the finding that expected
needs did not influence service use except indirectly by increasing reported
needs. In both intervention groups, study participants with higher levels of
expected needs were not significantly more likely to receive a higher dosage of
intervention. Thus, clients with multiple problems and risks were not engaged in
a higher number of service provider contacts to address those issues. This
finding suggests that since service providers utilized a similar number of contacts
in treating lower expected risk and higher expected risk clients, then they must
either have diminished the time spent to address each client need in high risk
cases or they simply did not address all of the need areas in these cases. Either
of these alternatives probably did not contribute to the best possible outcomes

with families at highest need.

121

The reasons why intervention dosage was not greater for families with
more expected needs are open to speculation. However, all of the individuals
who met eligibility criteria for study participation had high levels of need relative
to the general population. Individual differences of need within this high need
group may have been relatively weaker than differences observed in a broader
population including individuals from all socioeconomic levels. Thus, it is
possible that expected need would have displayed a stronger direct effect if
sample eligibility had not restricted the range of need observed.

An alternative explanation for the lack of influence of expected needs on
intervention use involves the differential responses that individuals with high
expected needs might have displayed. Whereas, some clients at highest need
may have been highly attention-seeking, others with these characteristics may
have displayed high avoidance patterns. Berlin and her colleagues (1998) noted
these two opposite patterns among high-risk clients, stating that “...although in
some instances more vulnerable participants appear to derive more program
benefits, they may not necessarily be described as easily served by program
staff. Frequently, the more vulnerable participants are hard to reach and difficult
to engage. Alternatively, more vulnerable participants may be more visible and
more demanding, and, as a result, elicit more attention from staff and more
program services” (Berlin et al., 1998, p.11). The combination of these two
different responses among those at highest expected need may have hidden the

appearance of effects of expected need on actual intensity of service use.

122

@orted Needs. Whereas expected need index scores did not
significantly predict service use in either treatment group, increased reported
need scores were related to greater use in both interventions. In addition, the
strong positive path estimate from expected needs to reported needs found in
chapter one was replicated. So, heightened treatment intensity was provided in
cases where objective indicators suggested clients had more needs and clients
acknowledged the presence of these concerns. These results were consistent
with research that has found individuals with higher perceived needs related to
program goals were more likely to maintain long-term, consistent participation in
an intervention than those who reported lower needs in these areas (Navaie-
Waliser et al., 2000). The findings also supported family-centered care practice
models in which services are tailored to meet expressed client needs, thereby
promoting changes in objective circumstances facing the client. Thus, the
significant path from reported needs to service use in both groups lends
credence to the effectiveness of both interventions in attending to client concerns
and implementing the intended client-centered intervention models.

Chapter one identified a number of factors related to whether or not
individuals reported needs that were expected. It was posited that, in
combination, these factors may represent the client’s “readiness” to label a need
as a first step in committing to act or to change the circumstance creating the
need. If differential readiness to accept help and modify behavior partly was
embedded in the client’s tendencies to report needs, then the significant positive

path from reported needs to service use could have represented the client’s

123

willingness to “take” the intervention. In this alternative explanation, the
individual’s acceptance of the intervention influenced differential service use
among those with higher needs rather than differential treatment intensity
provided by service providers in response to heightened reported needs. Using
existing data, I cannot discern which of these processes accounted for the
positive effect of reported needs on service use. However, it is likely that future
investigations designed to more carefully consider these issues will reveal that
both the readiness of the client and the service provider response to identified
needs contribute to the achieved intensity of home-based treatment completed.
Regardless of whether increased service delivery to those with higher
reported needs is indicative of service provider behavior and/or client readiness,
findings in this model suggested that individuals who were less able or willing to
report their needs received fewer services. So, individual differences in reporting
needs did make a difference in the amount of help people received. Secondly,
this finding raised concerns about existing methodologies. Frequently, a target
population with high expected needs is recruited, offered intervention, and
assessed to document the effectiveness of the intervention in meeting their
needs. But, if only a fraction of the recruited population actually reported needs
that triggered increased intervention, then the effectiveness of treatment among
those who reported their needs may have been masked by the numerous
individuals receiving a different treatment process because they did not report
expected needs. This concern challenges researchers and program planners to

examine intervention outcomes separately for individuals who report needs

124

compatible with program goals and for those who do not. Future research and
interventions may need to incorporate the client’s readiness for treatment into the
type of intervention delivered and the type of expected outcomes expected and
examined.

Maternal Availapiligr to Receive Services. Two factors related to the
participant’s availability to receive services were expected to influence the
number of home visits completed. Both employment and continued residence in
the county displayed significant effects in the expected directions in both
treatment groups.

Postnatal employment. In both intervention groups, individuals who
were employed at some point between birth and 6 months after delivery received
fewer visits from service providers. Employment was operationalized not to
represent continuous employment during this period, but rather, reported
employment at some point during that time. Thus, employed individuals included
both those who maintained consistent employment in the same job and those
who cycled between various jobs and in and out of work. Given this operational
definition and the increased demands for individuals receiving TANF to be
employed, it seems unlikely that the negative relationship between employment
and service use entirely resulted from an overall effect of higher client functioning
necessitating less treatment. Kitzman and others actively involved in developing
and delivering interventions have observed that employment in low-income
families often was associated with a different set of challenging issues, including

child care problems, transportation difficulties, unusual and inconsistent work

125

schedules, and challenges juggling competing demands of parenting and
employment demands on time (Kitzman, Cole et al., 1997). “Fitting in” the
appointments with a service provider, particularly one who may have limited
flexibility around the participant’s scheduled work hours, may become
increasingly difficult and less of a priority for some participants. Indeed, the data
in the current study supported that employed individuals successfully completed
fewer visits. While further work is needed to pinpoint why this occurred, these
findings suggested that the effective implementation of interventions was
influenced by the competing demands on participants of the program. Programs
may need to restructure policies (e.g., scheduled work hours for service
providers), supports (e.g., to insure safety of employees during evening home-
visits), format (e.g., continue home-based services or transfer to wrap-around
support in center-based setting that provides child care), and curricula content
and dosage for employed families. Likewise, communities may need to consider
how shifting employment policies influence the extent to which low-income
families can access and utilize the services designed to provide support to those
at risk.

Although postnatal employment reduced service intensity in both groups,
employment undermined service use to a greater degree in Nurse-CHW Team
care intervention than in SOC treatment. While both programs ostensibly offered
flexible appointments to clients, subtle differences in employee adoption of this
practice may have existed. On average, service providers were older in the SOC

intervention and few of them had young children of their own. Thus, without

I26

need to address their own childcare challenges, it may have been easier for SOC
nurses to schedule evening or weekend appointments than for nurses and CHWs
in the other intervention. The stronger path estimate in the Nurse-CHW Team
care intervention also could have resulted from two other possibilities. First, the
more restricted range of total visits in the SOC group may have weakened the
apparent differences in intensity of service delivery attributed to employment in
that group. Finally, the individuals who participated in postnatal employment may
have differed between the two intervention groups. Though significant in both
groups, a stronger relationship was observed between maternal education and
employment in the SOC group than in the Nurse-CHW Team care group. Thus,
the Nurse-CHW Team care group may have been more successful at engaging
women with more limited educational histories and fewer job skills in employment
during this period. Individuals with limited educational histories and fewer job
skills probably found the task of balancing work, family, and involvement in the
intervention more challenging than people who had successfully juggled time
management challenges in school settings previously. The observed difference
in the strength of the path from employment to service use between intervention
groups might then reflect this heightened engagement of vulnerable families to
the workforce by the Nurse-CHW Team care intervention.

Time points in Kent county. Another factor found to influence the
amount of services received in both programs was the length of time the
individual was in the county where the services were delivered. While the effect

of county residence on receipt of county-based services was not surprising, it is

127

important not to overlook its significance. Many social services and intervention
programs are provided at either the local or county level. Yet, many individuals
targeted to receive services frequently move between communities, counties and
states. No comprehensive network exists to connect individuals with services as
they make transitions between communities. Likewise, no system is in place to
transfer any information from one provider to the next about identified needs or
past services delivered to the family. In addition, from community to community
similar services have different names and are associated with different
institutions (e.g., health systems, school systems, departmental branches within
the government) making it difficult for people to seek out similar services when
they move to other communities. Furthermore, funding sources and eligibility
guidelines for many of these services vary across locales, rendering some
individuals eligible for services in one community and ineligible in others. In the
current study, those families who moved between counties received less
intervention. This limited receipt of intervention due to transience undermines
the program’s ability to have a positive impact on maternal and child outcomes
and attenuates the perceived effectiveness of interventions when research
results simply report findings on an intent-to-treat target group without
consideration of transience.

Service Helpfulness. Research has noted enhanced service utilization
and/or compliance with intervention regimens when people find that treatments
“matter to them” or are particularly relevant to their lives (Vivian 8 Wilcox, 2000).

Consistent with this work, the findings in the current study generally supported

128

the hypothesis that individuals who perceived the maternal and infant support
services to be more helpful to them would display greater use of intervention
services. In both groups there was a positive path estimate from service
helpfulness to intensity of service use. However, this path only reached
statistical significance in the Nurse-CHW Team care intervention. This result
also was consistent with expectations. Since the Nurse-CHW Team care
intervention was a relationship-based intervention (between mother and service
provider) with considerable flexibility to increase or decrease intensity of services
to meet individual case demands, the number of intervention visits was expected
to be more related to the individual’s perceived helpfulness of the program (and
of the service provider and the relationship underlying the program). In contrast,
the SOC intervention officially viewed the infant as the primary client (rather than
the mother) and was more educationally-oriented than relationship-based. The
wording of the helpfulness question asked individuals to rate how helpful the
maternal and infant support services have been “to you”. In some cases,
participants may have responded with perceptions about their personal benefit
from the program rather than the combined benefit to themselves and their
babies. Thus, the answer to perceived helpfulness might be more related to
service use for a relationship-based approach focused on the mother than for the
information-based service approach focused more on the infant. An alternative
explanation could also lie in the ranges of responses to the variables in each

group. Both groups exhibited similar variation in helpfulness ratings, however

129

there was considerably greater variability in the number of intervention visits
received in the Nurse-CHW Team care group than in the SOC intervention.

Perceived Stress and Service Utilization. As predicted, statistically
significant positive relationships were found between both expected needs and
postnatal perceived stress and perceived stress and depressive symptoms.
These relationships were consistent with a large literature establishing links
betweens difficult life events or circumstances and the perception of stress
(Barnfather 8 Ronis, 2000; Brooks-Gunn et al., 2000; Kessler, 1997) and
between different responses to those difficult events through perceptions of
stress and expression of depressive symptoms (Barnet et al., 1996; Dunkel-
Schetter et al., 2001; Ennis et al., 2000; Hall, 1990).

Contrary to study hypotheses, perceived stress did not exhibit a significant
direct effect on service use in either intervention group. Thus, it appeared that
perceived stress was more influential in altering individual’s tendencies to label or
report a need (see Chapter 1) than it was in decreasing or increasing individual
tendencies to use services to meet those needs. The lack of relationship
between perceived stress and service use is inconsistent with existing work
identifying reductions in service use intensity when individuals experienced
considerable stress (Eisenstadt 8 Powell, 1987; Kitzman, Cole et al., 1997). A
potential explanation for this inconsistency lies in the populations referenced in
the studies. Both Kitzman and her colleagues and Eisenstadt and Powell
discuss changes in service use among individuals engaged in programs.

However, the participants in the current study included both those who

I30

participated in the program at varying levels of intensity and those who never
were successfully engaged in the intervention (i.e., received zero visits from
providers). It is possible that dual processes occurred in the current study that
counter-balanced one another. For instance, among those using the program,
individuals with higher stress levels may have reduced their intensity of service
use. But, the group of individuals with no visits from providers overall may have
included a subgroup of individuals with relatively low levels of stress and high
levels of coping (i.e., generally functioning very well), who did not feel that they
needed services. In this example, the subgroup with no provider visits could
have masked the influence of high levels of perceived stress on service intensity
among those participating in the program. Future analyses comparing
participators to non-participators could investigate this the validity of this
explanation.

An alternative interpretation for the nonsignificant influence of perceived
stress on service use should also be explored. This study examined
relationships between stress and service use in the first 6 months after delivery
of an infant. The period immediately after the birth of a child brings many
changes to families (Tietjen 8 Bradley, 1985; Williams 8 Williams, 1997). Among
low-income mothers, there are many different factors that could contribute to
perceptions of stress. It is possible that some events evoking stress (e.g.,
uncertainty about baby’s behavior, difficulty helping baby adjust to day-night
sleep schedule, concerns about baby’s illness) during this period might lead to

increased service use, whereas other events evoking stress (e.g., finding and

131

adjusting to employment, problems in relationship with father of the baby,
financial problems) might reduce service use. The use of methodology that
measured only overall levels of perceived stress rather than stress associated
with different originating factors may have obscured relationships to service use.
Future research is needed that explores global ratings of perceived stress as well
as more specific indices on domains of stress (e.g., parenting-specific stress).
Perceived Stress andflrentiadlnteractions. The hypothesis that higher
levels of perceived stress would have a direct negative influence on maternal
interactions in a structured teaching task was not supported in either intervention
group. This result was surprising because a number of studies have found a
direct negative effect of perceived stress on maternal interactions with their
children (Crnic, Greenberg, 8 Slough, 1986; McKelvey et al., 2002; Planos et al.,
1997; Sachs et al., 1999). Other studies not necessarily finding a direct effect
between perceived stress and parenting interactions have found that perceived
stress exerts an indirect influence on interactions through its relationship with
depressive symptoms (Planos et al., 1997; Sachs et al., 1999), authoritarian
parenting style (Conger et al., 1984), ease of anger expression (Rodriguez 8
Green, 1997), or family disruption/changes in personal relationships (Taylor,
Roberts, 8 Jacobson, 1997). However, much of the existing work involves
parental interactions with either children across a wide range of ages (Sachs et
al., 1999), or older toddlers (Crnic et al., 1986) and children (Conger et al., 1984;
Planos et al., 1997; R. D. Taylor et al., 1997). Interactions with six-month-old

infants have some important differences from those with older children (Crnic 8

132

Acevedo, 1995) and mother-infant interactions have not demonstrated uniform
stability in style as their infants develop from six to thirteen months of age (Pettit
8 Bates, 1984). Thus, changes in infant behavior may be related to different
influences of maternal stress on observed interactions across ages. For
instance, young infants may not be as likely to interact in ways that could be
perceived as willful disobedience (e.g., walking away, throwing toys, talking back)
by a stressed parent and trigger negative interactions (Middlebrook 8 Forehand,
1985; Webster Stratton, 1990).

Existing research also tends to identify relationships between stress and
the socio—emotional components of parenting, such as maternal affect,
acceptance, sensitivity to cues, and participation in dyadic synchrony (Crnic et
al., 1986; McKelvey et al., 2002). However, the current study utilized all four
maternal subscales on the NCATS. This included a cognitive growth fostering
subscale that may have been more influenced by maternal education, verbal
fluency, and intelligence than by maternal stress and depressive symptoms.

A final explanation for the lack of direct effect of perceived stress on
maternal interactions involves the NCATS assessment itself. When
administering the NCATS, mothers are asked to select a task that they have not
yet observed their infants perform and to teach the child to perform the chosen
skill. Common activities selected in the six-month age range included the
following: transfer block from one hand to the other, squeak a squeaky toy, pick
up a food object and eat it, or scribble on a piece of paper. While not addressed

as a significant scoring variable by instrument developers (Sumner 8 Spietz,

133

1994), research observers noticed considerable differences in maternal behavior
depending on the task utilized with children at this age. Mothers teaching their
infants to scribble, for instance, displayed more negative reactions to the child’s
attempts to mouth the crayon than mothers teaching the child to use a squeaky
toy or to eat food. In addition, considerable differences in caregiver interactions
have been reported between mothers of different racial and ethnic groups
(Sumner 8 Spietz, 1994). However, to maintain sufficient power to test the
proposed hypotheses in each intervention group, the current study collapsed
individuals of various racial and ethnic backgrounds and teaching various tasks
into one group. Thus, it is possible that the existing variability in NCATS
maternal scores associated with the teaching task itself and the racial/ethnic
group of the mother masked the influence of perceived stress on the outcome.
@ressive Symtoms. In both intervention groups, increased depressive
symptoms were directly related to reduced maternal interactions. However,
contrary to study hypotheses, these effects did not reach statistical significance
in either group. A number of other studies have found significant negative
relationships between depressive symptoms and parent-child interactions (Field,
1998; Planos et al., 1997; Sachs et al., 1999). However, much of this work has
been conducted with alternative measures of parent-child interaction (Field,
Healy, Goldstein, Perry, 8 Bendell, 1988; Planos et al., 1997; Sachs et al., 1999).
The developers of the NCATS scale report mixed results in studies relating
depressive state to teaching scores (Sumner 8 Spietz, 1994). Research

providing simultaneous, independent ratings of maternal affect and teaching

134

interventions reported only correlations ranging from .36 to .41 between affect
and maternal teaching scores (Crnic, Ragozin, Greenberg, Robinson, 8 Basham,
1983). Other researchers have suggested that nonsignificant relationships
between maternal internal state and teaching interactions may have occurred
because “We might expect that all but the most depressed mothers could
marshall sufficient internal resources to ‘perform’ adequately in brief teaching and
feeding situations” (Booth at al., 1987, p. 304).

The NCATS instrument also combines multiple types of positive caregiver
behaviors into one composite maternal score. This approach may contribute to
inconclusive relationships between depressive symptoms and interactions with
young infants. Since six-month-old infants have not yet begun to speak words,
parents may emphasize modeling rather than instructing and verbalizing
feedback to children”. Parents who indicate high sensitivity to and
responsiveness to child cues, avoid punitive expressions and actions, offer
verbal and nonverbal instruction, feedback, praise, and description of task
materials will score higher in the NCATS interaction. Using another
methodology, Planos and her colleagues (1997) found that individuals with high
parenting stress displayed less inquiry and praise and more modeling and
mothers with depressive symptoms used more negative feedback, directives,
and modeling behavior with their children. If these findings were translated to the

current study with six-month-olds where fewer moms verbalized instructions with

 

26 Although I have not noticed mention of this in the literature, individuals conducting the
observations in the current study anecdotally noted considerable changes in parent behavior
(increased language, instruction, feedback and increased punishment, spanking, scolding, and
yelling with child age) between experiences scoring 6-month and 12-month interactions.

I35

children, then high modeling behavior and feedback would enhance the NCATS
maternal score. Thus, unless the depressive symptoms also influenced other
aspects of the interaction (especially punitive responses or lack of affection
toward the child), interaction differences between mothers with and without
depressive symptoms might be difficult to extract from the overall maternal score.
Evidence also suggests that difficult interactions might be even less readily
distinguishable in the current study sample due to the socioeconomic similarity of
the group. Past research has documented that working-class dyads exhibited
differences from middle class mothers that included many of the same
characteristics27 as reported in mothers with depressive symptoms (Phinney 8
Feshbach,1980)

Lastly, the effect of depressive symptoms on maternal interactions may
not have reached statistical significance, because its influence was partly
confounded with task and racial/ethnic variables not included in this analysis.
Future research that examines the influence of these potentially confounding
variables may further inform understanding of the influence of depressive
symptoms on maternal interactions in low-income women.

Mate__r_n_al and Child Proggm Ogtcomes.

The second set of questions investigated by testing the conceptual model
shown in Figure 6 related to the indirect influences of need mediated through

service utilization on parenting and child outcomes. Did the individuals who

 

27 Middle class mothers displayed significantly less intrusive behavior and used significantly fewer
negative statements and imperatives in interactions than did working-class mothers. Middle class
mothers also asked significantly more questions of their children. Two maternal styles emerged
within working-class moms - a negative intrusive style and a positive, non-directive style.

136

reported higher needs and subsequently used more maternal and infant support
services display better outcomes in parenting interactions, provision of nurturing
home environments, and have children with higher scores on mental
development assessments? The simple answer was no. Increasing service use
did not result in differential performance on these outcome measures in either
intervention group. However, a closer examination of the relationships modeled
raised some interesting questions.

Maternal lnteracjons. Earlier in this chapter’s discussion, I identified that
neither perceived stress nor depressive symptoms had statistically significant
effects on maternal interactions. While these relationships were of interest, a
more primary focus of the study was on understanding how service use
influenced maternal-child interactions. It was hypothesized that greater intensity
of service use would result in more positive maternal-infant interactions in
structured teaching tasks. Individuals who received a higher dosage of treatment
were expected to benefit more than those who participated at lower levels of
involvement. However, contrary to predictions, results suggested that increased
service use (either in actual number of visits or as a ordinal ranking of visit
quantity) did not significantly alter maternal interactions with their children for
individuals in either intervention group. Heightened service use did not
necessarily lead to more beneficial parenting outcomes. Upon closer scrutiny, a
number of factors may have contributed to this outcome.

Other researchers have discussed the difficulty of measuring engaged

participation in an intervention (Emde et al., 2000; Scott 8 Sechrest, 1989). In

137

the current study, problems associated with measurement of intervention dosage
may have masked a dosage treatment effect. Number of intervention visits (or
even rank order of amount of visits received in its ordinal form) may only tell part
of the intervention story. lmportantly, this measure of intervention dosage does
not provide information about specific aspects of the treatment. Some of these
might have included the following: the overall length of the intervention (e.g., a
35-minute visit and a two hour visit both counted as a single visit, receiving very
frequent visits for a short length of participation may have counted the same as
completing regular visits spaced evenly throughout the whole intervention
period), the content of the intervention (e.g., number of issue areas addressed,
types of activities undertaken, or whether content was crisis-oriented or issue-
oriented), the extent to which the client was fully engaged in and actively
participated in the visit, the extent to which others in the client’s social network
(e.g., father of the baby or her own mother) were included as part of the
intervention and perhaps continued to reiterate concepts to client at additional
intervals when the service provider was not present, the impact of additional
treatment dosage offered by other professionals that the service provider may
have contacted on the client’s behalf (e.g., referrals to social workers,
nutritionists, play groups, quality child care providers), or the contribution of
phone-based conversations with the client to answer questions or lend very brief
encouragement and support. Examination of the number of visits received
provides a rough approximation of the amount of intervention support received by

one client relative to others. In this case, it was the only comparable piece of

138

intervention dosage information available in both treatment groups and it was a
service utilization outcome that is commonly reported in many studies. However,
before drawing final conclusions about intervention effectiveness on maternal
interactions, more research is needed that explores the influence of these more
specific types intervention involvement differences.

Research also is needed that considers differences between those who
participated in the intervention and those who did not. In the current project, all
study participants were included, regardless of whether or not they had actually
participated in any of the intervention that they were randomly assigned to
receive. As a result, low service use intensity was confounded with a number of
individuals who did not participate at all. Individuals who either actively chose
not to participate or were simply missed by the services may have critical
differences from individuals who accepted the services and participated in
relatively few visits. Indeed, prior work has found differences both between
intervention participants and non-participants (Apodaca et al., 1997; lreys et al.,
2001) and between those who participated at different levels of intensity or
duration within the group involved in the intervention (Eisenstadt 8 Powell, 1987;
Navaie—Waliser et al., 2000; Olds 8 Korfmacher, 1998; Unger 8 Wandersman,
1988). The results of these studies have been mixed. Some of the prior
research has found non-participants to be more vulnerable and needy than
individuals who participate in intervention programs (Apodaca et al., 1997; Lindy,
Grace, 8 Green, 1981; Minde et al., 1980; Vachon, Rogers, Freedman, 8

Freedman, 1980), whereas other studies have indicated that non-participants

139

appeared to be coping well (Duggan et al., 2000; lreys et al., 2001; Vinokur,
Price, 8 Caplan, 1991 ). Another study further validated both of these claims by
finding that both a high functioning group and a highly vulnerable group may exist
within those who do not participate in services (Brown, 1978). Associations
between service utilization at varying levels of intensity and quality of maternal
interactions may have been clouded by the presence of a vulnerable group, a
high functioning group, or both of these subgroups within the cluster of
individuals who did not participate in the intervention.

The potential for a lack of natural fluctuation in service visits also may
have interfered with finding a relationship between service use intensity and
maternal interactions. A number of external constraints in each intervention
group may have reduced the possibilities for study participants to receive the
number of intervention visits they desired. In the standard of care intervention
nine nurse visits with clients were reimbursed during the pregnancy period and
the first 2 months after delivery (maternal support services) and again from 2-12
months postnatally. Nurses who wished to exceed this level of intervention in a
particular case had to acquire the approval of a supervisor and complete
additional paperwork indicating why the needs of the case merited further
intervention. The presence of this policy may have created an obstacle that
interfered with natural fluctuation in number of visits in this middle-intensity
range, and perhaps influenced the amount of intervention that a subset of

individuals received.

140

In the Nurse-CHW Team care group, a different external constraint may
have influenced service use intensity. Unlike SOC nurses, the Nurse-CHW Team
care service providers did not have to justify higher intervention dosages for
clients. However, in this intervention program, turnover issues28 influenced many
of the clients’ intervention histories. Out of 196 cases29 in this group, only 43
(21.9%) individuals maintained the same nurse and CHW handling their cases
throughout the whole intervention period (See Appendix R for more specific
information on turnover characteristics in the sample). The impact of changes in
service providers depended upon the specific circumstances of each staff
turnover event. In some cases, negative consequences could have been
minimized by maintaining one consistent provider with the client during the
transition. Yet, the possibility of an intervention lag or a gap in services occurred
with each turnover incident. Clients may have received less intense services or
experienced a short period without services while the case was reassigned,
caseloads were adjusted for remaining staff members, or new employees were

trained. Thus, the final number of visits clients received may have been

 

2" Staff turnover issues were unusually high during the study period for several reasons. A
number of service providers developed highly marketable skills and, during the low
unemployment economic times, accepted alternative job offers and promotions. Additional staff
were hired prior to the outset of the evaluation to enhance capability to serve the large number of
expected program participants. After a period of training, and in many cases some case
management, a portion of these individuals could not handle the demands of the job and were
terminated or chose to resign. However, the largest cause of turnover occurred midway through
the study when all staff in the program (and the health system) were given 2% pay reductions due
to an economic shortfall resulting from the merger of two hospitals. This pay cut significantly hurt
morale and many providers chose to leave the program to pursue other job opportunities.
Notably, these changes in pay did not occur in the SOC intervention and turnover was extremely
limited in that program. Other studies also have reported higher rates of turnovers in
interventions staffed with paraprofessionals as opposed to those who have been trained through
more traditional professional educational settings (Korfmacher, O’Brien, Hiatt, 8 Olds, 1999;
Wells, 1997).

29 Two cases were missing data on staff turnover and nineteen cases were excluded because no
postnatal visits from service providers were noted in their records.

141

influenced by program turnover influences as much as by individual
characteristics of the client and her situation. The program may have been less
able to provide the desired intensity of services or the client may have shifted in
her willingess to continue the same intervention intensity with the new provider”.
In addition to influencing the fluctuation of service use intensity variable
characteristics, turnover also may have had implications for progress in sensitive
intervention content areas. Especially in cases with difficult relationship histories,
even very smooth transitions between service providers had the potential to
undermine the stability of the provider-client relationship and to reduce the
potential effectiveness of a relationship-based intervention.

Home Environment. The quality of mothers’ interactions with their infants
was expected to directly affect the quality of the home environment provided to
nurture each infant’s growth and development. As predicted, in both intervention
groups, mothers displaying more positive interactions with their infants also
provided more nurturing home environments for their infants. This result was
consistent with prior research supporting positive associations between the
NCATS and HOME scales (Sumner 8 Spietz, 1994) and between both of these
scales and naturalistic observations of mother-child behavior in the home (Tesh
8 Holditch-Davis, 1997). Exploratory analyses suggested that individual
differences in maternal educational experience accounted for some of the
relationship between NCATS and HOME scores in both groups. Past work also

has reported relationships between maternal education and each of these

 

3° For instance, some individuals may have chosen to stop using services rather than to establish
a new relationship with a provider. Others may have decided to reduce the intensity of visits with
a new provider due to perceived differences in the client-provider relationship.

I42

instruments (Bradley, 1994; Bradley et al., 2001a; Sumner 8 Spietz, 1994). So,
while it was not surprising to observe a weaker relationship between NCATS and
HOME after taking education into account, the difference between intervention
groups in this regard was more unusual. After partialling out maternal education
differences, maternal-infant interactions did not exert a significant influence on

HOME scores in the Nurse-CHW Team care group, whereas a significant

 

I.
positive effect remained in the SOC group. There were not significant A
differences in mean education completed between individuals in either .
1
intervention group (See Table 4). However, more subtle variation in the i

distribution of educational experiences did exist (Nurse-CHW Team Care: 9.2%
< 8th grade, 46.6% 8-11th grade, 24.8% 12th grade, 19.4% > 12 years of
education; soc Care: 7.7% < 8‘“ grade, 49.8% 8-11th grade, 32.1% 12th grade,
10.4% > 12 years of education). If extremes on the distribution exerted a greater
influence in NCATS and HOME scores, then this difference might explain the
stronger influence of education on both NCATS and HOME scores in Nurse-
CHW Team care and the weaker relationship between NCATS and HOME
scores after partialling out this effect. Alternatively, variation in programmatic
emphases should be explored in treatment approaches with mothers who have
more limited educational backgrounds. Perhaps, SOC service providers actively
encouraged books, reading, and toys to mothers of young infants more
aggressively than Nurse-CHW Team care providers in these homes.

The hypothesis that higher quality home environments (as indexed by total

HOME scores) would have a direct positive effect on infant mental development

143

was not supported in the current study. No statistically significant effect of home
environment on BSID mental development index scores was observed in either
treatment group. The lack of observed relationship was surprising because prior
research has found that home environments containing high levels of responsive
and non-restrictive parent involvement, developmentally appropriate play
materials, organization, and variety were related to improved cognitive
development in children (Bradley, 1993, 1994; Bradley 8 Caldwell, 1988;
Burchinal, Campbell, Bryant, Wasik, 8 Ramey, 1997).

Several factors may explain the lack of support for a direct effect of home

environment on mental development in the current study. First, based on
eligibility requirements, all families in the current study had very limited financial
resources. Research on the HOME has found that poverty status explains
considerable variance in scores and nearly accounts for the observed variation in
scores between ethnic groups (Bradley et al., 2001a). Changes in family income
during the first four years of a child’s life also have been associated with
increased HOME scores in those families (Garrett, Ng'andu, 8 Ferron, 1994).

So, restricting the range of potential scores by selecting a sample that was
entirely low-income families may have reduced some of the instrument’s
discriminability. The correlation tables between HOME and MDI scores printed in
one study showed a reduction in size31 when values among lower socioeconomic
class families were compared to the larger group with varied socioeconomic

backgrounds (Bradley et al., 1989). In addition, unpublished study results from

 

3‘ Correlations between 12 month total HOME scores and 12 month MDI scores were .25 in the
overall sample, .03 in the lower class sample, .17 in the lower middle class sample, and .25 in the
middle class sample (Bradley et al., 1989).

144

Koniak—Griffen and her colleagues (data were later reported in Sumner et al.,
1994) found no statistically significant association between concurrent HOME
scores and MDI scores at 24 months in a very homogenous white, married,
middle-class sample of mothers participating in an intervention (Sumner 8
Spietz, 1994).

Secondly, the current study measured the characteristics of the home
environment at the same point in time as the child’s mental development
outcome. Although recent studies have documented reasonable stability""2 in
home environments across time, changes also do occur. So, if high HOME
scores at six months were not necessarily indicative of similar environments in
the preceding months, then these supportive home environments might not
contribute to noticeable changes in developmental outcome until months or even
years later. A time delay may be necessary in order for infants to benefit from
the learning materials, parental involvement, and caregiving practices found in
the home environment at six-months after delivery. Correlations from another
research project supported this possibility. Specifically, larger correlations were
reported between 12 month HOME scores and 24 month MDI scores (r = .50)
than between 12 month HOME scores and 12 month MDI scores (r = .25)
(Bradley et al., 1989). A number of other studies documenting the influence of
HOME scores on subsequent developmental outcomes also have measured

HOME scores at an earlier time point and developmental status at a later one

 

32 Burchinal and her colleagues (1997) found that the correlation between 6 and 12 month HOME
scores was .47.

145

(Bradley 8 Caldwell, 1981; Bradley et al., 2001b; Burchinal et al., 1997;
Thompson, Catlett, Oehler, Gustafson, 8 Goldstein, 1998).

In addition to the absence of a delay between measures, the actual child
age at assessment also may have influenced the nonsignificant HOME-MDI
finding. Interviewers in the current study noted that many families viewed their
children as “babies” at six month assessments and engaged in relatively little toy-
based play with children, whereas much greater emphasis on play with toys
occurred at twelve month assessments. Indeed, studies do report changes in
widespread availability of learning materials on the HOME by child age (Bradley
et al., 2001a). Researchers also have acknowledged a tendency to observe
higher HOME scores with increasing child age, a finding that may be indicative of
age-related differences in learning materials (Bradley et al., 1989; Caldwell 8
Bradley, 1984). Studies have demonstrated that variability in available learning
materials in the HOME subscale was heavily weighted in the predictive ability of
the HOME for subsequent intellectual development (Bradley 8 Caldwell, 1980;
Bradley et al., 2001b; Stevens 8 Bakeman, 1985). If parental perceptions of
learning materials for infants in the sample changed markedly from six to twelve
months, the relationships between HOME and MDI scores may also change
during this period. Work by Bradley and Caldwell (1980) supported this assertion.
After controlling other relevant factors in their sample, twelve month HOME
scores were a significant predictor of Stanford-Binet scores at age three,
whereas six-month HOME scores were not. This effect persisted even after

partialling out six month HOME scores from the twelve month analysis.

146

 

Finally, limited relationships may have been observed between HOME
Scores and MDI scores because of the characteristics of the BSID assessment
for 6-month—olds. At the six-month assessment the BSID contains items
designed to tap the developing cognitive and linguistic skills of the child.
However, many of the items on the mental subscale at younger ages include a
considerable motoric component that becomes somewhat minimized at older
ages. For example, scale items noted to be at the 6 month cognitive and
language developmental levels on the assessment include the following: Pulls
string adaptively to secure ring, cooperates in game, retains two of three cubes
for three seconds, rings bell purposely, lifts cup by handle, looks at pictures in
book, and uses gesture to make wants known. The development of some of
these skills may be less influenced by home Ieaming materials and interactions
than items at subsequent ages that involve fine manipulation of puzzle-like toys,
pointing to pictures, following verbal commands, expressing words, and
anticipating future actions in play event sequences. In a recent article Bradley
and his colleagues (2001b) supported this notion, acknowledging that child
performance on instruments measuring motor and self—care routines “are not as
likely as cognitive and language development to be affected by the kinds of
indicators included in the HOME-SF....” (p. 1879).

Infant Mental Development. Maternal interactions as evidenced in
performance during the NCATS teaching task were hypothesized to have a
direct, positive effect on infant mental development scores. However, the path

from maternal interactions to mental development index scores on the BSID-ll

147

 

was not statistically significant in either intervention group. This nonsignificant
finding was surprising because it is through joint interactions with primary
caregivers that infants acquire language, develop symbolic representations for
people and objects, and learn to identify and solve problems in their
environments (Fogel, 1990; Lang, 1984; Rogoff, 1998; Saami, Memme, 8
Campos, 1998). Furthermore, research has established positive relationships
between maternal interactions with children and subsequent child competence
on indicators of cognitive and language performance (Barnard 8 Morisset, 1995;
Bee et al., 1982).

Probable reasons for finding no effect of NCATS performance on MDI
scores were previously enumerated in the section discussing relationships of
HOME scores with MDI outcomes. These included: (a) no time lag existed
between measurement of NCATS scores and MDI outcomes, (b) greater
emphasis on motoric skills and less emphasis on language skills in the
assessment of mental competencies at 6 months than at subsequent ages, and
(c) potential restriction of predictive instrument discriminability through restriction
of sample to low-income population. Prior work that has established the
influence of maternal interactions on cognitive outcomes has routinely assessed
child outcomes after some passage of time following the initial interaction
assessment (Barnard 8 Morisset, 1995; Bee et al., 1982). The strong
relationships found between the maternal interactions in these studies and
subsequent child language outcomes also underscores the significance of

positive caregiver-child interactions in the mastery of symbolic and language

I48

 

skills that were less prominent in the six-month MDI assessment. Research also
suggests that child cognitive development is more negatively impacted by
contextual influences of poverty (including differential home environments and
proximal interactions) than motor development outcomes (Petterson 8 Albers,
2001). So, as assessments increasingly separate assessment of cognitive
mastery from motor skills, the influences of background characteristics for the
child would be expected to relate more strongly to the individual differences
observed. Indeed, NCATS instrument validation studies have identified only
minimal associations between parent-infant interaction prior to twelve months of
age and subsequent child cognitive ability (Sumner 8 Spietz, 1994).
Chilfierformance Factors. Several factors included in the conceptual
model were expected to reflect individual differences between infants that would
influence their interactions with their mothers and with the administrators of the
Bayley Scales of Infant Development. The inclusion of these variables was
intended to more accurately model individual differences in key outcome
variables by taking into account developmental histories of children and their
tendencies to exhibit unusually intense or ambiguous responses in interactions.
Test day behavior. The Behavior Rating Scale (BRS) of the BSID-ll
reflects the child’s orientation/engagement (to tasks, the examiner, and the
caregiver), emotional regulation, and quality of movement during the testing
situation. As predicted, in the current study BRS scores displayed significant

positive effects on infant MDI scores in both treatment groups. This finding is

149

consistent with a long history of previous research indicating that the BRS” and
the constructs that it taps are powerful positive predictors of both proximal infant
assessment scores and subsequent cognitive performance in childhood
(Burchinal et al., 1997; Matheny, 1980; Matheny et al., 1974; Roth et al., 1984).
The assessment situation exposes children to a variety of frustrating tasks and it
demands developmentally-appropriate periods of sustained attention and
engagement. Thus, it was not surprising that infants who were more alert,
socially responsive, persistent, and able to regulate their emotions following
periods of distress would receive higher overall scores. While it was not a
primary focus of this study, further research is needed to understand factors that
contribute to individual differences on the BRS since successful negotiation of
many of these behaviors will enhance the child’s subsequent social competence
and probable success in more structured school situations in the future.

Clarity of cues. Previous work has established individual differences in
the clarity of cues that infants exhibit in social interactions (Parke 8 Tinsley,
1983). Babies who display more ambiguous or confusing cues are less likely to
elicit appropriate responses from their caregivers, and therefore are less effective
in maintaining positive interaction sequences (Beckwith, 1984; Minde, 2000).
However, in this study the hypothesis that children sending clearer cues would
elicit more positive interactions from their parents was only supported in the SOC

group. For SOC intervention families, babies with clearer cues engaged mothers

 

33 Some research discusses the lBR rather than the BRS. The BRS was named the Infant
Behavior Record (lBR) in the previous edition of the Bayley Scales of Infant Development. The
scale was renamed the Behavior Rating Scale after revisions and further validation associated
with the restandardization of the BSID-ll (Bayley, 1993).

ISO

 

in significantly more positive interactions. However, cues were unrelated to
maternal performance in the Nurse-CHW Team Care mothers. Careful
examination of the data found no difference in mean quality of interactions
between intervention groups (Nurse-CHW Team Care M = 31.54, SD = 4.41;
SOC M = 31.93, SD = 4.87, t(1, 436) = -.864, p = .388) despite the fact that
infants in the Nurse-CHW Team Care group displayed significantly more
ambiguous cues than infants in Nurse-CHW Team care group (Nurse-CHW
Team Care M = 7.30, SD = 1.19; SOC M = 7.65, SD = 1.25, t(1, 436) = -2.94, p

= .003). So, mothers in Nurse-CHW Team care more effectively compensated

 

for their children’s lack of responsiveness than did mothers in SOC group. ’
Differences in the interventions may have contributed to this effect. Nurse-CHW

Team care providers may have helped mothers identify and read ambiguous

cues more effectively than SOC treatment. However, the mean difference in

infant cues observed was not huge. Typical infants in both groups exhibited at

least seven out of ten cues. Future work should explore whether participation in

the intervention actually contributed to the nonsignificant impact of cues on

maternal behavior or if similar compensatory maternal behavior was observed

among those who did not participate in the intervention as well.

Poor infant health. The hypothesis that poor infant health would
contribute to decreased MDI scores and more ambiguous cues was not
supported in this study. In both treatment groups, no significant effects of infant
health were observed. Although the health index used in this study was

composed of factors noted in the literature to be related to poorer developmental

15]

outcomes in children, it had not been designed or tested in prior research. Thus,
the instrument may simply not be effective at discriminating high and low risk
infants. For instance, the use of a cumulative approach that encompassed birth,
neonatal, and postnatal health factors may have equally weighted risks that
should have been weighted differently.

Study sample selection also may have contributed to the nonsignificant
relationships between infant health and behavior. For selection in the
subsample, individuals had to complete the late pregnancy interview. This
interview was typically collected at 34—38 weeks gestation. Data were not
collected if individuals delivered prior to completion of the late pregnancy
interview. This selection criterion had the effect of excluding the highest risk
preterm infants from the sample. With the exception of one three pound baby
born at 32 weeks gestation“, all of the babies in this subsample were at least 34
weeks of age at delivery and weighed at least four pounds at birth. While some
babies were still born prematurely and met the low birth weight criterion, this
group of infants typically has better developed lungs and has a considerably
better prognosis overall than infants born at 24-32 weeks gestation and weighing
under 1500 grams (Paneth, 1995; Widerstrom 8 Nickel, 1997). The exclusion of
infants at highest risk based on birth outcomes and the use of corrected rather

than actual age at assessment (both NCATS and BSID assessment) probably

 

3‘ In this case, the interview was conducted early based upon a wrong due date listed in the
research chart. Although the due date had been altered based on an early ultrasound, the chart
listed the date given based on last menstrual period. The client delivered this baby within a
couple days of the late pregnancy interview. The baby weighed 1375 grams and gestational age
was 224 days.

152

primarily accounted for the lack of explanatory power of the child health variable
in this study.

M-

A focal interest in this study was to investigate not only what factors
influenced initial reports of need, but also how expected and reported needs
related to participation in prevention/intervention programs and to individual
maternal and child outcomes. The findings suggested that understanding
reported needs is critical. Actual use of service was driven not by individuals’

expected needs, nor by their stressor responses to those needs, but by their self-

. ’ man-u
”5'3”. TH

reported needs. Thus, this work corroborated discussion in chapter one
indicating that reported needs may be the first step, even a beginning
commitment or personal goal, toward the individual change process.

However, embedded within the data presented was the clear finding that
service use was not simply a function of reported needs. Instead, reported need
was a key element within a bigger system of influences on use of services.
Program involvement also depended on access to the services, competing
demands on the individuals' time and resources, and their perceptions of the
overall helpfulness of the program itself. None of these insights are earthshaking
revelations. But they are critical considerations for reaching target populations
with services designed to support them. Services were used with greater
intensity when they were personally relevant, related to a reported need,
accessible, and approachable within the schedules and competing interests of

individuals.

153

Finally, the results reiterated that program use does not always translate
into improved outcomes. Greater dosage did not necessarily promote more
change. Understanding the factors influencing program engagement will not
provide simple solutions to the complicated issue of how best to enhance
outcomes among eligible families. Creating a network of services that reaches
the families that may benefit and benefits the families it reaches will be an

ongoing challenge.

154

 

CHAPTER 3

REPORTED NEEDS AMONG LOW-INCOME WOMEN:
DISCUSSION AND IMPLICATIONS

Study Aims.

This dissertation study investigated individual differences in reporting

perceived needs within a population of low-income pregnant women. Initially, I a

‘ N I A“?!

conceptually distinguished between three types of needs: expected, perceived,

and reported. Through several sets of analyses, I then considered if there was 1

 

an empirical, as well as conceptual, basis for distinguishing between expected I'
and reported needs. To this end, I particularly examined observed differences
between the extent of concern identified by participants (reported needs) and that
which might have been expected by providers based on more objective
characteristics of each participant’s self—described situation at the outset of
intervention (expected needs). Specifically, how were expected and reported
needs related to service use and intervention program outcomes? What
influenced individuals’ tendencies to report the presence of a need? What are
the implications of these findings for research and practice?

Is the Distinction Between Expecfiggnd Reported Needs MM?

The findings suggested that reported needs do matter. In this sample,
reported needs were based upon substantive expected needs, but they were not
interchangeable with them. Unlike expected needs, reported needs predicted

subsequent service utilization behavior. Simply disclosing a need--labeling that a

155

concern was present in personal experience—increased people’s tendencies to
use community services designed to assist with those needs. Thus, it appeared
that reporting the need was a step toward action or behavior change, even
perhaps an indicator of some commitment to engaged program participation.

So the distinction between expected and reported needs mattered. Why
is that significant? Numerous studies and programs begin with the premise of
producing change or attenuating a likely negative impact on groups “at risk” or “in
need”. Yet, little consideration has been given to the initial definitions of these
groups based on expected needs versus self-reported needs. The current
investigation suggested that the presence or absence of reported needs had
important implications for active participation in the treatment being tested.
Although it was not directly investigated, levels and types of reported needs also
probably produced substantive influences on the character and course of the
treatment itself. Thus, even when offered the same treatment, samples based on
reported versus expected needs would not be directly comparable with respect to
the intervention received. In the current study, need reporting seemed to reflect
a higher level of individual “readiness” for services. Differences between
randomized trials based on expected needs and quasi-experimental pre-
test/post-test comparisons often involve samples with divergent levels of reported
needs at the outset of the study. Yet, typically reported needs are not measured
and included as a factor in the analyses and interpretation of findings. Attention

to reported needs in future research might result in more precise descriptions of

156

whom an intervention “works” for35 rather than reiterating that a program is not
effective for the whole population with expected needs or that it displays very
weak effects across the general population with expected needs. Alternative
intervention approaches may be designed to enhance readiness to use services
among those who do not report the presence of expected needs.

While the findings from the current study supported the assertion that
development of different interventions for people with diverse need reporting
characteristics (or perhaps more specific readiness criteria to be developed
later), it is not the first to invite such a call to action. Indeed, client-centered care
approaches employed by many programs already recognize the usefulness of
individualizing interventions based on reported needs and choices of program
participants (Allen 8 Petr, 1998; Nelson 8 Allen, 1995). However, an important
next step is to further tailor programs and services to address the different set of
intervention issues among people reporting few needs, including both non-
participants and underreporting36 participants. In past work, Prochaska and his

colleagues have advocated identifying individual stages of change and adapting

 

35 The tone in current home visiting evaluation literature often underscores that interventions do
not appear to be effective across a broad population with expected needs. Home visiting is not
“THE” universal answer. However, studies may have defined the population too broadly, thereby
undermining evidence of effective outcomes. For instance, neither Alcoholics Anonymous or
Weight Watchers programs probably would appear to be effective across the broad populations
with treatment of their respective addictive behaviors. But, when individuals achieve a level of
need recognition and readiness, participation in these programs appears to produce more
positive results than individual efforts alone. Likewise, specific home-visiting programs may be
effective interventions among people who have a certain level of readiness and receptiveness to
information and intervention. Measurement of reported needs and other factors indicating a
readiness to change may be the first step to investigating the validity of this possibility.

36Although this terminology of underreporting implies that individuals are inaccurate in their
representation of their needs, this IS not the connotation that Is intended. The perceptions of
individuals, regardless of their match with others” perceptions, are real and are influential in
guiding behavior. In this context the use of the “underreport” term provides a reference not to the
individuals “real" needs, but rather to the needs that policy makers, program planners, and
researchers might attribute to them based on expected need characteristics.

157

specific addictive behavior treatment strategies appropriate for clients in each
stage (Prochaska, 1996; Prochaska et al., 1992). However, to my knowledge,
development of a similar protocol has not been discussed within the preventive
intervention domain. Application of these ideas into the design, marketing, and
implementation of interventions may improve the abilities of agencies to target
appropriate service delivery to community members and to maximize potential
changes in outcomes over time. Likewise, evaluations that take into account
client readiness in investigations of program effectiveness may define
intervention “success” differently depending on the initial characteristics of the
clients. Modification of research designs and outcome indicator selection to
consider differences in client readiness may serve as an important first step to
understanding which interventions are most effective with which clients.

What Influences RewrtedrNegis?

Although reported needs were based on substantive expected needs,
expected needs influenced the level of reported needs through both direct and
indirect pathways. Stressor responses were found to mediate relationships
between expected needs and reported needs. Future research will need to
identify whether these stressor responses heightened the perception of individual
threat associated with inaction or acted as a cue to expedite action among
individuals who already had perceived a threat. Yet, the data suggested that the
presence of stressor responses increased need-reporting behaviors. This finding
might be useful in developing intervention efforts to enhance individual readiness

to use services. For instance, a program could help pregnant women or parents

158

with expected needs vicariously experience the stress37 associated with not
developing a coherent parenting strategy or the uncertainty associated with not
knowing strategies to handle difficult child behaviors. Exposure to these kinds of
experiences within a safe context might increase individual tendencies to report
needs for parenting support and subsequently either seek or accept information
and assistance from others.

Individuals who viewed themselves as having general competence to take
efficacious action at some point related to the issue and who felt that benefits of
action would be greatest due to low competence in specific issue area
characteristics also expressed more needs. Since high expected needs were
negatively related to feelings of personal control, an important component of
some interventions may involve empowering people sufficiently so that they
begin to identify their own needs. In some cases, the presence of more reported
needs at the end than at the beginning of an intervention may represent
movement toward effective use of preventive intervention services that can
promote change. Thus, programs and evaluations may not necessarily expect to
see a reduction in reported needs, or a greater appearance of need fulfillment,
following treatment. Attention to the multiple meanings of evaluation data
collected for clients with different levels of readiness may provide greater clarity

in understanding the impact of home-based intervention.

 

37 Examples might include highly realistic vignettes in television commercials (cereal, shoes, and
alcohol establish a “need” for products through advertising, why not use these outlets to
encourage people to learn about parenting?) Alternatively, interventions could be developed for
patients in obstetric or pediatric office waiting rooms. For example, people could be exposed to
tapes with periods of extensive infant crying, virtual reality games handling common toddler
discipline issues, or other simulated challenges that ultimately highlight the usefulness of having
additional parenting knowledge and support.

159

Taken together, the current study results suggested that people who saw
greater relevance of the issue in their lives (e.g., because it was contributing to
heightened stress responses), who believed that they possessed the capability to
eventually take action related to the issue, and who thought that their actions
would provide some tangible benefit beyond their existing levels of competence
in that specific issue area, were the people who tended to report higher needs.
Thus, this pattern combining general mastery resources with heightened
relevance and potential benefit played a significant role in the active reporting of
expected needs that existed. This pattern of characteristics and apparent
appraisals helped people label a personal need that existed. Furthermore, this
labeling process was later shown to be more relevant than the presence of the
simple expected need in determining the intensity of later service utilization.
Thus, this combination of factors among low-income women contributed to a
“readiness” for service utilization at an individual level.

Whfiat Contritytes to MSSIISS Sivice Utiflzation?

lmportantly, actual service use intensity was not a direct function of
expected needs in this sample of low-income women. Instead, a combination of
individual, environmental, and programmatic characteristics contributed to the
intensity of service utilization. Influential factors that were supported in the study
included the following: individual readiness to use services (as indicated by
reported needs), broad access to services (i.e., residence in the county),
minimized competing responsibilities (i.e., employment), and the program itself.

Program characteristics influenced not only how many visits were necessary to

I60

indicate greater intensity of service use, but also how much client evaluation of
the program as personally helpful was important in determining the subsequent

amount of use. Although factors at a number of these levels were listed among

 

those described in McCurdy and Daro’s (2001) recent theoretical model of
service utilization, the inclusion of employment and residence in the county are

notable additions to their model. These findings draw attention to the

 

h
significance of factors at a number of levels in contributing to effective service
use. Attention of communities toward the range of influential factors in service
use may help establish more effective support to families or more flexible '

adaptation of services for families with different needs (e.g., unemployed and
employed) in order to maximally facilitate service use whenever individuals and
families wish to participate.

How Does Service Utili_zation Relate to Outcomes?

The results of this investigation also underscore the reality that, although
maximizing service use is a critical programmatic concern, more service does not
always result in better outcomes. Use of services is necessary to benefit from
intervention, but it is not sufficient. Programs and researchers must carefully
document which aspects of interventions were implemented and utilize more
effective coding schemes for the type of intervention conducted with each family.
Comparisons of actual needs addressed versus needs repOIted will provide
further information about how effectively family-centered care models are
implemented. In addition, repeated assessments of client and family

engagement in the intervention, perceptions about the relevance and usefulness

I61

of the intervention, activities and issues addressed during the intervention, and
relationships with service providers will offer maximal information to improve
program implementation as well as contribute to a better understanding of links
between the intervention provided and subsequent behavioral indicators of family
and environmental improvement as a result of participation.

Whv Focus on Needs in a Strengths-Based Setting?

Currently, in both service delivery and research realms there is a shift
away from examination of “needs” in favor of focusing on the less deficit-oriented
notions of assets or strengths (Nelson 8 Allen, '1 995; Reed 8 Brown, 2001).
Although this dissertation uses terminology involving individual needs, the
findings herein raise questions that are highly relevant to those espousing a
strengths-based approach to their work. The results presented point to critical
interrelationships between the constructs of needs and strengths. Individuals
with greater self-esteem and mastery (indicative of people highly aware of their
strengths) were more likely to subsequently report the presence of needs. That
is, needs were not identified as the absence of a given competence or strength.
Reported needs did not appear simply to be the opposite pole on the “strengths”
continuum. Indeed, this study on reported needs may have identified part of the
process explaining why asset-oriented approaches work so well with some
individuals and families. The focus on strengths during intervention may
enhance individuals’ abilities to identify, report, and ultimately take action to
overcome substantive needs. Future work is needed to examine how existing

constructs of need and strengths are theoretically, empirically, and experientially

162

related to each other. It is likely that considerable improvement in
prevention/intervention effectiveness in promoting behavioral changes will lie at
the intersection of these components.

Mimitations arg Recommendations for F dtgre Resm.

Although this investigation has provided some important contributions to
the understudied literature on expected and reported needs, a number of
limitations must be acknowledged. First, although the distinction between
expected, perceived, and reported needs has been identified, unfortunately the
study methodology confounded perceived and reported needs. While this is
likely to be an ongoing problem in many studies, future work must search for
ways to measure these constructs effectively. These efforts are needed in order
to differentiate whether factors that contribute to reported need are tied to the
need appraisal process or to the determination about whether or not to disclose a
need that is perceived to the audience. To understand that process as well,
future explorations should include more careful measurement of disclosure
constructs, including both social desirability and lie scale instruments.

Although the data set used was quite rich with information about low-
income pregnant and parenting women, future studies may benefit from the
collection of some additional information. First, the introduction of greater
socioeconomic variability into research would help explore whether factors and
processes described are specific to women of low-socioeconomic classes or are

more broadly applicable.

I63

 

 

Second, inclusion of a larger sample for future work is essential. In this
case, the sample size lacked sufficient power to examine models stacked by
racial/ethnic group or neighborhood characteristics. Yet, the attitudes and
experiences prominent within one’s proximal community and within one’s racial
or ethnic group are likely to have a considerable impact on attitudes toward
service use in general and in perceptions of specific programs more particularly
(McCurdy 8 Daro, 2001).

The inclusion of more specific measures to this research question also
would contribute greatly to potential understanding of this phenomenon. A
critical analysis of how reported need patterns vary depending on the structure of
need related questions, the contexts in which they are asked, and the variations
observed across need topic areas is needed. Also, assessing each individual’s
progress along various dimensions of Prochaska’s stages of change would help
identify if need reporting emerges and is maintained or if need-reporting
tendencies fluctuate with personal efforts to change behavior. Other measures
too, including coping styles, measures of cognitive and verbal abilities, perceived
threat, and reported competence, stress, and social support related to specific
need constructs, would maximize the potential explanatory power of future work.
Finally, distinctions between reported need, intention to enroll, actual enrollment,
and ongoing retention/participation in services would help elaborate the links

between model components.

164

Closing Comments.

With relative ease, many scientists have categorized individuals by levels
of needs based on relatively objective criteria. However, closer examination
suggests that experiential definitions of need are much more complex. As
people identify whether or not current situations constitute a personal need, they
appear to be making significant judgments about their capabilities and readiness
to act as well. While existing work examining the influences on and the potential
implications of various reported need behaviors remains in its infancy, it appears
that continued investigation in this area holds promise for expanding
understanding across a wide range of areas. Components in models of coping,
empowerment, service utilization, risk reduction and preventive intervention
efficacy may be revisited and/or developed as further research unfolds. But, the
greatest potential benefit from investigating differences between the scientific
and "lived” dimensions of need will be our eventual ability to more successfully
apply scientific knowledge to address the all too real needs of families in our

midst.

I65

TABLES

166

Table 1. Comparison of treatment characteristics across study intervention

 

 

groups.
Characteristic Nurse-CHW Standard of Care
Team Care
Service provider Nurse and Community Nurse

responsible for case
management

Relationships to
nutrition and social
service providers

Relationships to
health care
providers

Format of service
delivery

Primary client —
prenatal

Primary client —
postnatal

Health Worker team

Encouraged to refer to
nutritionists and social
workers

Regular communication
about client needs and
ongoing treatment with
health care providers

Home-visiting, visits in
other locations at client’s
request

Mother, but also
encourage inclusion of
informal support network
members

Mother and infant, but
also encourage inclusion
of informal support
network members

Encouraged to refer to
nutritionists and social
workers

Refers to and receives
referrals from health
care providers. Not
regular communication
about client needs.

Primarily home-
visiting, however one
subgroup of
contracted prenatal
providers relies on
face-to-face contact at
health clinic instead of
home visits in many
cases

Mother

Infant

 

167

 

 

Table 1. (Cont) Comparison of treatment characteristics across study

intervention groups.

 

Characteristic

Nurse-CHW
Team Care

Standard of Care

 

Intervention treatment

Assessment followed by
individualization of
curriculum modules to
client’s situation, her
priorities, and her
intervention format
preferences. Repeated
re—assessment of client’s
needs and priorities
throughout the
intervention. Curriculum
modules addressed the
following issues‘:
maternal mental health,
life skills, pregnancy and
childbirth issues, risky
behaviors, service use
social relationships,
infant care and
development, and
parenting.

Assessment followed
by developing a
multidisciplinary
treatment plan that
recognizes client’s
situation and her
priorities. Nurse
provided information
and resources
identified in the care
plan. Common topics
included: nutrition,
parenting, risky
behaviors, health
conditions and
practices, pregnancy,
infant development,
and infant care.
Information provision
not based on a
specific documented
curriculum.

 

 

1 Examples of curriculum content included the following: maternal life skills (e.g., budgeting and
finance, cooking and nutrition, personal goal setting), maternal mental health (e.g., postpartum
blues and depression, anxiety, stress), pregnancy issues (e.g., fetal growth and development,
emotional changes, rest, diet, and exercise), childbirth issues (e.g., preterm labor, when to go to
the hospital, birthing plan), health behaviors and n'sky behaviors (e.g., pregnancy do’s and don'ts,
tobacco, drugs, sexually transmitted diseases), service use (e.g., prenatal care, telephone tips,
WlC, when to call your doctor, baby's health care, community resources), social relationships
(e.g., conflict resolution, relationships and support, communication through nurturing and touch,
getting to know your baby), and infant care and development (e.g., bathing your baby, care of
belly button, feeding your baby, playing to Ieam, baby's growth and develop-ment) and parenting
(e.g., things you’ll need for baby, SIDS, crying and colic, helping your child Ieam, parenting young
children, accident prevention, health and hygiene, smoke free baby and me).

168

 

 

Table 1. (Cont) Comparison of treatment characteristics across study

intervention groups.

 

Characteristic

Nurse-CHW
Team Care

Standard of Care

 

Referral mechanism

Frequency of visits
according to clinical
path — prenatal

Frequency of visits
according to clinical
path - postnatal

Professional provider
referral not necessary.
Program directly initiated
contact with women
served by the health
clinics.

CHW: First month:
weekly home visits,
Thereafter. 2 home visits
and 2 phone calls per
month

Nurse: 1 home visit and
one phone call per
trimester with additional
predelivery home visit.

Delivenr- 6 Months
CHW: 2 home visits and
2 phone calls per month
Nurse: postdelivery home
visit plus 2 additional
home visits and 3 phone
calls during the six month
pefiod

6 Months — 1 2f Months
CHW: 1 home visit and 1
phone call per month
Nurse: 1 home visit and 1
phone call during the six
month period, plus
additional home visit for
transition planning
purposes and case
closure

Primarily professional
provider referral
through the health
clinic providers 1 "

Nurse: Up to 9 home ._
visits from pregnancy ‘
until 2 months after '
delivery. These are

typically delivered as

one visit per month

 

2 Months - 12 Months
Nurse: Up to 9 home
visits during this time
period. These are
typically delivered as
one visit per month.

 

169

 

Table 1. (Cont) Comparison of treatment characteristics across study
intervention groups.

 

 

 

 

Characteristic Nurse-CHW Standard of Care
Team Care
Appointment Appointments were Appointments were
persistence scheduled around client’s scheduled around
needs. Considerable client’s needs.
effort (phone calls, in- Follow-up phone 9.3
person attempts) calls (if had a
expended with repeated phone) to clients .j
attempts in cases of who did not keep
clients who did not keep appointments. But 3
appointments. it was primarily the i
client’s
responsibility to '
reschedule
appointment, if
desired.
Process for exceeding Nurse-CHW Team Special approval
recommended visits discuss reasons and required for agency
beyond those outlined adjust treatment plan as reimbursement.
on the clinical path needed Nurse provides
written justification
for necessity of
visit(s).
170

 

 

Table 2. Characteristics of Medicaid-eligible mothers in the Michigan
Maternal Health Services Study sample (n=613) and in dissertation
subsample (n=438) at study enrollment.

 

 

Excluded Subset Total Compare
Characteristic Cases Selected Sample (Excluded
(n=175) (n=438) (n=613) vs.
n (%) n (%) n (%) Selected)
Treatment group X2 = .178
MOMS Intervention group 90 (51.4) 217 (49.5) 307 (50.1) df = 1
Standard of Care group 85 (48.6) 221 (50.5) 306 (49.9)
Age M = 22.70 M = 22.27 M = 22.39 t= -1.02
SD = 4.75 SD = 4.73 SD = 4.74 df = 611
16-19 43 (24.6) 135 (30.8) 178 (29.0) X2 = 3.39
20-24 85 (48.6) 200 (45.7) 285 (46.5) df = 3
25-29 32 (18.3) 62 (14.2) 94 (15.3)
30-39 15 (8.6) 41 (9.4) 56 (9.1)
Race/Ethnicity X2 = 2.27
Caucasian 73 (42.0) 182 (41.6) 255 (41.6) df = 4
African American 54 (31.0) 116 (26.5) 170 (27.7)
Hispanic 32 (18.4) 101 (23.1) 133 (21.7)
Native American 4 (2.3) 9 (2.1) 13 (2.1)
Other 11 (6.3) 30 (6.8) 42 (6.9)
Marital Status
Married 21 (12.0) 84 (19.2) 105 (17.1) x2: 11.15"
Divorced, Separated, or 27 (15.4) 34 (7.8) 61 (10.0) df = 2
Vlfidowed
Single, Never Married 127 (72.6) 320 (73.1) 447 (72.9)
Living with father of baby 81 (46.3) 208 (47.5) 289 (47.2) X2 = .086
df =
Education
Years completed M = 11.02 M = 10.70 M = 10.79 t= -1.71‘
SD = 1.93 SD = 2.37 SD = 2.26 df = 611
< 12 years 100 (57.1) 248 (56.6) 348 (56.8) X2: .014
df = 1
Employed 77 (44.0) 190 (43.4) 267 (43.6) x2 = .020
df = 1

 

‘ .05 < p < .10 ** pg .01 pg .001

171

 

Table 2. (Cont) Characteristics of Medicaid-eligible mothers in the
Maternal Health Services Study (n=613) and in dissertation subsample

(n=438) at study enrollment.

 

 

Excluded Subset Total Compare
Characteristic Cases Selected Sample (Excluded
(n=175) (n=438) (n=613) vs.
n (%) n (%) n (%) Selected)
Weeks gestation M = 10.89 M = 12.24 M = 11.85 t= 3.24***
SD = 4.48 SD = 5.04 SD = 4.92 df = 611
Parity X2 = .505
None 72 (41.1) 194 (44.3) 266 (43.4) df = 1
One or more 103 (58.9) 244 (55.7) 347 (56.6)
Abuse History
Ever physically abused 100 (57.5) 223 (50.9) 323 (52.8) X2 = 2.15
df = 1
Ever emotionally abused 108 (62.1) 241 (55.1) 349 (57.1) X2 = 2.43
df = 1
Ever sexually abused 42 (24.0) 92 (21.1) 134 (22.0) X2 = .592
df = 1
Psychological Characteristics
Perceived Support M = 47.93 M = 47.32 M = 47.50 t= -.948
(MSPSS) SD = 6.89 SD = 7.17 SD = 7.09 df = 611
Perceived Stress M = 26.04 M = 26.42 M = 26.31 t= .541
(Cohen) SD = 7.67 SD = 7.86 SD = 7.80 df = 611
Self-Esteem M = 31.06 M = 30.75 M = 30.84 = -.861
(Rosenberg) SD = 4.18 SD = 3.96 SD = 4.02 df = 611
Mastery (Pearlin) M = 20.74 M = 20.63 M = 20.66 t= -.453
SD = 2.67 SD = 2.85 SD = 2.80 df = 611
Depressive Symptoms M = 19.42 M = 19.55 M = 19.51 t= .137
(CES-D) SD = 11.34 SD = SD = df = 611
10.93 11.04
CESD 316 102 (58.3) 248 (56.6) 350 (57.1) X2 = .141
df = 1

 

f.05 < p < .10 ** p: .01 p_<_ .001

172

 

Table 3. Author-generated expected and reported need indices.

Note: For expected need, individuals were only given one score per category. So, if multiple
characteristics fit the individual’s particular situation, then the one with the highest assigned need
rating was given to them.

 

 

 

 

Expected Need Reported Need
Lack of Information About Pregnancy/Childbirth
Have you ever been pregnant before? Since enrollment, did you have
Yes 0 No 2 a need/concem for pregnancy
educafion?
Yes 2 No 0
Have you ever delivered a live birth? Since enrollment, did you have
Yes 0 No 2 a need/concern for childbirth
educafion?
Yes 2 No 0

 

Lack of Information on ParentingICaring for Children
Is participant currently caring for children Since enrollment, did you have

 

 

(biological or othenlvise): a need or concern for infant
Yes 0 No 2 care education?
Yes 2 No 0
Caretaking ability confidencezz Since enrollment, did you have
(range 0-2) a need or concern related to
o I worry that I will not know what to do parenting?
if my baby gets sick Yes 2 No 0

. It will be difficult for me to know what
my baby wants.

0 I often worry that I will be forgetful and
cause something bad to happen to
my baby.

a I am afraid I will be awkward and
clumsy when handling my baby.

‘/2 point for each of following questions if answer
uncertain, mainly true, or completely true on it.

 

 

 

 

 

2 Items taken from caretaking ability subscale of the Maternal Self-Report Inventory. Reliability
and validity information about this scale are noted under the mothering ability esteem measures
section.

173

 

Table 3. (Cont) Author-generated expected and reported need indices.

 

Expected Need

Reported Need

 

 

12 years of education or greater
regardless of current school status 0

 

Educafion

<12 years of education 8 not Since enrollment, did you have a
currently in School 2 need or concern regarding your
<12 years of education 8 in own education?

school currently 1 Yes 2 No 0

 

Health Insurance

 

At enrollment:
Enrolled in neither medicaid

 

Since enrollment, did you have a
need or concern related to

 

nor private insurance 2 getting/using Medicaid or
Medicaid or private insurance MichCare?

is pending official status 1 Yes 2 No 0

Enrolled in either medicaid or

private insurance 0

Emplgyment

 

Expected need is based on whether or not
employed at both enrollment and 34 weeks
and if at enrollment mom expressed desire
to have job (suggests needed/wanted one)
by saying she was looking for work.

Not employed both waves 8

Looking for a job at enrollment 2
Became unemployed between
waves 2

Not employed both waves 8

Not looking for job at enrollment 1
Employed at both waves 8

Looking for new job at enroll 1
Employed at both waves 8 Not

looking for new job at enrollment 0

 

 

Since enrollment, did you have a
need or concern about finding a
job or work?

Yes 2 No 0

 

174

 

 

 

 

Table 3. (Cont) Author-generated expected and reported need indices.

 

 

 

 

 

 

 

 

 

 

 

Expected Need Reported Need
Housing
Regardless of number of reported Since enrollment, did you have a
moves between data waves, if living need or concern related to
in motel, shelter or if homeless at housing?
enrollment 2 Yes 2 No 0
2 or more moves between waves
8 live in home at enrollment 2
1 move between waves 8 live
in a home at enrollment 1
No moves between waves 8 live
In a home at enrollment 0
Food
Response at enrollment to question: Since enrollment, did you have a
need or concern about getting
How often do you and your family have food and/or menu planning?
the amount of food you need? Yes 2 No 0
Sometimes, rarely, or never 2
Most of the time 1
Always 0
Personal Safety
If reported physical, sexual, or Since enrollment, did you have a
emotional abuse at any time during need or concern related to

pregnancy or since enrollment, then domestic violence?
client received 2 points regardless of Yes 2 No 0

reported feelings about safety living in
home 2

If no abuse reported since enrollment,
expected need based on answer at
enrollment to the question:

How often do you feel safe living in

 

 

 

your home?

Sometimes, rarely, or never 2
Most of the time 1
AIWflS 0

 

175

 

Table 3. (Cont) Author-generated expected and reported need indices.

 

Expected Need

Reported Need

 

Counseling/Mental Health

 

At enrollment individuals were asked to
report their feelings about the
pregnancy when they first found out
and at the time of enrollment.

Reported feelings about pregnancy at
study enrollment were reportedly:

Unhappy or very unhappy 2
Mixed, happy, or very happy
feelings 0

Regardless, of feelings about
pregnancy, if a participant reports
a history of or current experience
of physical, emotional, or sexual
abuse, the expected need for
counseling is revised to 2

Since enrollment, have you had a
need or concern for counseling or
mental health assistance?

Yes 2 No 0

 

Drug Use

 

Self-reported use of pcp, crack,
cocaine, heroin, uppers, downers,
LSD, or marijuana during current
pregnancy (based on enrollment
8 34 wk info) 2

Self-reported history of using
pcp, crack, cocaine, heroin,
uppers, downers, LSD, or
marijuana prior to current

 

Since enrollment, have you had a
need or concern related to drug
use?

Yes 2 No 0

 

 

 

 

pregnancy
No prior reported illicit drug use 0
Pregnancy Support
Participant is not living with any Since enrollment, have you had a
other adult in her household 2 need or concern for pregnancy
Participant lives with one or more support?
other adults in her household 0 Yes 2 No 0

 

 

176

 

 

 

Table 4. Comparison of percentages of participants in each intervention
group reporting needs or concerns on each of the reported need index
variables.

 

 

 

 

Nurse- Standard Total
CHW of
Reported Need Team Care Sample Compare
Care
(n=217) (n=221) (n=438)
n (%) n (%) n (%)
Pregnancy Education 95 (43.8) 72 (32.6) 167 (38.1) x2 = 5.82*
Pregnancy Education by
Age
16-19 36 (59.0) 25 (33.3) 61 (45.2) x2 = 8.60”
20-24 45 (43.3) 32 (33.3) 77 (38.5) x2 = 2.08
25-29 8 (23.5) 10 (35.7) 18 (29.0) x2 = 1.1 1
30-39 6 (33.3) 5 (21.7) 11 (26.8) x2: .691
Childbirth Education 99 (45.6) 95 (43.0) 194 (44.3) = .308
Infant Care Education 58 (26.7) 50 (22.6) 108 (24.7) 12 = .993
Parenting 46 (21.2) 51 (23.1) 97 (22.1) = .224
Education 113 (52.1) 118 (53.4) 231 (52.7) = .077
Health Insurance 180 (82.9) 178 (80.5) 358 (81.7) = .425
Coverage
Job/Work Assistance 70 (32.3) 79 (35.7) 149 (34.0) = .594
Housing 86 (39.6) 95 (43.0) 181 (41.3) 12: .508
Getting Food and/or 53 (24.4) 54 (24.4) 107 (24.4) = .000
Menu Planning
(nutritious meals)
Domestic Violence 7 (3.2) 11 (5.0) 18 (4.1) x2 = .852
Counseling! Mental 15 (6.9) 20 (9.0) 35 (8.0) X2 = .680
Health
Drug Use 4 (1.8) 5 (2.3) 9 (2.1) x2 = .096
General Pregnancy 75 (34.6) 68 (30.8) 143 (32.6) X2 = .716

Support

 

*p<.05**p5.01

177

Note: degrees of freedom for each Chi-Square assessment = 1

 

Table 5. Enrollment characteristics of study participants by intervention
group (n = 438).

 

 

 

Nurse- Standard
CHW Of Total
Characteristic Team Care Sample Compare
Care (SOC)
(n = 217) (n = 221) (n = 438)
n (%) n (%) n (%)
Age M = 22.26 M = 22.27 M = 22.27 t= -.019
SD = 4.51 SD = 4.95 SD = 4.73 df = 436
16-19 61 (28.1) 74 (33.5) 135 (30.8) X2 = 2.73
20-24 104 (52.0) 96 (43.4) 200 (45.7) df = 3
25-29 34 (15.7) 28 (12.7) 62 (14.2)

30-39 18 (8.3) 23 (10.4) 41 (9.4)
Race/Ethnicity X2 = 1.64
Caucasian 93 (42.9) 89 (40.3) 182 (41.6) df = 3

African American 53 (24.4) 63 (28.5) 116 (26.5)
Hispanic 49 (22.6) 52 (23.5) 101 (23.1)
Other 22 (10.1) 17 (7.7) 39 (8.9)
Not Born in USA 33 (15.2) 35 (15.8) 68 (15.5) x2: .033
df = 1
Marital Status
Married 44 (20.3) 40 (18.1) 84 (19.2) f= 3.11
Divorced, Separated, or 12 (5.5) 22 (10.0) 34 (7.8) df = 2
Widowed
Single, Never Married 161 (74.2) 159 (71.9) 320 (73.1)
Living with father of 108 (50.0) 100 (45.2) 208 (47.5) X2 = .989
baby df = 1
178

 

 

 

Table 5. (Cont) Enrollment characteristics of study participants by
intervention group (n = 438).

 

Nurse- Standard

 

 

CHW Of Total
Characteristic Team Care Sample Compare
Care (SOC)
(n = 217) (n = 221) (n = 438) E
l
n (%) n (%) n (%)
Education
Years completed M = 10.74 M = 10.67 M = 10.70 t= .318 5
SD = 2.62 SD = 2.17 SD = 2.37 df = 436 i
< 12 years 121 (55.8) 127 (57.5) 248 (56.6) X2= .130 i
df = 1
Employed 92 (42.4) 98 (44.3) 190 (43.4) X2 = .169
df = 1
Weeks gestation M = 12.08 M = 12.39 M = 12.24 = -.646
SD = 5.11 SD = 4.97 SD = 5.04 df = 436
Parity x2 = .359
None 93 (42.9) 101 (45.7) 194 (44.3) df = 1
One or more 124 (57.1) 120 (54.3) 244 (55.7)
Abuse History
Ever physically abused 110 (50.7) 113 (51.1) 223 (50.9) X2 = .008
df = 1
Ever emotionally abused 119 (54.8) 122 (55.5) 241 (55.1) x2 = .017
df = 1
Ever sexually abused 46 (21.4) 46 (20.9) 92 (21.1) X2 = .015
df = 1

 

179

 

 

Table 5. (Cont) Enrollment characteristics of study participants by
intervention group (n = 438).

 

 

 

 

Nurse- Standard
CHW Of Total
Characteristic Team Care Sample Compare
Care (SOC)
(n = 217) (n = 221) (n = 438) 1’"
M M M i
(SD) (SD) (SD) 1
Psychological I
Characteristics
Perceived Support 47.53 47.13 47.32 t= .581
(MSPSS) (6.95) (7.39) (7.17) df = 436
Perceived Stress (Cohen) 26.11 26.72 26.42 t= -.810
(7.74) (7.98) (7.86) df = 436
Self-Esteem (Rosenberg) 30.79 30.71 30.75 t= .205
(3.85) (4.07) (3.96) df = 436
Mastery (Pearlin) 20.65 20.61 20.63 t = .115
(2.68) (3.02) (2.85) df = 436
Depressive Symptoms 19.14 19.95 19.55 t= -.777
(CES-D) (10.90) (10.98) (10.93) df = 436
CESD316 n=119 n=129 n=248 x2: .556
(54.8%) (58.4%) (56.6%) df = 1
180

 

 

 

 

FIGURES

 

181

 

:eEoB “cacao...“ 2:35-26. 5 once: corona.— mczofioi 1.. .322 _nsunoecoo .P 0.59".

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

M
m. N
m u a
m. m m N .2230 .2539.
O l x. e v tonnzm
w m mm r 368 1+ 82855
w .u. m
I m
I m. O m ..
n... o I
d m. N
m. o I
m. M I «53956
m nu... 0330.500
m
+
+ H / +
one 2.3.5 A/
T + hm u . n. +
.362 autonom— uooz quoaxm
A b

 

 

182

 

 

 

.eofiocm 95 £3,
990:8 2 3.:an 203 83%.? mzocmmoxm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m W O a
W W m. m o s... so. v s 8263 28588 2.85:8» 5;,
m. E m o W: W: _ocoE 05 c_ 35:58 53 3523ch .
mm M u m.
S m
e Epsom a___o<
3. 3550.2
+ 8.- o -
toaasw
.88 .t. 683.com
. 3 teams.
5.- momcoqmom
5895
at. an. t...
cotoaom 1 :V. uoaooaxm

 

 

 

 

 

 

.05. Engine “.0253: 53538:. 525 23259.3 952.53
use 333...; 38.52.: 9.3: —. _oooE _mzunoocoo 9568 mmEucc n_mzuca 5am .N 2:9“.

183

:5.

+

 

 

toaaaw

 

 

 

 

 

 

>IJOMI9N

ismisml

 

 

 

 

104950

teen >IJ°MISN

 

 

‘-
O

 

 

notwog
alnsoIosgcj

 

 

‘3

O)

o_
l

.8505 one 53
990:8 9 35an 203 33%? anecomoxm

Go. v 3 826128588 2.8566 5;,
.308 05 E moumEzmo £8 nouuemucflw ..

 

Epsom z___a<
$55522

 

 

 

 

Ecosamsow

 

 

muooz
notoaom

 

 

 

+
be.

new moist: amen—ES: use: _. _euoE _aaaaoocoo mazes“ ems—E: n_mzwca 5.8 .n 0.59....

“_on s\\\

 

 

«m _.. w roam—ms.

 

 

 

 

 

momcoamom

 

Lommobw

 

 

+ .3

 

.x. 882

 

A

«EV.

nouooaxm

 

 

 

denounce moans—om «one. pouzflocom a 53> fleas—:22. uczaEzmo

184

 

 

050:0 0:0 53>
0.90.30 0. 0055.00 0.0; 0030:? 0380096

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N W W W m0 m .00. v s 0020.... 200580 202.000 5:,
m. w- ml- w w. m .085 05 s 0905.00 50.. 800.0220 -
w m. m m m 8m
m 0
e 500.00 0:5...
8. 056:5:
+ «.O _- .l Of [@N.
0000.5 1
_0_0ow st. E00.0m-:0m
. .0 30.002
5.- 000:0000m
500050
+ .
00002 .Mm. LN 00002
000000”. f 3... 00.0098
0
.0. 0030...; 00:30:00 .2200 0:00.00

=0 05x5. 500 000205 .000... 2.000.000 0oo5_00=_ E3505 :5? 0.33.0.0: 05.05.00
0:0 00_n0_.0> «00.50:. 050.. _. .009: 0:300:00 05.00. 00505. 0.0220 50... .0 0.39“.

185

Figure 5. Structural equation model testing latent factor
structure of conceptual model 1; Parameters estimated with
maximum likelihood methods.

 

Parnt 12*

 

49"

\
#8 Exp ' >
.06*

1
48A eed\ 3.83*

-.93* ’
/ StressResp '55,,
all
1.81“ ¥56*\81 5;"

 

Basic

 

 

 

Life E

 

 

A_—

 

Parentin

 

 

StrE SW31] [DepE Dep34

 

 

 

 

 

 

 

 

*
5.39 $

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8
3’2
(D
'0
d)
C)
Q; ~75
otheer -.22* : '0‘
. 3.. 53* Q —> z
'52 .68" \7 § .2
8’ :8
co
-.77* {058; 93 85. 8 L
g 0 8 5 to 0 0 III E
b m l— co 8'
.33* -
Easy
TalkE
Mis- 38*
trustEd—

 

 

 

 

 

 

 

 

 

 

 

- .70*

’
1 a"? l _ III
Usefl /4. A 1'85

Note: Standardized parameter estimates do ppt have a maximum value of 1.0
* Path estimates in the model with significant t—values, where p < .05
" LISREL uses this path for scaling and as such does not estimate t-values

 

 

186

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

187

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

eomfiwfififi 5.8:
.o_>mcom @220
“mm m. _
._. 2 :0 +9 manE>w $95
$30 9220 9:39qu + 828.9.
.6 3:30
+ F -
+ |
99699:. 3.29:. 332 +
20an.93 + 9... . +
.anmm + mm: 83.8 All notoaom
.855. @220 i .6 3:90 . _EmSmon.
\4 D 1/
+ +
4/ Eofimmficm \ €9.56.qu + 3.32
+ + omfioaxm
.0 3:30 _EmSmom 392% so
$3535:
382 ootoaom w oEF cocoamom cozoeon.
868$ :85me 358 E3.
scams. “6 8.30

 

 

 

 

 

 

3558 2:35-26. 5 02.533 «SEA—035“. 22o
£333..» .3 coco: ho 9.5.2.2 3585:. no «owns: 2: 95235 .8 .26.: _aafioocoo m 959".

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
   

 

 

 

 

 

 

 

 

 

 

188

 

 

 

 

 

 

 

 

 

 

 

 

r iii
mEoEE>w . macaw
5.8... 028230 .0. \. oozoeon.
.o.>m...om Em... 5 .oon. _Smsmon. _EmEmon.
>8 33 B -+
mm. vor 330 E35
. F B €20
Eoan.o>on. N0. + 36:95 moooaz .
_EcmE «amt.— Al..l 9.53093:— mmD 02me A «mm Etc mm «2.
mo - .mEQmS. . Po. _Smfimon. .mvi/
_msEo A/
5.- No 382
«m F. .3. fimNW/ .mm coaomaxm.
EoEco..>cw
9:01 .3550 E3. 5 EoE>oEEm $9.336:
£50m oE_._. _Smcfion. 8.20m

 

 

 

 

 

 

 

down-53o 53229.2. 33:9... E:E_xaE 5.3 250 E3... 3:0 69.32
3 tau—zone! Econ—.093 5 N 3.3:. _aafioocoo @532 So...— mmsucc airman 5.3 K 2:9".

 

 

 

 

 

mEoEE>w $25
5.3... ozmmoamn. .on. 328.9.

 

 
 
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

189

 

 

 

 

 

 

 

 

 

 

 

EmE. .oom _SmSwon. Al. _mESmon.
.o.>mcom
>8 amp 8. +
.NN. % mo mozoEmE.
E 3.5.0
Eoan.o>oQ .3. + 3.225 macaw...“ .
_mEms. EmE. All. 2.0.8925 mm: ooEow ? «N? no... m .9
mo - .mEoEs. Alfll _Emchn. .mvi/
_msEo 4/
mo. mo wuomz
«om. R F. HimoW/ mo uoEmaxm.
EoEco._>cw
080... 3580on 5 EmE>o_aEm $65336:
mE_on. mg...» .3259. ooEow

 

 

 

 

 

 

 

.cozaEzmo 5825.3 305.9... :55qu 5.3 95.5 00m
ow uoN_Eou:E Ecmflfituo 5 N _ouoE 3333.30 @583 E9: .353: mtg—mg find .m 959".

APPENDICES

190

 

 

 

ABM-
Eligibility criteria for enrollment into the Michigan Maternal Health Services

Study.

 

Eligibility Criteria

 

16 years of age or older

Eligible for Medicaid according to self-reported income and
number of dependents

Resident of Kent County with no stated plans to move out of the
county in the next 18 months

Less than or equal to 24 weeks gestation based on the date of
the woman's last menstrual period

Able to verbally understand either English or Spanish

In the previous 2 years, no reported participation in mental
health therapy from a social worker, psychiatrist, and/or
psychologist or use of medications to treat:

~ Depression

~ Bipolar disorder

- Schizophrenia

- Delusional disorders

~ Multiple personalities

- Panic attacks

~ Post-traumatic stress syndrome

Not currently receiving maternal support services from a nurse
at study enrollment

 

Note: The Michigan Maternal Health Services Study is the large data source (n =
613). The dissertation sample itself is a subset of these participants (n = 438)
who met other specific criteria.

191

 

 

 

Appendix 3.
Reasons that Michigan Maternal Health Services Study cases were
excluded from data analysis for dissertation study.

 

 

Criteria for Exclusion 8. Specific Reason Number of Cases
No delivery of live birth surviving until 6 months 55
Spontaneous abortion 39
Elective abortion 7
Stillbirth delivery 4
Infant death 5
No maternal intent to maintain custody of the 4

child following birth

Infant given up for adoption 4

 

 

Note: Table continued on following page.

192

Appendix B. (Cont)
Reasons that Michigan Maternal Health Services Study cases were

excluded from data analysis for dissertation study.

 

Criteria for Exclusion 8: Specific Reason Number of Cases

 

Completion of both enrollment and late 116
pregnancy research interview

Missing 34 wk interview, 6 month interview, 20
MDI, NCATS, and medical abstract

Missing 34 wk interview, 6 month interview, 56
MDI, and NCATS

Missing 34 wk interview, MDI, NCATS and 4
had only partial information on 6 month
interview

Missing 34 wk interview and had only partial 2

information on the 6 month interivew

Missing 34 wk interview, MDI, and NCATS 2
Missing 34 wk and medical abstract 1
Missing 34 wk interview 31
Total Cases Excluded 175

 

 

193

Appendix C.
Measurement model of perceived personal control construct.

     
   
 

Perceived
Personal
Control

  
 
  

 

 

M.thering

 

 

 

 

 

 

 

Mastery Esteem .bility
53* 7270\\
Can Do good Will be a Can
teach caring good work out
baby for baby mother problems

 

 

 

 

 

 

 

 

 

752 (df= 8, n = 438) = 13.26; p = .103; RMSEA = .039; GFI = .99

* Standardized path estimates with statistically significant (p < .05) t-values

194

 

 

 

Appendix D.

Factor analysis and reliability analysis of 13 Network Orientation Scale

 

 

 

 

items.
Descriptive Statistics

Mean Std. Deviation
NEVTRUST 2.58 .714
OPINIONS 2.35 .621
ADVANTAG 2.398 I 7 .629399
NEVUNDER 2.23662 .591474
PERSONAL 2.74347 .705 893
FIGPROBS 2.24184 .7 I 3419
KEEPPROB 2.07 .631
ANGRY 2.99 .529
GETUPS ET 2.99928 .488631
HELPOUT 2.92 .687
ADVICE 2.89 .536
ASKFAVOR 2.93 .406
SECRET 3.03 .618

 

 

 

 

Item Descriptions

8. NEVTRUST: You can never trust people to keep a secret.

13. OPINIONS: In the past, I have rarely found other people’s opinions helpful

when I’ve had a problem.

18. ADVANTAG: If you confide in other people, they will take advantage of you.

10. NEVUNDER: Other people never understand my problems.

15. PERSONAL: Some things are too personal to talk to anyone about.

12. F IGPROBS: If you can’t figure out your problems, nobody can.

5. KEEPPROB: People should keep their problems to themselves.

14. ANGRY: It really helps when you are angry to tell a friend what happened.
9. GETUPSET: When a person gets upset they should talk it over with a friend.
7. HELPOUT: In the past, friends have really helped me out when I’ve had a

problem.
2. ADVICE: Friends often have good advice to give.

19. ASKFAVOR: It’s okay to ask favors of people.

11. SECRET: Almost everyone knows someone they can trust with a personal

secret.

 

 

 

195

Appendix D. (Cont)
Factor analysis and reliability analysis of 13 Network Orientation Scale
items.

Rotated Factor Comonent Matrix

 

Item Item Name Mistrust Usefulness
No.

(29.29%)* (12.55%)

 

 

8 NEVTRUST .690 -.063
13 OPINIONS .681 -.064
18 ADVANTAG .677 -.203
10 NEVUNDER .639 -.126
15 PERSONAL .589 -.146
12 FIGPROBS .580 -.129

5 KEEPPROB .477 -.248
14 ANGRY -.008 .723

9 GETU PSET -.084 .719

7 HELPOUT -.215 .667

2 ADVICE -.099 .653
19 ASKFAVOR -.226 .51 7
11 SECRET -.218 .466

 

* Indicates variance explained by the extracted factor. There is a correlation of
-.412 between the 2 factors in the sample.

Factors were extracted from the correlation matrix using principal components analysis.
Factors were rotated using varimax rotation with Kaiser Normalization.

196

 

Appendix D. (Cont)

Factor analysis and reliability analysis of 13 Network Orientation Scale

items.

 

Scree Plot
4

 

 

Eigenvalue

The extraction of 2
factors accounted for
41.84% of the total
variance. Each factor had

. an eigen value greater

 

 

Component Number

 

6

1'0 711} 13

than 1.

 

 

Reliability Analysis

 

 

 

 

 

Network Usefulness
Item Item Name Corrected ltem- Alpha if Item
No. Total Correlation Deleted
14 ANGRY .4775 .6614
9 GETUPSET .5155 .6526
7 HELPOUT .5046 .6538
2 ADVICE .4702 .6634
19 ASKFAVOR .3837 .6907
1 1 SECRET .3506 .7044

 

Alpha = .7107 Standardized Item Alpha = .7179

 

197

 

Appendix D. (Cont.)

Factor analysis and reliability analysis of 13 Network Orientation Scale

items.

Reliability Analysis

 

 

Network Mistrust
Item Item Name Corrected ltem- Alpha if Item
No. Total Correlation Deleted
8 NEVTRUST .5071 .7186
13 OPINIONS .5090 .7191
1 8 ADVANTAG .5338 .71 37
10 NEVUNDER .4771 .7262
15 PERSONAL .4477 .7324
12 FIGPROBS .4419 .7340
5 KEEPPROB .4055 .7402

 

Alpha = .7560 Standardized Item Alpha = .7581

 

198

 

Appendix D. (Cont.)
Factor analysis and reliability analysis of 13 Network Orientation Scale

items.

Reliability Analysis

Network Usefulness and Network Mistrust as 2 Subscales of Scale

 

 

Item Item Name Corrected ltem- Alpha if Item II
No. Total Correlation Deleted I
8 NEVTRUST .4521 .7762
13 OPINIONS .4500 .7762
1 8 ADVANTAG .5339 .7682
10 NEVUNDER .4505 .7763
1 5 PERSONAL .4329 .7782
12 FIGPROBS .4107 .7806
5 KEEPPROB .4161 .7793
14 ANGRY (reversed) .3545 .7844
9 GETUPSET (reversed) .4167 .7801
7 HELPOUT (reversed) .4661 .7746
2 ADVICE (reversed) .3801 .7824
1 9 ASKFAVOR (reversed) .3948 .7828
1 1 SECRET (reversed) .3503 .7853

 

Alpha = .7924 Standardized Item Alpha = .7952

 

199

 

Appendix E.
Distribution of expected needs, reported needs and discrepancies between

reported and expected needs among study participants.

 

Total Expected Needs Scale Score

 

n

 

 

 

Frequency

1.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00
3.00 5.00 7.00 9.00 11.00 13.00 15.00 17.00 19.00 21.50

Total Expected Needs Scale Score

 

 

Total Reported Needs Scale Score

 

an
Du

 

 

Frequency

00 400 800 1200 1600 2000
2.00 6.00 10.00 14.00 18.00 22.00

Total Reported Needs Scale Score

 

 

 

200

Apgndix E: (Cont.)
Distribution of expected needs, reported needs and discrepancies between

reported and expected needs among study participants.

 

Frequency

 

PDIFF: Reported Need Minus Expected Need

 

40

30‘

2oJ

1o.

 

0 IIIIIII

-14.50 -11.50 -8.50 -5.50 -2.50 .50 3.50 6.50 10.00

-13.00 -10.00

PDIFF

 

my: _ul. rant?

 

 

-7.00 -4.00 -1.00 2.00 5.00 8.00

 

 

201

 

Appendix F.
Means, standard deviations, and correlations used in path analysis of

conceptual model 1.

 

 

Descriptive Statistics
PTOT 8.2055 4.76676
ZSTRS RSP .000000 3.2264707
FOBFAM34 7.0388 2.09300
ETOT 10.0890 4.08394
ZROSMAST .000000 I .86042 I 8
EASYTALK 2.54 .677
MISTRUST 16.6270 2.94230
TLKUSEFL 17.7664 2.11820
MOMABIL 18.4087 1.64034
MAGE I8 .2055 .40451
MAGIS I9 .3082 .46229
AGE 22.27 4.731
C PSPARNT .08 .268

 

 

 

 

 

Variable Names
PTOT: Reported Needs
ZSTRSRSP: Stress Responses
FOBFAM34: Social Support
ETOT: Expected Needs
ZROSMAST: Mastery & Self-Esteem
EASYTALK: Disclosure Comfort
MISTRUST: Mistrust Network
TLKUSEFL: Network Use Useful
MOMABIL: Mothering Ability Feelings
MAGE18: Maternal Age 5 18
MAGE19: Maternal Age 5 l9
AGE: Maternal Age (Continuous Variable)
CPSPARNT: CPS Involvement as a Parent

 

202

 

_ mc_. 52.- ”3... N8. cccr ccc. occ. mac... wvcr 02 .- _2. Ncc. Hzm<mmmu

 

 

 

 

 

 

 

 

 

 

 

 

_ 39. «mm: «no. on _ .- omc. cccr mccr nnmr c: .. chr we _ .- m0<
_ New. mac... 2 _. mncr Sc. wmc. mum. 3 _. ncc. c2. EMU;
_ c_ _... ccc. chcr 2c. cvc. N—m. Rc. ccc. hm _. w _ mOSZ

_ :L . R— .- ch. ch. N3... _cc. SN... 2: .- 45320.2

_ N3..- _nm. Em. cmc. v3. car 5c. Ammmbvih

_ ccmr _nmr ncc. cvcr _2. c5. HmDmHE—z

_ N2 . wmcr nmc. Nm _ .- 2 _.- métm<m

_ hm _ .- c2. c3..- wvcr rmémOMN

_ chcr mam. mmv. POEM

_ ”3.- _cc.. vm2<mm0m

_ wvm. mmmmy—HmN

_ HOE

 

CAM/$9.0 m0< Smog ”$052 1:202 i_n—mmeqh HmDMhB—z Mégm<mhm§mOmN HOHm vm2<mmOm mmMthmN HOE

 

833.230

._. .060:— _a:uamocoo no «3295 sunn— :_ now»: neon—20.9.00 ten .m:0_ua_>0u Uhmucwum .9505
3:8. m. x. 55mm?

203

Appendix G.
Fit indices for path analysis of conceptual model 1 using manifest variables

and estimating parameters with maximum likelihood methods.

Degrees of Freedom = 29
Minimum Fit Function Chi-Square = 20.37 (P = 0.88)
Normal Theory Weighted Least Squares Chi-Square = 19.86 (P = 0.90)
Estimated Non-centrality Parameter (NCP) = 0.0
90 Percent Confidence Interval for NCP = (0.0 ; 3.44)
Minimum Fit Function Value = 0.047
Population Discrepancy Function Value (F0) = 0.0
90 Percent Confidence Interval for F0 = (0.0 ; 0.0080)
Root Mean Square Error of Approximation (RMSEA) = 0.0
90 Percent Confidence Interval for RMSEA = (0.0 ; 0.017)
P-Value for Test of Close Fit (RMSEA < 0.05) = 1.00
Expected Cross-Validation Index (ECVI) = 0.14
90 Percent Confidence Interval for ECVI = (0.14 ; 0.15)
ECVI for Saturated Model = 0.21
ECVI for Independence Model = 1.47
Chi-Square for Independence Model with 36 DF = 616.81
Independence AIC = 634.81
Model AIC = 51.86
Saturated AIC = 90.00
Independence CAIC = 680.55
Model CAIC = 133.17
Saturated CAIC = 318.70
Normed Fit Index (NFI) = 0.97
Non-Normed Fit Index (NNFI) = 1.02
Parsimony Normed Fit Index (PNFI) = 0.78
Comparative Fit Index (CFI) = 1.00
Incremental Fit Index (IFI) = 1.01
Relative Fit Index (RFI) = 0.96
Critical N (CN) = 1064.80
Root Mean Square Residual (RMR) = 0.21
Standardized RMR = 0.031
Goodness of Fit Index (GFI) = 0.99
Adjusted Goodness of Fit Index (AGFI) = 0.98
Parsimony Goodness of Fit Index (PGFI) = 0.64

204

 

Appendix H.
Fit indices for path analysis of conceptual model 1 using manifest variables

and estimating parameters with generalized least squares methods.

Degrees of Freedom = 29
Minimum Fit Function Chi—Square = 19.77 (P = 0.90)
Normal Theory Weighted Least Squares Chi-Square = 22.99 (P = 0.78)
Estimated Non-centrality Parameter (NCP) = 0.0
90 Percent Confidence Interval for NCP = (0.0 ; 8.22)
Minimum Fit Function Value = 0.045
Population Discrepancy Function Value (F0) = 0.0
90 Percent Confidence Interval for F0 = (0.0 ; 0.019)
Root Mean Square Error of Approximation (RMSEA) = 0.0
90 Percent Confidence Interval for RMSEA = (0.0 ; 0.026)
P-Value for Test of Close Fit (RMSEA < 0.05) = 1 .00
Expected Cross-Validation Index (ECVI) = 0.14
90 Percent Confidence Interval for ECVI = (0.14 ; 0.16)
ECVI for Saturated Model = 0.21
ECVI for Independence Model = 0.71
Chi-Square for Independence Model with 36 DF = 285.87
Independence AIC = 303.87
Model AIC = 54.99
Saturated AIC = 90.00
Independence CAIC = 349.61
Model CAIC = 136.31
Saturated CAIC = 318.70
Normed Fit Index (NFI) = 0.93
Non-Normed Fit Index (NNFI) = 1.05
Parsimony Normed Fit Index (PNF I) = 0.75
Comparative Fit Index (CFI) = 1.00
Incremental Fit Index (IFI) = 1.04
Relative Fit Index (RFI) = 0.91
Critical N (CN) = 1097.03
Root Mean Square Residual (RMR) = 0.24
Standardized RMR = 0.036
Goodness of Fit Index (GFI) = 0.99
Adjusted Goodness of Fit Index (AGFI) = 0.98
Parsimony Goodness of Fit Index (PGFI) = 0.64

205

 

 

 

 

Appendix 1.
Fit indices for path analysis of conceptual model 1 that incorporates path

from mastery/seIf-esteem to mothering ability feelings.

Degrees of Freedom = 27
Minimum Fit Function Chi-Square = 38.53 (P = 0.070)
Normal Theory Weighted Least Squares Chi-Square = 38.30 (P = 0.073)
Estimated Non-centrality Parameter (NCP) = 11.30
90 Percent Confidence Interval for NCP = (0.0 ; 31.80)
Minimum Fit Function Value = 0.088
Population Discrepancy Function Value (F0) = 0.026
90 Percent Confidence Interval for F0 = (0.0 ; 0.074)
Root Mean Square Error of Approximation (RMSEA) = 0.031
90 Percent Confidence Interval for RMSEA = (0.0 ; 0.052)
P-Value for Test of Close Fit (RMSEA < 0.05) = 0.93
Expected Cross-Validation Index (ECVI) = 0.17
90 Percent Confidence Interval for ECVI = (0.15 ; 0.22)
ECVI for Saturated Model = 0.21
ECVI for Independence Model = 1.47
Chi-Square for Independence Model with 36 DF = 616.81
Independence AIC = 634.81
Model AIC = 74.30
Saturated AIC = 90.00
Independence CAIC = 680.55
Model CAIC = 165.78
Saturated CAIC = 318.70
Normed Fit Index (NFI) = 0.94
Non-Normed Fit Index (NNFI) = 0.97
Parsimony Normed Fit Index (PNFI) = 0.70
Comparative Fit Index (CFI) = 0.98
IncrementaI Fit Index (IFI) = 0.98
Relative Fit Index (RF 1) = 0.92
Critical N (CN) = 533.64
Root Mean Square Residual (RMR) = 0.25
Standardized RMR = 0.045
Goodness of Fit Index (GFI) = 0.98
Adjusted Goodness of Fit Index (AGFI) = 0.97
Parsimony Goodness of Fit Index (PGFI) = 0.59

206

 

 

 

 

Appendix J.
Standardized estimates of error terms modeled in latent factor model of
conceptual model 1.

 

Error Term Value

 

Latent Variable Error - Psi Matrix

1. ExpNeed Fixed at 1.0
2. Reported Need1 -.74
3. StressResp .02
4. Social Support .51
5. Disclosure Fixed at 1.0
6. PerControI .13
7. Mothering Abilities .84
Manifest Variable Error -Theta Epsilon Matrix
1. Parnt .99
2. Basic .77
3. LifeE 1.0
4. Parenting .72
5. Basic Needs .52
6. Life Exp .89
7. StrsE .35
8. Str34 .69

 

 

1 See footnote related to this unusual value in text section on measurement model fit in chapter
one.

207

 

Appendix J. (Cont.)
Standardized estimates of error terms modeled in latent factor model of

conceptual model 1.

 

Error Term Value

 

Manifest Variable Error -Theta Epsilon Matrix

 

9. DepE .35
10. Dep34 .70
11. P88 Fam .75
12. Low Farn Involve .89
13. Fam Helper 1.0
14. Low FOB Involve .94
15. FOB Helper .98
16. Easy Talk E .86
17. Mistrust E .51
18. Talk Usefl E .76
19. MastE .54
20 Sele .53
21. Teach Baby .72
22. Care Baby .47
23. Good Mom .54
24. Work Out .72

 

208

Appendix K.
Standardized estimates of covariation in error terms modeled in structural

equation model of conceptual model 1.

 

Error Terms Value

 

Modeled Covariation in Errors

18-1 Talk Useful E & Parnt .17*
14-3 Low FOB Involve & Life E .13*
17-3 Mistrust E & Life E -.11*
21-4 Teach Baby & Parenting -.17*
234 Good Mom & Parenting -.11*
22-3 Care Baby & Basic Nds .12*
8-7 StrE & Str34 .14*
10-8 Dep34 & Str34 .36*
10-9 Dep34 & DepE .06*
17-10 MistrustE & Dep34 .12*
12-11 Low Fam Involve & PSS Family -.53*
13-11 Fam Helper & PSS Family .31*

 

209

 

Appendix K. (Cont.)
Standardized estimates of covariation in error terms modeled in structural

equation model of conceptual model 1.

 

Error Terms Value

 

Modeled Covariation in Errors

13-12 Fam Helper & Low Fam Involve -.26*
14-13 Low FOB involve & Fam Helper .17*
15-13 FOB Helper & Fam Helper -.30*
19-13 Mastery & Fam Helper .06*
15-14 FOB Helper & Low FOB Involve -.39*
18-16 Talk Usefl E & Easy Talk E .15*
19-18 Mastery & Talk Usefl E .01

20-19 Self E & Mast E .28*
21-19 Teach Baby & Mast E .07*

 

210

Appendix L.
Fit indices for conceptual model 1 latent factor model.

Degrees of Freedom = 224
Minimum Fit Function Chi-Square = 592.13 (P = 0.0)
Normal Theory Weighted Least Squares Chi-Square = 547.72 (P = 0.0)
Estimated Non-centrality Parameter (NCP) = 323.72
90 Percent Confidence Interval for NCP = (258.86 ; 396.27)
Minimum Fit Function Value = 1.35
Population Discrepancy Function Value (F0) = 0.74
90 Percent Confidence Interval for F0 = (0.59 ; 0.91)
Root Mean Square Error of Approximation (RMSEA) = 0.058
90 Percent Confidence Interval for RMSEA = (0.051 ; 0.064)
P-Value for Test of Close Fit (RMSEA < 0.05) = 0.022
Expected Cross-Validation Index (ECVI) = 1.60
90 Percent Confidence Interval for ECVI = (1.45 ; 1.77)
ECVI for Saturated Model = 1.37
ECVI for Independence Model = 7.26
Chi-Square for Independence Model with 276 DF= 3123.31
Independence AIC = 3171.31
Model AIC = 699.72
Saturated AIC = 600.00
Independence CAIC = 3293.28
Model CAIC = 1085.97
Saturated CAIC = 2124.67
Normed Fit Index (NFI) = 0.81
Non-Normed Fit Index (NNFI) = 0.84
Parsimony Normed Fit Index (PNFI) = 0.66
Comparative Fit Index (CFI) = 0.87
Incremental Fit Index (IFI) = 0.87
Relative Fit Index (RFI) = 0.77
Critical N (CN) = 204.81
Root Mean Square Residual (RMR) = 1.47
Standardized RMR = 0.087
Goodness of Fit Index (GFI) = 0.90
Adjusted Goodness of Fit Index (AGFI) = 0.87
Parsimony Goodness of Fit Index (PGFI) = 0.68

211

Appendix M.

Conceptual model 2 variable means and standard deviations for Nurse-

CHW Team care group before and after missing data imputation.

 

 

Pre- Post-
Variable2 Imputation Imputation
M (SD) M (SD)
(n) In)
Nurse-CHW Team Care (n=217)
Expected needs 9.72 (4.11) 9.72 (4.11)
(217) (217)
Reported needs 8.30 (4.76) 8.30 (4.76)
(217) (217)
Service helpfulness-34 wks 8.46 (1.89) 8.49 (1.88)
(213) (217)
Service helpfulness-6 mo 8.21 (2.04) 8.20 (2.04)
(187) (217)
Service helpfulness-Composite 16.75 (3.27) 16.69 (3.27)
(185) (217)
Perceived stress-6 wks 19.98 (7.79) 20.05 (7.71)
(205) (217)

 

 

 

2 Items not requiring imputation included: 13 individual items and total scores for expected needs
and reported needs, time points in Kent county, infant sex, pre and postnatal service use,
occurrence of major and/or minor medical condition. Also, “composite” variables, while listed in
the table and utilized in analyses, were not actually imputed. Scores at each wave were imputed

and then transformed to create the final composite variable.

212

 

 

Appendix M. (Cont.)

Conceptual model 2 variable means and standard deviations for Nurse-
CHW Team care group before and after missing data imputation.

 

 

 

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
(n) (n) I-
Perceived stress-6 mo 22.28 (7.90) 22.53 (8.03)
(192) (217) I!
Perceived stress-Composite 42.36 (13.93) 42.58 (1 3.80) '
(188) (217)
Depressive symptoms-6 wks 12.60 (9.13) 12.63 (9.03)
(205) (217)
Depressive symptoms-6 mo 12.36 (9.28) 12.02 (9.08)
(191) (217)
Depressive symptoms — 24.91 (16.21) 24.65 (15.77)
Composite (187) (217)
Time points in Kent County 2.73 (.663) 2.73 (.663)
(217) (217)
Employment-6 weeks .40 (.490) .38 (.486)
(205) (217)
Employment-6 months .59 (.494) .59 (.493)
(194) (217)
Employment -Composite .67 (.471) .67 (.470)
(1 89) (217)

 

213

Appendix M. (Cont.)
Conceptual model 2 variable means and standard deviations for Nurse-

CHW Team care group before and after missing data imputation.

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
(I1) ("I
NCATS Caregiver Total 31.64 (4.45) 31.54 (4.41)
(Maternal interactions) (186) (217)
Clarity of Child’s Cues 7.32 (1.24) 7.30 (1.19)
(186) (217)
HOME Total Score 34.38 (5.17) 34.16 (5.08)
(Home Environment) (180) (217)
Mental Development Index 95.68 (7.15) 95.83 (7.12)
Score (188) (217)
Behavior Rating Scale Total 123.61 (9.67) 124.00 (9.44)
Score (Test Day Behavior) (183) (217)
Infant Sex .52 (.501) .52 (.501)
(217) (217)
Postnatal Service Use (visits) 13.85 (10.46) 13.85 (10.46)
(217) (217)
Pre & Postnatal Service Use 26.87 (16.24) 26.87 (16.24)
(217) (217)

 

214

 

 

 

Appendix M. (Cont.)
Conceptual model 2 variable means and standard deviations for Nurse-

CHW Team care group before and after missing data imputation.

 

 

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
("I (I1)
Days gestation at birth 277.3 (12.39) 277.3 (12.36)
(216) (217)
Birth weight (in grams) 3348.6 (600.9) 3351.7 (601.3)
(216) (217)
Hospital days prior to discharge 2.69 (3.5) 2.69 (3.5)
of infant (216) (217)
NICU days .79 (3.72) .79 (3.71)
(216) (217)
Child Cold at 6 months .44 (.497) .43 (.496)
(185) (217)
Maternal rating of child health 2.13 (1.03) 2.10 (1 .01)
at 6 months (190) (217)
Infant health index (6 item .639 (1.03) .636 (1.03)
composite) (216) (217)
Proximal infant health index .797 (.741) .788 (.734)
(2 item composite) (182) (217)

 

215

 

 

Apgndix N.

Conceptual model 2 variable means and standard deviations for Standard

of Care group before and after missing data imputation.

 

 

Pre- Post-
Variable" Imputation Imputation
M (SD) M (SD)
In) (0)
Expected needs 10.45 (4.04) 10.45 (4.04)
(221) (221)
Reported needs 8.11 (4.79) 8.11 (4.79)
(221) (221)
Service helpfulness-34 wks 7.79 (2.23) 7.78 (2.22)
(219) (221)
Service helpfulness-6 mo 7.61 (2.48) 7.59 (2.48)
(187) (221)
Service helpfulness-Composite 15.39 (3.55) 15.37 (3.45)
(186) (221)
Perceived stress-6 wks 20.47 (8.05) 20.48 (7.96)
(204) (221)
Perceived stress-6 mo 22.39 (8.52) 22.63 (8.39)
(194) (221)

 

 

3 Items not requiring imputation included: 13 individual items and total scores for expected needs
and reported needs, time points in Kent county, infant sex, pre and postnatal service use,
occurrence of major and/or minor medical condition. Also, “composite” variables, while listed in
the table and utilized in analyses, were not actually imputed. Scores at each wave were imputed
and then transformed to create the final composite variable.

216

 

- -127

 

 

Appendix N. (Cont.)

Conceptual model 2 variable means and standard deviations for Standard

of Care group before and after missing data imputation.

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
I") (I1)
Perceived stress-Composite 42.65 (14.50) 43.10 (14.12)
(185) (221)
Depressive symptoms-6 wks 12.95 (8.75) 13.09 (8.71)
(206) (221)
Depressive symptoms-6 mo 12.38 (9.20) 12.74 (9.38)
(196) (221)
Depressive symptoms - 25.08 (15.74) 25.83 (15.80)
Composite (189) (221)
Time points in Kent County 2.77 (.642) 2.77 (.642)
(221) (221)
Employment-6 weeks .36 (.481) .37 (.483)
(206) (221)
Employment-6 months .56 (.498) .56 (.498)
(195) (221)
Employment-Composite .63 (.483) .64 (.482)
(188) (221)

 

217

 

 

 

Appendix N. (Cont.)
Conceptual model 2 variable means and standard deviations for Standard

of Care group before and after missing data imputation.

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
(n) In)
NCATS Caregiver Total 31.90 (4.82) 31.93 (4.87)
(Maternal interactions) (188) (221)
Clarity of Child’s Cues 7.64 (1.28) 7.65 (1.25)
(188) (221)
HOME Total Score 33.93 (5.67) 33.96 (5.60)
(Home Environment) (185) (221)
Mental Development Index 96.14 (6.83) 96.10 (6.71)
Score (190) (221)
Behavior Rating Scale Total 122.80 (11.2) 122.95 (10.84)
Score (Test Day Behavior) (189) (221)
Infant Sex .50 (.501) .50 (.501)
(221) (221)
Postnatal Service Use (visits) 5.66 (6.81) 5.66 (6.81)
(221) (221)
Pre & Postnatal Service Use 9.13 (9.37) 9.13 (9.37)
(221) (221)

 

218

 

 

 

Appendix N. (Cont.)
Conceptual model 2 variable means and standard deviations for Standard

of Care group before and after missing data imputation.

 

 

Pre- Post-
Variable Imputation Imputation
M (SD) M (SD)
I") ("1
Days gestation at birth 277.9 (10.62) 277.9 (10.60)
(220) (221)
Birth weight (in grams) 3406.8 (503.1) 3407.0 (502.0)
(220) (221)
Hospital days prior to discharge 2.69 (3.6) 2.69 (3.6)
of infant (220) (221)
NICU days .69 (3.68) .68 (3.68)
(220) (221)
Child Cold at 6 months .35 (.477) .35 (.478)
(185) (221)
Maternal rating of child health 2.05 (.972) 2.02 (.974)
at 6 months (193) (221)
Infant health index (6 item .559 (.917) .557 (.916)
composite) (220) (221)
Proximal infant health index .678 (.719) .683 (.723)
(2 item composite) (183) (221)

 

219

 

‘— .~h—~. - _ _
.‘I
I,

 

 

Appendix 0.
Fit indices for path analysis of conceptual model 2 using manifest variables

and estimating parameter with maximum likelihood methods.

OVERALL FIT INDICES

Degrees of Freedom = 156
Minimum Fit Function Chi-Square = 122.69 (P = 0.98)
Normal Theory Weighted Least Squares Chi-Square = 117.73 (P = 0.99)
Estimated Non-centrality Parameter (NCP) = 0.0
90 Percent Confidence Interval for NCP = (0.0 ; 0.0)
Minimum Fit Function Value = 0.28
Population Discrepancy Function Value (F0) = 0.0
90 Percent Confidence Interval for F0 = (0.0 ; 0.0)
Root Mean Square Error of Approximation (RMSEA) = 0.0
90 Percent Confidence Interval for RMSEA = (0.0 ; 0.0)
P-Value for Test of Close Fit (RMSEA < 0.05) = 1.00
Expected Cross-Validation Index (ECVI) = 0.61
90 Percent Confidence Interval for ECVI = (0.61 ; 0.61)
ECVI for Saturated Model = 0.49
ECVI for Independence Model = 1.82
Chi-Square for Independence Model with 182 DF = 755.04
Independence AIC = 811.04
Model AIC = 225.73
Saturated AIC = 420.00
Independence CAIC = 953.34
Model CAIC = 500.17
Saturated CAIC = 1487.27
Normed Fit Index (NFI) = 0.84
Non-Normed Fit Index (NNFI) = 1.07
Parsimony Normed Fit Index (PNFI) = 0.72
Comparative Fit Index (CPI) = 1.00
Incremental Fit Index (IFI) = 1.06
Relative Fit Index (RFI) = 0.81
Critical N (CN) = 711.77

220

.. ';

 

 

 

Appendix 0. (Cont.)
Fit indices for path analysis of conceptual model 2 using manifest variables
and estimating parameter with maximum likelihood methods.

Nurse-CHW TeaLm Care

Contribution to Chi-Square = 69.41
Percentage Contribution to Chi-Square = 56.57
Root Mean Square Residual (RMR) = 3.06

Standardized RMR = 0.055
Goodness of Fit Index (GFI) = 0.96

Standard Of Care

Contribution to Chi-Square = 53.28
Percentage Contribution to Chi-Square = 43.43
Root Mean Square Residual (RMR) = 2.77

Standardized RMR = 0.048
Goodness of Fit Index (GFI) = 0.97

221

 

 

 

Appendix P.
Standardized estimates of error terms modeled in path analysis of

conceptual model 2 — Nurse-CHW Team care intervention group.

 

Error Term Value

 

Manifest Endogenous Variable Error
(Theta Epsilon Matrix)

1. Reported needs .79
2. Ordinal service use 1.19
3. Maternal interactions .89
4. Home environment .87
5. Infant mental development .97
6. Clarity of infant cues .94
7. Perceived stress .94
8. Depressive symptoms .44

Manifest Exogenous Variable Error
(Theta Delta Matrix)

.—L

. Expected needs 1.02
. Service helpfulness .95
. Poor infant health 1.16
. Time points in Kent County 1.03
. Postnatal employment .98
. Test day behavior (BRS) .86

‘7 .
. .

 

 

222

 

 

Appendix Q.

Standardized estimates of error terms modeled in path analysis of

conceptual model 2 - SOC intervention group.

 

 

Error Term Value
Manifest Endogenous Variable Error
(Theta Epsilon Matrix)
1. Reported needs .79
2. Ordinal service use .42
3. Maternal interactions 1.02
4. Home environment 1.05
5. Infant mental development .97
6. Clarity of infant cues 1.06
7. Perceived stress .99
8. Depressive symptoms .51
Manifest Exogenous Variable Error
(Theta Delta Matrix)
1. Expected needs .98
2. Service helpfulness 1.05
3. Poor infant health .89
4. Time points in Kent County .97
5. Postnatal employment 1.02
6. Test day behavior (BRS) 1.14

 

223

 

 

 

 

 

 

Appendix R.
Descriptive information about service provider turnover for individuals in
the Nurse-CHW Team care intervention group (n = 196).

 

 

Service Provider Contact n (%) Total
Throughout Entire Intervention Category
n 1%)
Consistent Nurse 8. CHW 43 (21.9)
One Change in Service Provider 45 (23.0)
Consistent Nurse, changed CHW 29 (14.8)
Consistent CHW, changed Nurse 16 (8.2)
Two Changes in Service Providers 61 (31.1)
Consistent Nurse, 2 changes in CHWs 21 (10.7)
Consistent CHW, 2 changes in Nurses 1 (0.5)
Changed Nurse & changed CHW 39 (20.0)
Three Changes in Service Providers 36 (18.4)
Had 2 Nurses and 3 CHWs 24 (12.2)
Had 2 CHWs and 3 Nurses 12 (6.1)
Four Changes in Service Providers 9 (4.6)
Had 2 Nurses and 4 CHWs 2 (1.0)
Had 3 Nurses and 3 CHWs 7 (3.6)
Five Changes in Service Providers 2 (1 .0)
Had 3 Nurses and 4 CHWs 1 (0.5)
Had 3 CHWs and 4 Nurses 1 (0.5)

 

Note: 2 cases missing data and 19 cases excluded because, although
randomized to care, they never accepted/received services from program staff.

224

 

 

REFERENCES

225

 

REFERENCES

Abortion. (1993). In F. Cunningham, P. MacDonald, N. Grant, K. Leveno,
& L. Gilstrap, III (Eds), Willia_m's obstetrics (19th ed., pp. 666). Nowvalk, CT:
Appleton and Lange.

Achat, H., Kawachi, I., Levine, S., Berkey, C., Coakley, E., & Colditz, G.
(1998). Social networks, stress and health-related quality of life. Qualiy of Life
Research j (8), 735-750.

Acosta, O., & Toro, P. A. (2000). Let's ask the homeless people
themselves: A needs assessment based on a probability sample of adults.
American Journal of Cpmmgnity Psychology, 28(3), 343-366.

Adler, N. E., Marmot, M., McEwen, B. S., & Stewart, J. (Eds). (1999).
Socioeconomic stays and health in industrial naftions: Social, psychological, and
biological pathways. New York, NY, US: New York Academy of Sciences.

Ahluwalia, I. B., Dodds, J. M., & Baligh, M. (1998). Social support and
coping behaviors of low-income families experiencing food insufficiency in North
Carolina. Health Education and Behavior 25(5), 599-612.

 

Ali, J., & Avison, W. R. (1997). Employment transitions and psychological
distress: The contrasting experiences of single and married mothers. Journal of
_lie_alth & Social Behavior. 38(4), 345-362.

Allen, R. I., & Petr, C. G. (1998). Rethinking family-centered practice.
American Journal of Orthopsvchiatrv. 68(1), 4-15.

Antonovsky, A. (1973). The utility of the breakdown concept. Social
Science & Megicine, 7, 605-612.

Antonovsky, A. (1987). The salutogenic perspective: Toward a new view
of health and illness. Advances 4(1), 47-55.

 

Apodaca, J. X., Woodruff, S. I., Candelaria, J., Elder, J. P., & Zlot, A.
(1997). Hispanic health program participant and nonparticipant characteristics.
American Journal of Health Behavior. 21(5), 356-363.

Bandura, A. (1986). Social foundations of thought and action: A social
cognitive theogy. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. (2001). Social Cognitive Theory: An agentic perspective.
Armal Review of Psychology. 5;, 1-26.

226

Banyard, V. L. (1999). Childhood maltreatment and the mental health of
low-income women. American Journal of Orthopsvchiatry, 69(2), 161—171.

Barnard, K., Eyres, S., Lobo, M., & Snyder, C. (1983). An ecological
paradigm for assessment and intervention. In T. B. Brazelton & B. M. Lester
(Eds), New Approaches to Developmental Screening of Infm (pp. 199-218).
New York, NY: Elesvier Science Publishing Co.

Barnard, K., & Kelly, J. (1990). Assessment of parent-child interaction. In
S. Meisels & J. Shonkoff (Eds), Handbook of Early Chilghooglntgvention. New
York: Cambridge University Press.

Barnard, K. E., & Eyres, S. J. (1979). Child health assessment, part;
The first year of life (Publication Number DHEW No. HRA 79-25). Washington,
DC: US. Government Printing Office.

Barnard, K. E., Hammond, M. A., Sumner, G., Kang, R., Johnson-
Crowley, M, Snyder, C., Spietz, A., Blackburn, 8., Brandt, P., & Magyary, D.
(1987). Helping parents with preterm infants: Field test of a protocol. Early Child
Development and Care¥(2), 255-290.

Barnard, K. E., & Morisset, C. E. (1995). Preventive health and
developmental care for children: Relationships as a primary factor in service
delivery with at risk populations. In H. E. Fitzgerald & B. M. Lester (Eds),
Children of poverty: Research. health, and policy issues (pp. 167-195). New
York, NY, USA: Garland Publishing, Inc.

Barnet, B., Joffe, A., Duggan, A. K., Wilson, M. D., & Repke, J. T. (1996).
Depressive symptoms, stress, and social support in pregnant and postpartum
adolescents. Archives of Pediatrics and Adolescent Me_d(icine, 150(1), 64-69.

Barnett, A. P. (1987). Consumer perceptions of Black adolescent mothers.
Chilg and Youth Services 9(1), 53-73.

 

Barnett, O. W., Martinez, T. E., & Keyson,, M. (1996). The relationship
between violence, social support, and self-blame in battered women. Journal of
Interpgsonal Violence. 1 1(2), 221-233.

Barnfather, J. 8., & Ronis, D. L. (2000). Test of a model of psychosocial
resources, stress, and health among undereducated adults. Research in Nursiryq
& Health, 23(1), 55-66.

Baum, A., & Posluszny, D. M. (1999). Health psychology: Mapping
biobehavioral contributions to health and illness. Annual Review of Psychology,
50, 137-163.

227

Baumrind, D. (1993). The average expectable environment is not good
enough: A response to Scarr. Chilg_DeveIopment, 64(5), 1299-1317.

Bayley, N. (1993). Bayley Scales of Infant Development (2nd ed.). San
Antonio, TX: The Psychological Corporation.

Becker, M. H. (1974). The Health Belief Model and personal health
behavior. Health Education Monographs, 2, 324-508.

Beckwith, L. (1984). Parent interaction with their preterm infants and later
mental development. Early Child Development and Care 16, 27-40.

 

Bee, H. L., Barnard, K. E., Eyres, S. J., Gray, C. A., Hammond, M. A.,
Spietz, A. L., Snyder, C., & Clark, B. (1982). Prediction of IQ and language skill
from perinatal status, child performance, family characteristics, and mother-infant
interaction. Child Develginent. 53(5), 1134-1156.

Belle, 0., Dill, D., & Burr, R. (1991). Children's network orientations.
Journflf Community Psychology, 19(4), 362-372.

Benasich, A. A., Brooks-Gunn, J., & Clewell, B. C. (1992). How do
mothers benefit from early intervention programs? Journal of Applied
Developmental Psychology, 13(3), 31 1-362.

Berlin, L. J., O'Neal, C., & Brooks-Gunn, J. (1998). What makes early
intervention programs work? The program, its participants, and their interaction.
Zero to Three. 18(4), 4-15.

Bertera, E. M. (1999). Assessing perceived health promotion needs and
interests of low-income older women. Journal of Women's Health and Gender
Based MegicineJU 0), 1323-1336.

Bettes, B. A. (1988). Maternal depression and motherese: Temporal and
intonational features. Child Development, 59(4), 1089-1096.

Binsacca, D. 8., Ellis, J., Martin, D. G., & Petitti, D. B. (1987). Factors
associated with low birthweight in an inner-city population: The role of financial
problems. American Jofiurnal of Pablic Health, 77(4), 505-506.

Birenbaum-Camreli, D., & Carmeli, Y. S. (1996). Patients who get what

they want: Policy and power in the context of prenatal care. Human
Organization, 55(2), 178-182.

228

 

 

 

Blascovich, J., & Tomaka, J. (1991). Measures of self-esteem. In J. P.
Robinson & P. B. Shaver & L. S. Wrightsman (Eds), Measures of Personam
and Social Psychological Attitudes (pp. 121-123). San Diego, CA: Academic
Press.

 

Bogat, G. A., Caldwell, R. A., Guzman, B., Galasso, L., & Davidson, W. 8.,

II. (1998). Structure and stability of maternal support among pregnant and
parenting adolescents. Journal of Commgnity Psychology, 26(6), 549-568.

Bolton, K. D., Chalmers, 3., Cooper, P. A., & Wainer, E. S. (1993). Black
mothers' experiences of a neonatal care unit. Journal of Reproductive and Infant
Psycholagy, 11(4), 229-234.

Booth, C. L., Barnard, K. E., Mitchell, 8. K., & Spieker, S. J. (1987).
Successful intervention with multi-problem mothers: Effects on the mother-infant
relationship. Infant MeLaI Health Journal 8(3), 288-306.

 

Bradley, C. F. (1983). An evaluation of family-centered maternity care.
Women ang Health, 8(1), 35-47.

Bradley, R. H. (1993). Children's home environments, health, behavior
and intervention efforts: A review using the HOME Inventory as a marker
measure. Genetic, Social. anggenegl Psychology Monographs, 119(4), 437-
490.

Bradley, R. H. (1994). The HOME Inventory: Review and reflections.
Aglvances in Child Development and Behavior 25, 241-288.

 

Bradley, R. H., & Caldwell, B. M. (1978). Screening the environment.
American Journal of Orthopsychiatry. 48(1), 114-130.

Bradley, R. H., & Caldwell, B. M. (1980). The relation of home
environment, cognitive competence, and IQ among males and females. Child
Develcmment. 51 (4), 1 140-1148.

Bradley, R. H., & Caldwell, B. M. (1981). The HOME Inventory: A
validation of the preschool scale for Black children. Child Development, 52(2),
708-710.

Bradley, R. H., & Caldwell, B. M. (1988). Using the HOME Inventory to
assess the family environment. E_e<_:liatric Nprsing, 14(2), 97-102.

229

Bradley, R. H., Caldwell, B. M., Rock, S. L., Ramey, C. T., Barnard, K. E.,
Gray, C. A., Hammond, M. A., Mitchell, S. K., Gottfried, A. W., Siegler, L., &
Johnson, D. L. (1989). Home environment and cognitive development in the first
3 years of life: A collaborative study involving six sites and three ethnic groups in
North America. Developmental Psychology, 25(2), 217-235.

Bradley, R. H., Conrvyn, R. F., McAdoo, H. P., & Garcia Coll, C. (2001a).
The home environments of children in the United States Part I: Variations by
age, ethnicity, and poverty status. Child Develomentizw), 1844-1867.

Bradley, R. H., Conrvyn, R. F., McAdoo, H. P., & Garcia CoII, C. (2001b).
The home environments of children in the United States Part II: Relations with
behavioral development through age thirteen. Chilflevelmnent, 72(6), 1868-
1886.

Brammer, L. (1990). Teaching personal problem solving to adults. Journal
of Cognitive Psychothtiapv. 4(3), 267-279.

Brazelton, T. B. (1992). Pregnancy: The first touchpoint, Touchpoints:
Your child's emofipnal and behavioral development (pp. 3-22). Reading, MA:
Perseus Books.

Brooke, O. G., Anderson, H. R., Bland, J. M., Peacock, J. L., & Stewart, C.

M. (1989). Effects on birth weight of smoking, alcohol, caffeine, socioeconomic
factors, and psychosocial stress. British Medical Journal 298, 795-801.

 

Brooks-Gunn, J., Leventhal, T., & Duncan, G. J. (2000). Why poverty
matters for young children: Implications for policy. In J. D. Osofsky & H. E.
Fitzgerald (Eds), WAIMH Hancflgook of Infant Mental Hglth: Parenting and
Child Care (Vol. 3, pp. 89-131). New York, NY: John Wiley & Sons, Inc.

Brown, B. B. (1978). Social and psychological correlates of help-seeking
behavior among urban adults. American Journal of Commanity Psycholqu6(5),
425-439.

Brown, M., & Cudeck, R. (1993). Alternative ways of assessing model fit.

In K. Bollen & J. Long (Eds), Testing structural eguation models (pp. 136-162).
London: Sage.

Bugental, D. B., & Goodnow, J. J. (1998). Socialization processes. In N.

Eisenberg (Ed), Social, emotional, and personalig development (5th ed., Vol. 3,
pp. 389-462). New York: John Wiley & Sons.

Burak, L. J., & Meyer, M. (1997). Using the Health Belief Model to

examine and predict college women's cervical cancer screening beliefs and
behavior. with Care for Women Interrfltional, 18(3), 251-262.

230

 

-o.“ FIT"
.

 

 

Burchinal, M. R., Campbell, F. A., Bryant, D. M., Wasik, B. H., & Ramey,
C. T. (1997). Early intervention and mediating processes in cognitive
performance of children of low-income African American families. Child
Development, 68(5), 935-954.

Burnard, P., & Morrison, P. (1992). Selfficlosarg A contemporary
analysis. Brookfield, USA: Avebury.

Burns, E. I., Doremus, P. C., & Potter, M. B. (1990). Value of health,
incidence of depression, and level of self-esteem in low-income mothers of pre-
school children. Pediatric Nursing, 13, 141-153.

Cairns, K. V., Collins, 8. D., & Hiebert, B. (1994). Adolescents' self-

perceived needs for sexuality education. Canadian Journal of Human Sexualiy,
3(3), 245-251.

Caldwell, B., & Bradley, R. H. (1984). Home Olgervation for Measurement
of the Environmapt. Little Rock, AK: Universiyt of Arkansas at Little Rock.

Campbell, 8. B., Cohn, J. F., & Meyers, T. (1995). Depression in first-time
mothers: Mother-infant interaction and depression chronicity. Developmental
Psycholggr. 31(3), 349-357.

Canty-Mitchell, J., & Zimet, G. D. (2000). Psychometric properties of the
Multidimensional Scale of Perceived Social Support in urban adolescents.
American Joarpal of CommanitsLPsychology, 28(3), 391-400.

Catania, J. A., Kegeles, S. M., & Coates, T. J. (1990). Towards an
understanding of risk behavior: An AIDS risk reduction model (ARRM). Health

Education Quarterly, 17(1), 53-72.

Cecil, H., Stanley, M. A., Carrion, P. G., & Swann, A. (1995).
Psychometric properties of the MSPSS and NOS in psychiatric outpatients.
Mrnal of glinical Psychology, 51(5), 593-602.

Cherry, N. M., & Gear, R. (1987). Young people's perceptions of their
vocational guidance needs: I. Priorities and pre-occupations. British Joarnal of
Guidance and CouLnseIIing, 15(1), 59-71.

Chung, M. J. (1993). Parent education for kindergarten mothers: Needs
assessment and predictor variables. Early Child Development anp Care, 85, 77-
88.

Ciliska, D., Hayward, 8., Thomas, H., Mitchell, A., Dobbins, M.,
Undenrvood, J., Rafael, A., & Martin, E. (1996). A systematic overview of the

231

 

effectiveness of home visiting as a delivery strategy for public health nursing
interventions. Canadian Journal of nglic Health 87(3), 193-198.

 

Clark, R., Paulson, A., & Conlin, S. (1993). Assessment of developmental
status and parent-infant relationships: The therapeutic process of evaluation. In
C. H. Zeanah, Jr. (Ed), Handbook of Infant Mental Health. New York: Guilford
Press.

Clinton, B. (1992). The Maternal Infant Health Outreach Worker Project:
Appalachian communities help their own. In M. Larner & R. Halpern & O.
Harkavy (Eds), Fair start for chiliLren: Lessons learneq froszeven
demonstratiormroiects. New Haven, CT: Yale University Press.

Cohen, S. (1986). Contrasting the Hassles Scales and the Perceived
Stress Scale: Who's really measuring appraised stress? American Paychologis_t,
4_1_(6), 716-718.

Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of
perceived stress. Journal of Health and Social Behavior 24, 385-396.

 

Cohen, S., & Williamson, G. M. (1988). Perceived stress in a probability
sample of the United States. In S. Spacapan & S. Oskamp (Eds), The Social
Psychology of Healfith (pp. 31-67). Newbury Park: Sage Publications.

Collins, N. L., DunkeI-Schetter, C., Lobel, M., & Scrimshaw, S. C. M.
(1993). Social support in pregnancy: Psychosocial correlates of birth outcomes

and postpartum depression. _Jo_urnal of Parsonality and Social Psychology, 65(6),
1243-1258.

Conger, R. D., McCarty, J. A., Yang, R. K., Lahey, B. B., & Kropp, J. P.
(1984). Perception of child, child-rearing values, and emotional distress as
mediating links between environmental stressors and observed maternal
behavior. Chilagevelgment, 55, 2234-2247.

Coon, H., Fulker, D. W., DeFries, J. C., & Plomin, R. (1990). Home
environment and cognitive ability of 7-year—old children in the Colorado Adoption

Project: Genetic and environmental etiologies. Developmental Psychology, 26(3),
459-468.

Covington, M. V. (2000). Goal theory, motivation, and school
achievement: An integrative review. Annaal Review of Psychology, 51, 171-200.

Craig, T. J., & Van Natta, P. A. (1976). Presence and persistence of

depressive symptoms in patient and community populations. American Journal of
Psychiatry, 133(12), 1426-1429.

232

Crnic, K., & Acevedo, M. (1995). Everyday stresses and parenting. In M.
H. Borstein (Ed), Applied and Practical Parenting (Vol. 4, pp. 277-297). Mahwah,
NJ: Lawrence Erlbaum Associates, Inc.

Crnic, K. A., Greenberg, M. T., & Slough, N. M. (1986). Early stress and
social support influences on mothers' and high-risk infants' functioning in late
infancy. mm Mental Health Journal 7(1), 19-33.

 

Crnic, K. A., Ragozin, A. S., Greenberg, M. T., Robinson, N. M., &
Basham, R. B. (1983). Social interaction and developmental competence of

preterm and full-term infants during the first year of life. ChiIC_I_Develoment. 54.
1 199-1210.

Crockenberg, S. (1988). Social support and parenting. In H. E. Fitzgerald
& B. M. Lester & M. W. Yogman (Eds), Theory and Resejarch in Behavioral
Pediatrics (Vol. 4, pp. 141-174). New York, NY: Plenum Press.

Dahlem, N. W., Zimet, G. D., &Walker, R. R. (1991). The
Multidimensional Scale of Perceived Social Support: A confirmation study.
Journal of Clinical Psychology, 47(6), 756-761.

Daro, D., & Gelles, R. (1992). Public attitudes and behaviors with respect
to child abuse prevention. Journal of Interparsonal Violence. 7(4), 517-531.

Daro, D. A., & Harding, K. A. (1999). Healthy Families America: Using
research to enhance practice. The Future of Children 9(1), 152-176.

 

Davis, P. (1982). Converting the need for care into demand for services.
International Dental Jogrnal, 32(3), 271-280.

Derlega, V. J., Metts, S., Petronio, S., & Margulis, S. T. (1993). Self-
disclosure. Newbury Park, CA, USA: Sage Publications, Inc.

Duggan, A., Windham, A., McFarlane, E., Fuddy, L., Rohde, C.,
Buchbinder, S., & Sia, C. (2000). Hawaii's Healthy Start program of home visiting
for at-risk families: Evaluation of family identification, family engagement, and
service delivery. Pediatrics 105(1 pt 3), 250-259.

 

233

DunkeI-Schetter, C., Gurung, R. A. R., Lobel, M., & Wadhwa, P. D. (2001).
Stress processes in pregnancy and birth: Psychological, biological, and
sociocultural influnces. In A. Baum & T. A. Revenson & J. E. Singer (Eds),
Handbook of Health Psychology (pp. 495-518). Mahwah, N.J.: Lawrence
Erlbaum Associates.

DunkeI-Schetter, C., Sagrestano, L. M., Feldman, P., & Killingsworth, C.
(1996). Social support and pregnancy: A comprehensive review focusing on
ethnicity and culture. In G. R. Pierce & B. R. Sarason & I. G. Sarason (Eds),
Headpook of Social Support and the Family (pp. 375-412). New York: Plenum
Press.

Dunn, J., & Brown, J. (1994). Affect expression in the family, children's
understandings of emotions and their interactions with others. Merrill-Palmer
Quarterly, 40, 120-137.

Dunst, C. J., Johanson, C., & Trivette, C. (1991). Family-oriented early

intervention policies and practices: Family-centered or not. Exceptional Children,
5_8(2), 115-126.

Dunst, C. J., & Trivette, C. M. (1988). A family systems model of early
intervention with handicapped and developmentally at-risk children. In D. R.
Powell (Ed), Parent Edgcation as Early Childhood Intervenflr: Emerging
Qrections in Theory. Reseflch and Practice (Vol. 3, pp. 131-179). Nonrvood, NJ:
Ablex Publishing Corporation.

East, P. L., Matthews, K. L., & Felice, M. E. (1994). Qualities of adolescent
mothers' parenting. Journal of Adolescent Heaih. 15(2), 163-168.

Eckenrode, J. (1983). The mobilization of social supports: Some individual
constraints. American Journal of Commgnity Psychology. 11(5), 509-528.

Eisenstadt, J. W., & Powell, D. R. (1987). Processes of participation in a
mother-infant program as modified by stress and impulse control. Journal of
Applied Developmental PsychologyL8(1), 17-37.

Eker, D., & Arkar, H. (1995). Perceived social support: Psychometric
properties of the MSPSS in normal and pathological groups in a developing
country. Social Psychiatry & Psychiatric Epidemiology. 30(3), 121-126.

Emde, R. N., Korfmacher, J., & Kubicek, L. F. (2000). Toward a theory of
early relationship-based intervention. In J. D. Osofsky & H. E. Fitzgerald (Eds),
WAIMH Ha_ncfimok of Infant Mental Health: Early Intervention Evaluation and
Assessment (Vol. 2, pp. 3-32). New York, NY: John Wiley & Sons, Inc.

 

234

Eng, E., & Blanchard, L. (1990). Action-oriented community diagonsis: A
health education tool. International Quarterly of Commnity Haalth Education
1_1_(2), 93-110.

 

Ennis, N. E., Hobfoll, S. E., & Schroder, K. E. (2000). Money doesn't talk,
it swears: How economic stress and resistance resources impact inner-city

women's depressive mood. American Journal of Commgnity Psychology, 28(2),
149-173.

Epperson, D. L., Bushway, D. J., &Wannan, R. E. (1983). Client self-
terminations after one counseling session: Effects of problem recognition,

counselor gender, and counselor experience. Journal of CounselinaPsychology,
Q0), 307-315.

Farran, D. C., Clark, K. A., & Ray, A. R. (1990). Measures of parent-child
interaction. In E. D. Gibbs & D. M. Teti (Eds), lnterdisciplinagj assessment of
infants. Baltimore, MD: Paul Brookes Publishing Company.

Faver, C. A., Crawford, 8. L., & Combs-Onne, T. (1999). Services for child
maltreatment: Challenges for research and practice. Children and Youth Services
Review 21(2), 89-109.

 

Feagans, L. V., & Proctor, A. (1994). The effects of mild illness in infancy
on later development: The sample case of the effects of otitis media (middle ear
effusion). In C. B. Fisher & R. M. Lerner (Eds), Applied Develgmental
Psychology (pp. 139-173). New York: McGraw-Hill.

Feagans, L. V., Sanyal, M., Henderson, F ., Collier, A., & Appelbaum, M. I.
(1987). The relationship of middle ear disease in early childhood to later narrative
and attention skills. do_urnal of Pediatric Psychology. 12, 581-594.

Ferketich, S. L., & Mercer, R. T. (1990). Effects of antepartal stress on

health status during early motherhood. Scholarly Inggiry for Nursing Practice,
5(2), 127-149.

Field, T. (1998). Maternal depression effects on infants and early
interventions. _P_r§ventive Medicine, gm), 200-203.

Field, T., Hallock, N., Ting, G., Dempsey, J., Dabiri, C., & Shuman, H. H.
(1978). A first-year follow-up of high-risk infants: Formulating a cumulative risk
index. Childpevelopment, 49(1), 1 19-131.

Field, T., Healy, B., Goldstein, 8., Perry, 8., & Bendell, D. (1988). Infants

of depressed mothers show "depressed" behavior even with nondepressed
adults. Child Development, 39, 1569-1579.

235

Fiske, S. T., & Pavelchak, M. A. (1986). Category-based versus
piecemeal-based responses: Developments in schema-triggered affect. In R. M.
Sorrentino & E. T. Higgins (Eds), Handbook of Motivflm and Cpgnition:
Foundations of Social Behavior (pp. 167-203). New York: Guilford Press.

Flanagan, P. (1998). Teen mothers: Countering the myths of dysfunction
and developmental disruption. In C. Garcia Coll & J. L. Surrey (Eds), Mothering
against the odds: Diverse voices of contemporary mothers (pp. 238-254). New
York, NY, USA: The Guilford Press.

 

Flaskerud, J. H., & Winslow, B. J. (1998). Conceptualizing vulnerable
populations health-related research. Ngrsing Research, 47(2), 69-78.

Fogel, A. (1990). The process of developmental change in infant
communicative action: Using dynamic systems theory to study individual
ontogenies. In J. Colombo & J. Fagen (Eds), Individual differences in infancy:
Relfliility. sta_bilitv. a. nd prediction (pp. 341-358). Hillsdale, NH: Lawrence
Erlbaum Associates.

Forbes, A., & Roger, D. (1999). Stress, social support and fear of
disclosure. British Joumaiof Health Psychology. 4(2), 165-179.

Ford, B. L. (1983). An overview of hot-deck procedures. In W. G. Madow
& H. Nisselson & I. OIkin & D. B. Rubin (Eds), Incomplete data in sample
surveys (pp. 185-207). New York: Academic Press.

Foster-Fishman, P. G., Salem, D. A., Chibnall, S., Legler, R., & Yapchai,
C. (1998). Empirical support for the critical assumptions of empowerment theory.
American Journal of Commgnity Psychology, 26(4), 507-536.

Garrett, P., Ng'andu, N., & Ferron, J. (1994). Poverty experiences of
young children and the quality of their home environments. Child__Develoment,
6_5_(2), 331-345.

Gifford, P. (1986). The meaning of lumps: A case study of the ambiguities
of risk. In C. R. Janes & R. Stall & S. M. Gifford (Eds), Anthropology and
epidemiology. Boston: Reidel.

Gochman, D. S. (1997). Health behavior research: Definitions and
diversity. In D. S. Gochman (Ed), Handbook of health behavior research 1:
Personal and social determinants (pp. 3-20). New York: Plenum Press.

Goldsmith, D. F., & Rogoff, B. (1995). Sensitivity and teaching by

dysphoric and nondysphoric women in structured versus unstructured situations.
_Develcymental Psychology, 31 (3), 388-394.

236

Gomby, D. S., Culross, P. L., & Behrrnan, R. E. (1999). Home visiting:
Recent program evaluations—analysis and recommendations. The Future of
Children 9(1), 4-26.

 

Gonzalez-Calvo, J., Jackson, J., Hansford, C., & Woodman, C. (1998).
Psychosocial factors and birth outcome: African American women in case
management. dppmal of Health Care for the Poor & Underserved, 9(4), 395-419.

Gorski, P. A. (1998). Perinatal outcome and the social contract-
interrelationships between health and humanity. Journal of Perinayilogy, 18(4),
297-301.

Gottman, J. M., Katz, L. F., & Hooven, C. (1997). Meta-emotion: How
families commuLnicate emotionally. Mahwah, NJ: Lawrence Erlbaum Associates.

Graham, M., Stabile, 1., Powell, A., Pruett, R., Zervigon Hakes, A., &
Butler, B. (1997). Serving pregnant women within Early Head Start: Lessons from
the Panhandle Healthy Start and Early Head Start. 2am to Threay18(2), 31-36.

Grant, T. M., Ernst, C. C., & Streissguth, A. P. (1999). Intervention with
high-risk alcohol and drug-abusing mothers: 1. Administrative strategies of the

Seattle model of paraprofessional advocacy. Journal of Commanity Psychology,
fl (1 ), 1-18.

Gravida-Payne, 8., & Stoneman, Z. (1997). Family predictors of maternal
and paternal involvement in programs for young children with disabilities. Child
Development, 68, 701-717.

Greenberg, M. H., Levin-Epstein, J., Hutson, R. Q., Ooms, T. J.,
Schumacher, R., Turetsky, V., & Engstrom, D. M. (2002). The 1996 welfare law:
Key elements and reauthorization issues affecting children. The Future of
Children, 12(1), 27-57.

Gross, D., Julion, W., & Fogg, L. (2001). What motivates participation and
dropout among low-income urban families of color in a prevention intervention?

Family Relations: Interdisciplinary Jogrnal of Applied Family Studies, 50, 246-
254.

Gruen, R. J., Schuldberg, 0., Nelson, E. A., Epstein, L., Weiss, L., &
Quinlan, D. M. (1994). Network orientation and depressive symptomology:
Development of the Network Utilization Scale. Mrnal of Social and Clinical
Psychology. 13(4), 352-365.

Hall, L. A. (1990). Prevalence and correlates of depressive symptoms in
mothers of young children. Public Health Nursing, 7(2), 71-79.

237

Hall, L. A., Williams, C. A., & Greenberg, R. S. (1985). Supports,
stressors, and depressive symptoms in low-income mothers of young children.
American Jou_rr_ral of nglic Health 75(5), 518-522.

 

Halpern, R. (1993). The societal context of home visiting and related
services for families in poverty. The thgre of Chflen, 3(3), 158-171.

Hardy-Brown, K., Miller, 3., Dean, J., Carrasco, C., & Thompson, S.
(1987). Home-based intervention: Catalyst and challenge to the therapeutic
relationship. Zero to Three, 8(1), 8-12.

Heinicke, C. M., Fineman, N. R., Ruth, G., Recchia, S. L., Guthrie, D., &
Rodning, C. (1999). Relationship-based intervention with at-risk mothers:
Outcome in the first year of life. Infant Mental Health Journal 20(4), 349-374.

 

Hickey, C. A., Cliver, S. P., Goldenberg, R. L., McNeaI, S. F., & Hoffman,
H. J. (1995). Relationship of psychosocial status to low prenatal weight gain
among nonobese black and white women delivering at term. Obstetrics and
Gynecology. 86(2), 177-183.

Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect.
Psychological Review 94(3), 319-340.

Hiroto, D. S., & Seligman, M. E. (1975). Generality of learned

helplessness in man. Journal of Personality and Social Psycholggy. 31(2), 311-
327.

Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiences
predicting changes in self-esteem of sixth- and seventh-grade students. Journal
of Educational Psychology, 82(1), 117-127.

I-Iuber, C., Holditch-Davis, D., & Brandon, D. (1993). High-risk preterm
infants at 3 years of age: Parental response to the presence of developmental
problems. Children's Health Care 22(2), 107-124.

 

lreys, H. T., DeVer, K. A., & Chernoff, R. (2001). Who joins a preventive
intervention? How risk status predicts enrollment. Journal of Commanity

Psychology, 29(4), 417-427.
Izard, C. E. (1993). Organizational and motivational function of discrete

emotions. In M. Lewis & J. M. Haviland (Eds), Handbook of emotions (pp. 631-
641). New York: The Guilford Press.

Joreskog, K., & Sorbom, D. (1996). LISREL 8: User's reference guide.
Lincolnwood, IL: Scientific Software International, Inc.

238

Josten, L. E., Mullett, S. E., Savik, K., Campbell, R., & Vincent, P. (1995).
Client characteristics associated with not keeping appointments for public health
nursing home visits. @lic Health flrsing, 12(5), 305-311.

Jourard, S. M. (1971). The transparent self. New York: Van Nostrand
Reinhold Company.

Kang, R., Barnard, K., Hammond, M., Oshio, 8., Spencer, C., Thibodeaux,
B., & Williams, J. (1995). Preterm infant follow-up project: A multi-site field
experiment of hospital and home intervention programs for mothers and preterm
infants. Mic Health Ngrsing, 12(3), 171-180.

Kaniasty, K. (2000). Help-seeking comfort and receiving social support:
The role of ethnicity and context of need. American Journal of Commgnity

Psychology, 28(4), 545-581.

Kawachi, 1., Kennedy, B. P., Gupta, V., & Prothrow-Stith, D. (1999).
Women's status and the health of women and men: A view from the States.
Social Science & Medicinay48(1), 21-32.

Kazarian, S. 8., & McCabe, S. B. (1991). Dimensions of social support in
the MSPSS: Factorial structure, reliability and theoretical implications. Journal of
Commupity Parchology. 39, 150-158.

Keane, A., Brennan, A. M. W., & Pickett, M. (2000). A typology of
residential fire survivors' multidimensional needs. Wastem Journal of Nursing
Research, 2_2(3), 263-278.

Keen-Payne, R., & Bond, M. L. (1997). Voices of clients and caregivers in
a maternity home. Western Journal of Nursing Research, 19(2), 190-204.

Kemp, V. H., & Page, C. (1987). Maternal self-esteem and prenatal

attachment in high-risk pregnancy. Maternal-Child Nursing Journal, 16(3), 195-
206.

Kessler, R. C. (1997). The effects of stressful life events on depression.
Annual Review of Psychology. 48, 191-214.

Kieffer, C. (1984). Citizen empowerment: A developmental perspective.
Prevention in Human Services 3, 9-36.

 

Kissman, K. (1999). Respite from stress and other service needs of
homeless families. Commgnity Mental Health Journal 35(3), 241-249.

Kitzman, H. J., Cole, R., Yoos, H. L., & Olds, D. (1997). Challenges
experienced by home visitors: A qualitative study of program implementation.
Mrnal of Commgnity Psychology, 25(1), 95-109.

 

239

 

Kitzman, H. J., Yoos, H. L., Cole, R., Korfmacher, J., & Hanks, C. (1997).
Prenatal and early childhood home-visitation program processes: A case
illustration. Journal of Commgnity Psychology, 25(1), 27-45.

 

Koch, R., Lewis, M. T., & Quinones, W. (1998). Homeless: Mothering at
rock bottom. In C. Garcia Coll & J. L. Surrey (Eds), Mothering against the odds:
Diverse voices of contemporary mom (pp. 61-84). New York, NY, USA: The
Guilford Press.

Korfmacher, J. (1998). Examining the service provider in early
intervention. Zero to Three, 18(4), 17-22.

Korfmacher, J., O'Brien, R. A., Hiatt, S. W., & Olds, D. (1999). Differences
in program implementation between nurses and paraprofessionals providing

home visits during pregnancy and infancy: A randomized trial. American Journal
of Pablic Health, 89(12), 1847-1851.

Korsch, B. M. (1984). What do patients and parents want to know? What
do they need to know? Pediatrics 74(5 Pt 2 Suppl), 917-919.

 

Kotch, J. B., Browne, D. C., Ringwalt, C. L., Stewart, P. W., Ruina, E.,
Holt, K., Lowman, B., & Jung, J. W. (1995). Risk of child abuse or neglect in a
cohort of low-income children. Child Abuse and Naqlect, 19(9), 1115-1130.

Kunkle, 8., & Gerrity, D. A. (1997). Gender, expressiveness,
instrumentality, and group social environment as predictors of self-disclosure.
Jamal for Specialists in Grogp Work, 22(3), 214-224.

Kurz, D. (1998). Old problems and new directions in the study of violence
against women. In R. K. Bergen (Ed), Issues in Intimate Violence (pp. 197-208).
Thousand Oaks, CA: Sage Publications, Inc.

Landenburger, K. M. (1998). The dynamics of leaving and recovering from
an abusive relationship. Journal of gastetric, Gynecologic, & Neonatal Nursing,
21(6), 700-706.

Landy, 8. (2000). Assessing the risks and strengths of infants and families
in community-based programs. In J. D. Osofsky & H. E. Fitzgerald (Eds),
WAIMH Handbook of Infant Mental Health: Early lnterventifion. Qaflationfl
Assessment (Vol. 2, pp. 335-376). New York, NY: John Wiley & Sons, Inc.

Lang, P. J. (1984). Cognition in emotion: Concept and action. In C. E.
Izard & J. Kagan & R. B. Zajonc (Eds), Emotions. cognition, and behavior (pp.
192-226). New York: Cambridge University Press.

240

Larner, M. B., Terman, D. L., & Behrrnan, R. E. (1997). Welfare to work:
Analysis and recommendations. The Fu_tt_rre of Children. 7(1), 4-19.

Lawes, G. (1992). Individual parent training implemented by nursery
nurses: Evaluation of a programme for mothers of pre-school children.
_Behaviodral Psychotherapy, 20(3), 239-256.

Lazarus, R. S. (1966). Psychological stress and the coping process. New
York: McGraw-Hill.

Lazarus, R. 8., & Folkman, S. (1984). Stress, appraisal. and coping. New
York: Springer.

Lein, L., Jacquet, S. E., Lewis, C. M., Cole, P. R., &Williams, B. B. (2001).
With the best intentions: Family violence option and abused women's needs.
Violence Against Women, 7(2), 193-210.

Lewis, S. K., Ross, C. E., & Mirowsky, J. (1999). Establishing a sense of
personal control in the transition to adulthood. Social Forces. 77(4), 1573-1599.

Lindy, J. 0., Grace, M. C., & Green, B. L. (1981). Survivors: Outreach to a
reluctant population. American Joranal of Ortmpsychiatjy. 51, 468-478.

Link, B. G., & Phelan, J. (1995). Social conditions as fundamental causes
of disease. Mmal of Health and Social Behavior Extra Issue, 80-94.

 

Linville, P. W. (1987). Self-complexity as a cognitive buffer against stress-

related illness and depression. Journal of Personaliy & Social Psychology, 52(4),
663-676.

Lobel, M., DunkeI-Schetter, C., 8 Scrimshaw, S. C. M. (1992). Prenatal
maternal stress and prematurity: A prospective study of socioeconomically
disadvantaged women. Health Psychology. 11(1), 32-40.

Lotas, M., Penticuff, J., Medoff-Cooper, B., Brooten, D., & Brown, L.
(1992). The HOME scale: The influence of socioeconomic status on the
evaluation of the home environment. Mysing Resaarch. 41, 338-341.

Lowenkron, A. H. (1999). Coping with the stress of premature labor.
Health Care for Women International, 2_0(6), 547-561.

Lozoff, B. (1995). Using the HOME inventory with infants in Costa Rica.
International Jogrnal of Behavioral Development, 18, 277-295.

241

Ludman, E. J., McBride, C. M., Nelson, J. 0., Curry, 8. J., Grothaus, L. C.,
Lando, H. A., & Pirie, P. L. (2000). Stress, depressive symptoms, and smoking
cessation among pregnant women. Health Psychology, 19(1), 21-27.

Lutenbacher, M., & Hall, L. A. (1998). The effects of maternal
psychosocial factors on parenting attitudes of low-income, single mothers with
young children. Nursing Research 47(1), 25-34.

 

Lynch, J. W., Kaplan, G. A., & Shema, S. J. (1997). Cumulative impact of
sustained economic hardship on physical, cognitive, psychological, and social
functioning. New England Journal of Mediaina 337, 1889-1895.

Maccoby, E., & Martin, J. A. (1983). Socialization in the context of the
family: Parent-child interaction. In P. Mussen (Ed), Handbook of Child
Psychology (pp. 1-101). New York: Wiley.

Maccoby, E. E. (1992). The role of parents in the socialization of children:
An historical overview. Developmental Psychology, 28(6), 1006-1017. '

 

Maddux, J. E., & DuChan'ne, K. A. (1997). Behavioral intentions in
theories of health behavior. In D. S. Gochman (Ed), Handbook of haajp
pehavior research 1: Personal and social determinants (Vol. 1, pp. 133-151).
New York: Plenum Press.

Maier, S. F ., & Seligman, M. E. (1976). Learned helplessness: Theory and
evidence. Journal of Experimental Ffiychology: General, 105(1), 3-46.

Marcenko, M. O., & Smith, L. K. (1992). The impact of a family-centered
case management approach. Social Work in Health Care. 17(1), 87-100.

Marcenko, M. O., & Spence, M. (1994). Home visitation services for at-risk
pregnant and postpartum women: A randomized trial. American Journal of
Orthopsychiatry, 64(3), 468-478.

Margolis, P. A., Lannon, C. M., Stevens, R., Harlan, C., Bordley, W. C.,
Carey, T., Leininger, L., Keyes, L. L., & Earp, J. L. (1996). Linking clinical and
public health approaches to improve access to health care for socially
disadvantaged mothers and children. Archives of Pedi_atrics & Adolescent
Macflcine, 150(8), 815-821 .

Matheny, A. P. (1980). Bayley's Infant Behavior Record: Behavioral
components and twin analyses. ChildQevelopment. 51(4), 1157-1167.

Matheny, A. P., Dolan, A. B., &Wilson, R. S. (1974). Bayley's Infant
Behavior Record: Relations between behaviors and mental test scores.
Develwmental Psychm, 10, 696-702.

242

 

 

McCarton, C. M., Brooks-Gunn, J., Wallace, 1. F., Bauer, C. R., & al, e.
(1997). Results at age 8 years of early intervention for Iow-birth-weight premature
infants: The Infant Health and Development Program. Journal of the American
Mecfiml Assaciation, @712), 126-132.

McCurdy, K., & Daro, D. (2001). Parent involvement in family support
programs: An integrated theory. Family Relations: Interdisciplinam Journal of
Applied Family Sttgies. 50(2), 113-121.

McDonald, R. P., & Ringo Ho, M.-H. (2002). Principles and practice in
reporting structural equation analysis. Psychological Methods, 7(1), 64-82.

McDonough, S. C. (2000). Preparing infant mental health personnel for
twenty-first century practice. In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH
Handflok of Infant Mental Healthzjany Intervention, Evaluation and
Assessment (Vol. 2, pp. 535-546). New York, NY: John Wiley & Sons, Inc.

 

McEwen, B. S. (1998). Protective and damaging effects of stress
mediators. New England Journal of Medicine, 338(3), 171-179.

McKelvey, L. M., Fitzgerald, H. E., Schiffman, R. F., & von Eye, A. (2002).
Family stress and parent-infant interaction: The mediating role of coping. Infant
Menfl Health Joarnal, ZEN-2), 164-181.

McKinIay, J. B. (1973). Social networks, lay consultation, and help-seeking
behavior. Social Forces. 51, 275-292.

McLeod, J. D., & Nonnemaker, J. N. (1999). Social stratification and
inequality. In C. S. Aneshensel & J. C. Phelan (Eds), Handbook of the sociology
of mental health (pp. 321-344). New York: Kluwer Academic/Plenum Publishers.

McLoyd, V. C. (1990). The impact of economic hardship on black families
and children: Psychological distress, parenting, and socioemotional
development. ChilcflDevelopment, 61(2), 311-346.

McLoyd, V. C. (1995). Poverty, parenting, and policy: Meeting the support
needs of poor parents. In H. E. Fitzgerald & B. M. Lester (Eds), Children of
poverty: Research, health. andpolicy issgaa (Vol. 23, pp. 269-303). New York,
NY, USA: Garland Publishing, Inc.

McNett, S. C. (1987). Social support, threat, and coping responses and
effectiveness in the functionally disabled. Narsing Research, 36(2), 98-103.
Melchior, L. A., Huba, G. J., Brown, V. B., & Reback, C. J. (1993). A short

depression index for women. Education and Psychological Measurermnt, 53.
1 1 17-1 125.

243

Mercer, R. T., & Ferketich, S. L. (1988). Stress and social support as
predictors of anxiety and depression during pregnancy. Advances in Nursing
Sciance. 10(2), 26-39.

Mercer, R. T., & Ferketich, S. L. (1994). Predictors of maternal role
competence by risk status. Narsiag ReseaLch, 43(1), 38-43.

Middlebrook, J. L., & Forehand, R. (1985). Maternal perceptions of
deviance in child behavior as a function of stress and clinic versus nonclinic
status of the child: An analogue study. Behavior Therapy, 16, 494-502.

Miller, W. R., & Seligman, M. E. (1975). Depression and Ieamed
helplessness in man. Journal of Abnormal Psychology, 84(3), 228-238.

Minde, K. (2000). The assessment of infants and toddlers with medical
conditions and their families. In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH
flfldbook of Infant Mentafil Health: Early lnterventionLEvaluation and
Assesasment (Vol. 2, pp. 81-117). New York, NY: John Wiley & Sons, Inc.

 

Minde, K., Shosenberg, N., Marlon, R, Thompson, 3., Ripley, B., & Burns,
8. (1980). Self-help groups in a premature nursery— a controlled evaluation.
Journal of Pediatrics. 96, 933-940.

Munoz, R. F., Gonzalez, G. M., & Starkweather, J. (1995). Automated
screening for depression: Toward culturally and linguistically appropriate uses of
computerized speech recognition. Hispanic Journal of Behavioral Sciences,
1_7(2), 194-208.

Myers, J. K., & Weissman, M. M. (1980). Use of a self-report symptom
scale to detect depression in a community sample. American Journal of
Psychiatry. 137(9), 1081-1084.

Nations, M. K., Camino, L. A., &Walker, F. B. (1988). 'Nerves': Folk idiom
for anxiety and depression? Social Science & Madicine. 2_6_(12), 1245-1259.

Navaie-Waliser, M., Martin, S. L., Campbell, M. K., Tessaro, I.,
Kotelchuck, M., & Cross, A. W. (2000). Factors predicting completion of a home
visitation program by high-risk pregnant women: The North Carolina Maternal
Outreach Worker Program. American Joml of Public Health. 90(1), 121-124.

Nelson, K., & Allen, M. (1995). Family-centered social services: Moving
toward system change. In P. Adams & K. Nelson (Eds), Reinventing human
services: Community ancflamily-centered practice (pp. 109-125). New York:
Aldin De Gruyter.

244

 

 

 

North, C. 8., & Smith, E. M. (1993). A systematic study of mental health
services utilization by homeless men and women. Social Psychiatm and
Psyc_hiatric Epidemiology, 28(2), 77-83.

Olds, D. (1988). Common design and methodological problems
encountered in evaluating family support services: Illustrations from the
prenatal/early infancy project. In H. B. Weiss & F. H. Jacobs (Eds), Evaluating
Family Programs (pp. 239-265). New York, NY: Aldine De Gruyter.

Olds, D. L., Eckenrode, J., Henderson, C. R., Jr., Kitzman, H., Powers, J.,
Cole, R., Sidora, K., Morris, P., Pettitt, L. M., & Luckey, D. (1997). Long-term
effects of home visitation on maternal life course and child abuse and neglect:
Fifteen year follow-up of a randomized trial. Journal of tha American Mediaal
Association, fllB). 637-643.

 

Olds, D. L., Henderson, C. R., Jr., Kitzman, H. J., Eckenrode, J. J., Cole,
R. E., & Tatelbaum, R. C. (1999). Prenatal and infancy home visitation by nurses:
Recent findings. The th_ure of Children, 9(1), 44-65.

Olds, D. L., & Kitzman, H. (1990). Can home visitation improve the health
of women and children at environmental risk? Pediatrics 86(1), 108-116.

 

Olds, D. L., & Korfmacher, J. (1998). Maternal psychological
characteristics as influences on home visitation contact. Journal of Commgnitv

Psychology, 26(1), 23-36.

Paarlberg, K. M., Vingerhoets, A. J., Passchier, J., Dekker, G. A., & Van
Geijn, H. P. (1995). Psychosocial factors and pregnancy outcome: a review with

emphasis on methodological issues. Journal of Psychosomatic Research 39(5),
563-595.

 

 

Paneth, N. S. (1995). The problem of low birth weight. The Future of
ChMren, 5(1), 19-34.

Parke, R., & Tinsley, B. (1983). The early environment of the at-risk infant:
Exploring the social context. In D. Bricker (Ed), Intervention with__at-risk a_nd
handicapped infants: From research to application (pp. 153-175). Baltimore:
University Park Press.

Parke, R. D., & Buriel, R. (1998). Socialization in the family: Ethnic and
ecological perspectives. In N. Eisenberg (Ed), Social emotional, and personalig
development (5th ed., Vol. 3, pp. 463-552). New York: John Wiley & Sons.

 

Patten, M. A. (1981). Self concept and self esteem: Factors in adolescent
pregnancy. Adolescence. 16(64), 765-778.

245

Patterson, E. T., Douglas, A. 8., Patterson, P. M., & Bradle, J. B. (1992).

Symptoms of preterm labor and self-diagnostic confusion. Musing Resgach.
51(6), 367-372.

Pearlin, L. I. (1999). The stress process revisited. In C. S. Aneshensel & J.
C. Phelan (Eds), Handbook of the Sociology of Mantal Healm (pp. 395-415).
New York: Kluwer Academic/Plenum Publishers.

Pearlin, L. 1., Lieberman, M. A., Menaghan, E. G., & Mullan, J. T. (1981).
The stress process. Journal of Hglth and Social Behavior 22, 337-356.

 

 

Pearlin, L. 1., & Schooler, C. (1978). The structure of coping. Journal of
Health and Social Behavior, 19, 2-21.

Penn, N. E., Levy, V. L., & Penn, B. P. (1986). Professional services

preferred by urban elderly Black women. Anyerican Journal of Social Psychiatgy,
5(2), 129-130.

Petterson, S. M., & Albers, A. B. (2001). Effects of poverty and maternal
depression on early child development. Child Developmenti2, 1794-1813.

Pettit, G. 8., & Bates, J. E. (1984). Continuity of individual differences in

the mother-infant relationship from six to thirteen months. Child Development,
,5_5_(3), 729-739.

Pharis, M. E., & Levin, V. S. (1991). "A person to talk to who really cared":
High-risk mothers' evaluations of services in an intensive intervention research
program. Child WaLare. 70(3), 307-320.

Phinney, J. 8., & Feshbach, N. D. (1980). Non-directive and intrusive
teaching styles of middle- and working-class English mothers. Sritish Jodrnal of

Educational Psychology, 50, 2-9.

Planos, R., Zayas, L. H., & Busch-Rossnagel, N. A. (1997). Mental health
factors and teaching behaviors among low-income Hispanic mothers. Families in
Society, 78(1), 4-12.

Poland, M. L., Giblin, P. T., Waller, J. B., Jr., & Bayer, I. S. (1991).
Development of a paraprofessional home visiting program for low-income
mothers and infants. American Jodrnal of Prevenjve Mermne. 7(4), 204-207.

Powell, C., & Grantham-McGregor, S. (1989). Home visiting of varying
frequency and child development. Pediatrics 84(1), 157-164.

 

Prochaska, J. O. (1995). Common problems: Common solutions. Clinical
Psychology: Science and Practice 2(1), 101-105.

 

246

Prochaska, J. O. (1996). A stage paradigm for integrating clinical and
public health approaches to smoking cessation. Addictive Behaviors 21(6), 721-
732.

 

Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. (1992). In search
of how people change: Applications to addictive behaviors. American
PsychologistA7(9), 1 102-1 114.

Prochaska, J. 0., Johnson, 8., & Lee, P. (1998). The transtheoretical
model of behavior change. In S. A. Schumaker & E. B. Schron & J. K. Ockene &
W. L. McBee (Eds), The handbook of healtaaehavior changa (2nd ed., pp. 59-
84). New York: Springer Publishing Company.

Prochaska, J. O., & Prochaska, J. M. (1999). Why don't continents move?
Why don't people change? Journal of Psychotherapy Integration, 9(1), 83-102.

Radloff, L. (1977). The CES-D Scale: A self-report depression scale for
research in the general population. Applied Psychological Measarement. 1(3),
385-401.

Rawlins, S. R. (1991). Using the connecting process to meet family
caregiver needs. Journal of Professional Ngrsing,_7(4), 213-220.

Reed, C. 8., & Brown, R. E. (2001). Outcome-asset impact model: Linking
outcomes and assets. Eyalgation and Program Planning, 24(3), 287-295.

Rhodes, J. E. (1993). Patterns of service utilization among pregnant and
parenting African-American adolescents. Psychologyof Women Quarterly, 17,
257-274.

Rhodes, J. E., & Jason, L. A. (1990). A social stress model of substance
abuse. Journal of Consulting and Cling Psychology, 58(4), 395-401.

Riessman, F. (1990). Restructuring help: A human services paradigm for
the 19905. American Joarnal of Commdnity Psycholaqy, 18(2), 221-230.

Rini, C. K., DunkeI-Schetter, C., Wadhwa, P. D., & Sandman, C. A.
(1999). Psychological adaptation and birth outcomes: The role of personal
resources, stress, and sociocultural context in pregnancy. Healt_h Psychology,
EM), 333-345.

Roberts, R. E. (1980). Reliability of the CES-D Scale in different ethnic
contexts. Psychiatry Research, 2, 125-143.

Rodriguez, 0., & Green, A. (1997). Parenting stress and anger expression
as predictors of child abuse potential. Child Abuse and Neglect, 21(4), 367-377.

247

Rogoff, B. (1998). Cognition as a collaborative process. In D. Kuhn & R. S.
Siegler (Eds), anitionaperceptipn, and language (5th ed., Vol. 2, pp. 679-744).
New York: John Wiley & Sons.

Rogoff, 3., Ellis, 8., & Gardner, W. (1984). Adjustment of adult-child
instruction according to child's age and task. _Developmental Psychology, 20(2),
193-199.

Roman, L. A., Lindsay, J. K., Moore, J. 8., & Barton, L. R. (in preparation).
Improving the health and development of low-income women: The conceptual
basis for devoloping a community-based maternal support intervention.

Roman, L. A., Lindsay, J. K., Moore, J. 8., Barton, L. R., Baer, L. J.,
Gebben, M. R., & Paneth, N. (in preparation). Engaging low-income, diverse
pregnant women in research: Design and implementation of a community-based
intervention trial.

Roman, L. A., Lindsay, J. K., Moore, J. 8., Paneth, N., Baer, L., Barton, L.
R., Gardiner, J., Goddeeris, J., Shoemaker, A., Fitzgerald, H. E., Gebben, M., &
Davis, J. (2001). Improving the health and development of low-income pregnant
women. Rockville, MD: Maternal and Child Health Bureau, HRSA, PHS, DHHS.

Rosenberg, M. (1979). Conceiving the self. New York, NY: Basic Books.

Rosenberg, M., Schooler, C., Schoenbach, C., & Rosenberg, F. (1995).
Global self-esteem and specific self-esteem: Different concepts, different
outcomes. American Sociological Review, 60(1), 141 -1 56.

Rosenhan, D. L., & Seligman, M. E. P. (1995). Abnormal psychology (3rd
ed). New York: W. W. Norton & Company.

Rosenstock, I. M. (1966). Why people use health services. Milbank
Memorial Fund Quarterly, 44, 94-124.

Ross, C. E., & Mirowsky, J. (1989). Explaining the social patterns of
depression: control and problem solving-or support and talking? Journal of
Health and Social Behavior. 30(2), 206-219.

Roth, K., Eisenberg, N., & Sell, E. R. (1984). The relation of preterm and
full-term infants' temperament to test-taking behaviors and developmental status.

Infant Behavior and Development, 7, 495-505.

Roth, P. L., & Switzer, F. 8., III. (1999). Missing data: Instrument-level
heffalumps and item-level woozles. Research Mews Forgm, 4, 1-5.

248

 

Saami, C., Memme, D. L., & Campos, J. J. (1998). Emotional
development: Action communication, and understanding. In N. Eisenberg (Ed),
Handjbook of childpg/chology (5th ed., Vol. 3, pp. 237-309). New York: John
Wiley & Sons.

Sachs, B., Hall, L. A., Lutenbacher, M., & Rayens, M. K. (1999). Potential
for abusive parenting by rural mothers with low birth weight children. Image: The
Journal of Nursing Scholarsllip231(9), 21-25.

Sattler, J. M. (1992). Assessment of children (3rd ed). San Diego: Jerome
M. Sattler, Publisher, Inc.

Scannapieco, M. (1994). Home-based service program: Effectiveness
with at risk families. Children and Yorgh Services Review, 16(5-6), 363-377.

Schoier, G. (1999). On partial nonresponse situations: The hot deck
imputation method. Paper presented at the meeting of the International Statistical
Institute.

Schumacker, R. E., & Lomax, R. G. (1996). A beginner's guide to
structural equation modeling. Mahwah, NJ: Lawrence Erlbaum Associates.

Scott, A. G., & Sechrest, L. (1989). Strength of theory and theory of
strength. Evaluation and Program Planning, 12(4), 329-336.

Seguin, L., Potvin, L., St. Denis, M., & Loiselle, J. (1995). Chronic
stressors, social support, and depression during pregnancy. Obstetrics and
Gynecolagv. 85(4), 583-589.

Seidman, R. Y., Williams, R., Burns, P., Jacobson, S., Weatherby, F., &
Primeaux, M. (1994). Culture sensitivity in assessing urban Native American
parenting. _Egplic Health Narsing, 11, 98-103.

Seligman, M. E. (1968). Chronic fear produced by unpredictable electric
shock. Journal of Comparative and Physiological Psycholpgy, 66(2), 402-411.

Seligman, M. E. (1974). Depression and learned helplessness. In R. J.
Friedman & M. M. Katz (Eds), The psycholggy of depression: Contemporagy

theory and research (pp. 318).

Seligman, M. E. (1975). Helplessness: On depression. development. and
death. San Francisco, CA: W. H. Freeman.

Seligman, M. E., & Maier, S. F. (1967). Failure to escape traumatic shock.
Journal of Experimental Psychology, 74(1), 1-9.

249

Shea, E., & Tronick, E. Z. (1982). The Maternal Self-Report Inventory: A
research and clinical instrument for assessing maternal self-esteem. In H. E.
Fitzgerald & B. M. Lester & M. W. Yogman (Eds), Theory and Research in
Behavioral Pediatrics (Vol. 4, pp. 101-139). New York, NY: Plenum Press.

Sheppard, M. (1993). The external context for social support: Towards a
theoretical formulation of social support, child care and maternal depression.
Social Work and Social Sciences Review, 4(1), 27-58.

Shonkoff, J. P., & Phillips, D. A. (Eds). (2000). From nagrons to
neighborhoods: The science of early childhood development. Washington, DC:
National Academy Press.

Silber, E., & Trippett, J. S. (1965). Self-esteem: Clinical assessment and
measurement validation. Psychological Reports, 16, 1017-1071 .

Simons, R. L., & Johnson, C. (1996). The impact of marital and social
network support on quality of parenting. In G. R. Pierce & B. R. Sarason & I. G.
Sarason (Eds), Hand_book of Social Sgpport and the Family (pp. 269-288). New
York: Plenum Press.

Simons, R. L., Whitbeck, L. B., Melby, J. N., & Wu, C. (1994). Economic
pressure and harsh parenting. In R. D. Conger & G. H. Elder (Eds), Families in

troubled times: Adapting to changa in rural America (pp. 207-222). New York:
Aldine.

Singer, M. I., Bussey, J., Song, L. Y., & Lunghofer, L. (1995). The
psychosocial issues of women serving time in jail. Social Wgrk, 40(1), 103-113.

Sontag, J. C., & Schacht, R. (1993). Family diversity and patterns of

service utilization in early intervention. Journal of Early Intervention, 17(4), 431-
444.

Spieker, S. J., Solchany, J., McKenna, M., DeKlyen, M., & Barnard, K. E.
(2000). The story of mothers who are difficult to engage in prevention programs.
In J. D. Osofsky & H. E. Fitzgerald (Eds), WAIMH Handbook of Infant Me__ntal
Health: Parentim and Child Care (Vol. 3, pp. 175-209). New York, NY: John
Wiley 8 Sons, Inc.

Steiger, J. H. (1990). Structural model evaluation and modification: An
interval estimation process. Mult_ivariate Bahavioral Research, 2_5_, 173-180.

Stein, R. E. K., Bauman, L. J., & lreys, H. T. (1991). Who enrolls in

prevention trials? Discordance in perception of risk by professionals and
participants. American Jogrnal of Commgnity Psychology, 19(4), 603-617.

250

Stevens, J. H., & Bakeman, R. (1985). A factor analytic study of the
HOME scale for infants. Bevelomental Psychology, 21(6), 1196-1203.

Strecher, V. J., Champion, V. L., & Rosenstock, I. M. (1997). The Health
Belief Model and health behavior. In D. S. Gochman (Ed), Handbook of health
behavior research 1: Personal and social determinants (Vol. 1, pp. 71-91). New
York: Plenum Press.

Sumner, G., & Spietz, A. (Eds). (1994). NCAST caragiver/parent-child
interaction teaching manual. Seattle: NCAST Publications, University of
Washington School of Nursing.

Tableman, B. (1999-2000a). Effective home visiting for very young
children-1. Best PracflcLBriefs. 17, 1-8.

Tableman, B. (1999-2000b). Effective home visiting for very young
children-2. Lest Practice Briefs. 19, 1-12.

Tableman, B. (1999-2000c). Effective home visiting for very young
children—3. Best Practice Briefs 20, 1-8.

 

Tableman, B. (1999-2000d). Summaries of home visiting models for very
young children. Best Practice Briefs, 18, 1-12.

Tableman, B., & Sorenson, P. (1999-2000). Issues in community-based
evaluation: Early intervention home visiting. Best Practicefiriefs, 16, 1-8.

Taylor, D., & Dower, C. (1997). Toward a women-centered health care
system: Women's experiences, women's voices, women's needs. Health Care for
Women International, 18(4), 407-422.

Taylor, R. D., Roberts, D., & Jacobson, L. (1997). Stressful life events,
psychological well-being, and parenting in African American mothers. Journal of
F_ami|y Psychology. 11(4), 436-446.

Taylor, S. E., Repetti, R. L., & Seeman, T. (1997). Health psychology:
What is an unhealthy environment and how does it get under the skin? Annual
Reviaw of Psychology. 48, 411-447.

Teele, D. W., Klein, J. 0., Chase, C., Menyuk, P., Rosner, B. A., & The
Greater Boston Otitis Media Study Group. (1990). Otitis media in infancy and
intellectual ability, school achievement, speech, and language at age 7 years.
Journal of Infectious Diseases, 162, 685-694.

251

Terry, D. J., Mayocchi, L., 8 Hynes, G. J. (1996). Depressive
symptomatology in new mothers: A stress and coping perspective. Journal of
Abnormal Psychology, 105(2), 220-231.

Tesh, E. M., 8 Holditch-Davis, D. (1997). HOME Inventory and NCATS:
Relation to mother and child behaviors during naturalistic observations. Research
in NursinL8 Health. 20, 295-307.

Thoits, P. A. (1995). Stress, coping, and social support processes: Where
are we? What next? Jagrnal of Health and Social Behavior Extra Issue, 53-79.

 

Thomas, J. M., Benham, A., 8 Guskin, K. A. (2000). Intervention-centered
assessment: Opportunity for early and preventive intervention. In J. D. Osofsky
8 H. E. Fitzgerald (Eds), WAIMH Hanabook of Infant MeniHealth: Early
Intervention, Evaluation and Assessment (Vol. 2, pp. 403-440). New York, NY:
John Wiley 8 Sons, Inc.

 

Thompson, R. J., Jr., Catlett, A. T., Oehler, J. M., Gustafson, K. E., 8
Goldstein, R. F. (1998). Home environment and developmental outcome of
african american and white infants with very low birthweight. Children's Health
Care 27(1), 1-14.

 

Tiedje, L. B., Kingry, M. J., 8 Stommel, M. (1992). Patient attitudes
concerning health behaviors during pregnancy: Initial development of a
questionnaire. Health Education anrterly, 19(4), 481-493.

Tietjen, A. M., 8 Bradley, C. F. (1985). Social support and maternal

psychosocial adjustment during the transition to parenthood. Canadian Journal of
Behavioral Science, 17(2), 109-121.

Tolsdorf, C. C. (1976). Social networks, support, and coping: An
exploratory study. Family Process. 15, 407-417.

Torrey, W. C., Mueser, K. T., McHugo, G. H., 8 Drake, R. E. (2000). Self-
esteem as an outcome measure in studies of vocational rehabilitation for adults
with severe mental illness. Psychiatric Service§251(2), 229-233.

Trivette, C., Dunst, C., 8 Hamby, D. (1996). Characteristics and
consequences of help—giving practices in contrasting human service programs.
American Joranal of Commgnity Psychology, 24(2), 273-293.

Trupin, E., Tarico, V., Low, B., Jemelka, R., 8 McClellan, J. (1993).
Children on child protective services caseloads: Prevalence and nature of
serious emotional disturbance. Child Abuse andfi Neglect, 17, 345-355.

Turner, R. J., 8 Lloyd, D. A. (1999). The stress process and the social
distribution of depression. J_o_r_rrna| of Healt_h 8 Social Behavior, 40(4), 374-404.

252

Turner, R. J., Lloyd, D. A., 8 Roszell, P. (1999). Personal resources and
the social distribution of depression. Americap Jaurnal of Somnamitv

Psychology, 27(5), 643-672.

Turner, R. J., 8 Noh, S. (1983). Class and psychological vulnerability
among women: The significance of social support and personal control. Journal
of Hea_lth anaSociaI Behavior 24(1), 2-15.

 

Twohy, K. M., 8 Reif, L. (1997). What do public health nurses really do
during prenatal home appointments? Epplic Health Ngrsiryq, 14(6), 324-331.

Unger, D. G., 8 Wandersman, L. P. (1988). A support program for
adolescent mothers: Predictors of participation. In D. R. Powell (Ed), Parent
Education asanrIy Chilfidhood Intarvention: Emerging Directions in Theom,
Research and Practice (Vol. 3, pp. 105-130). Nonrvood, NJ: Ablex Publishing
Corporation.

Vachon, M. L. 8., Rogers, J., Freedman, L. 8., 8 Freedman, S. J. J.
(1980). A controlled study of self-help intervention for widows. American Journal
of Psychiatry, 137, 1380-1384.

Vaux, A. (1985). Factor structure of the Network Orientation Scale.
Psychological Ramrts. 57. 1181 -1 182.

Vaux, A., Burda, P., 8 Stewart, D. (1986). Orientation toward utilization of
support resources. Jarrnal of Commgnitv Psycholagy. 14, 159-170.

Vines, S. W., 8 Williams-Burgess, C. (1994). Effects of a community
health nursing parent-baby (ad)venture program on depression and other
selected maternal-child health outcomes. Egplic Health Ngrsing, 11(3), 188-194.

Vinokur, A. D., Price, R. H., 8 Caplan, R. D. (1991). From field
experiments to program implementation: Assessing the potential outcomes of an
experimental intervention program for unemployed persons. American Journal of
Sommgnity Psychology. 19(4), 543-562.

Vivian, B. G., 8 Wilcox, J. R. (2000). Compliance communication in home
health care: A mutually reciprocal process. Qualitative Health Research 10(1),
103-116.

 

Wandersman, A., 8 Giamartino, G. A. (1980). Community and individual
difference characteristics as influences on initial participation. American Journal
of Community Psychology, 8(2), 217-228.

253

Wang, J., Siegal, H. A., Falck, R. S., 8 Carlson, R. G. (2001). Factorial
structure of Rosenberg's self-esteem scale among crack-cocaine drug users.
Structural Eguation Modeling,_8(2), 275-286.

Warda, M. R. (2000). Mexican Americans' perceptions of culturally
competent care. Western Jogmal of Ngfiing Research, 2_2(2), 203-224.

Wasserrnan, G. A., Rauh, V. A., Brunelli, S. A., Garcia Castro, M., 8
Necos, B. (1990). Psychosocial attributes and life experiences of disadvantaged
minority mothers: Age and ethnic variations. Chilflevelopment, 61(2), 566-580.

Webster Stratton, C. (1990). Stress: A potential disruptor of parent
perceptions and family interactions. Journal ofilinical Child Psychology. 19(4),
302-312.

Weiss, H., 8 Jacobs, F. (Eds). (1988). Evaluating family support
programs. Hawthorne, NJ: Aldine de Gruyter.

Weissbourd, B., 8 Kagan, S. L. (1989). Family support programs:
Catalysts for change. American Jogml of Qrthopsychia_try, 59, 20-31.

Weisz, S. (1982). Competence, contingency, and the development of
perceived control. miman Bevelopmentfi, 250-281.

Wells, W. M. (1997). Serving families who are hard to reach, maintain, and
help through a universal access home visiting program. Z_ero to Thre._e, 17(4), 22-
26.

Wheeless, L. R., 8 Grotz, J. (1977). The measurement of trust and its
relationship to self-disclosure. Human Commgnication Resgrch, 3(3), 250-257.

Widerstrom, A. H., 8 Nickel, R. E. (1997). Determinants of risk in infancy.
In A. H. Widerstrom 8 B. A. Mowder 8 S. R. Sandall (Eds), Infantaevelopment
aria risk: An introduction (pp. 61—87). Baltimore, MD: Paul H. Brookes Publishing
Company.

Williams, M. L., 8 Meredith, V. (1984). Physician-expectant mother
communication: An analysis of information uncertainty. Commgnication Research
Reports, 1(1), 110-116.

Williams, P. D., 8 Williams, A. R. (1997). Transition from hospital to home

by mothers of preterm infants: Path analysis results over three time periods.
Families, Systems 8 Health, 15(4), 429-446.

254

Wilson, K., Vercella, R., Brems, C., Benning, D., 8 Renfro, N. (1992).
Levels of Ieamed helplessness in abused women. Women and Therapy, 13(4),
53-67.

Woods, N. F., Lentz, M., Mitchell, E., 8 Oakley, L. D. (1994). Depressed
mood and self-esteem in young Asian, Black and White women in America.
Health Care for Women International. 15, 243-262.

Wu, T., 8 Serper, M. R. (1999). Social support and psychopathology in
homeless patients presenting for emergency psychiatric treatment. Journal of
Clinical Psychology, 55(9), 1127-1133.

Yamamoto, K. J., 8 Kinney, D. K. (1976). Pregnant women's ratings of
different factors influencing psychological stress during pregnancy. Psycholpgical
Reports, 39(1), 203-214.

Zayas, L. H., 8 Busch-Rossnagel, N. A. (1992). Pregnant Hispanic
women: A mental health study. Families in Society, 73(9), 515-521.

Zeanah, C. H., 8 McDonough, S. (1989). Clinical approaches to families in
early intervention. Seminars in Perinatology, 13(6), 513-522.

Zich, J. M., Attkisson, C. C., 8 Greenfield, T. K. (1990). Screening for
depression in primary care clinics: The CES-D and the 801. International Journal
of Psychiatry in Medicine, 2_0(3), 259-277.

Zimet, G. D., Dahlem, N. W., Zimet, S. G., 8 Farley, G. K. (1988). The
Multidimensional Scale of Perceived Social Support. Journal of Personalijg
Assessment, Q0), 3041.

Zimet, G. 0., Powell, 8. 8., Farley, G. K., Werkman, 8., 8 Berkoff, K. A.
(1990). Psychometric characteristics of the Multidimensional Scale of Perceived
Social Support. Journal of Personality Assessment, 55(3-4), 610-617.

Zimmerman, J. (1999). Qalaation of the Maternal Support S_ervices
Program. Lansing, MI: Michigan Public Health Institute for the Michigan
Department of Community Health Community Living, Children and Family
Administration Division of Family and Community Health.

Zimmerman, M. A. (1990a). Taking aim on empowerment research: On
the distinction between individual and psychological conceptions. American
Journal of Commnity Psychology, 18(1), 169-177.

Zimmerman, M. A. (1990b). Toward a theory of learned hopefulness: A
structural model analysis of participation and empowerment. Journal of Research
in Peraonalityja, 71-86.

255

Zimmerman, M. A., 8 Rappaport, J. (1988). Citizen participation,
perceived control, and psychological empowerment. American Joarnal of
Comradnity Psychology, 16(5), 725-750.

Zuckerman, B., Amaro, H., Bauchner, H., 8 Cabral, H. (1989). Depressive

symptoms during pregnancy: Relationship to poor health behaviors. American
Journal of Obstetrics and Gynecology. 160(5(pt 1)), 1107-1111.

256