 

.9; if!
I Ifhvlg‘. f ‘d .
v-Uflval’..u {8%, \f-
O to; 1.!
.... 4.? Itcoflntutd .a.
lifelvrbu‘v a:

.3 .13), u”...- . .

.01.”...

I. rtvnir

3.311%:
13.1
a 1.

sh»?

I .5 3L
.1 .r 3,4. 1"!
i j '
are}? no
u. .n
o . .ﬂo} :
com”!

5"”...-

. g
06 I I: It.
vhf...) two...
0 "4" "III I...
o.- ..lur! .arli.

1..., . at!» :5 ,
5A!- I: f ltflo'9'“..

u.) v Ir . ,fo

9.... : ;fc). {.5405} r‘

Illﬂlfiiuv. J a

I
inHXI—p: Nu... tacit )-
tllévﬂuva ‘ If?!)
g

a...
$3
5.4.4.3 .

”1. on,
I! I
a: t. .

1?! o
1.13 . lbw
u ﬁguonoJII-n .

t ‘v ‘
Alaisﬁa r. ”Wag“?
v 2.
4.35947 :01:

00
35341”...

if I
II
VI")

Ill.-

ﬂamm. .
:4»

VII

 

.11 . I
I).
all...

, .-
01.34.

f 3.. I.
:5 l! 1,2!)
1.”: . ff .3.
I I .13 o

l‘

l

.5} .

«.8...

r)! .0.
’l;‘
{K'nvffrlu’ a

A, on." . ’LINH‘.
If! ur pry I

If!» V

(I;

 

 

ITY LIBRARI

Illil’lilillili‘ilillMilli

3 1293 00791 4553

 

 

LIBRARY
latchlgan Shh
l University

 

 

 

This is to certify that the
dissertation entitled
The Influence of Acquiring
the Family Eldercare Role
on Female Employment Adaptation

presented by

Susan Tesch Franklin

has been accepted towards fulfillment
of the requirements for

Ph.D. degreein Family and
Child Ecology

 

 

Ww ,0. 4414/

Major professor

 

Date 11/4/92

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity lnditution
own”:

THE INFLUENCE OF ACQUIRING THE FAMILY ELDERCARE ROLE
ON FEMALE EMPLOYMENT ADAPTATION

By

Susan Tesch Franklin

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Family and Child Ecology

1992

 

Del

interaction
Pm’ious c
“Mined 1
0“ lhe emﬁ
impact of
ml'olveme:
UKUGQ
comallOng
ilCNemar p

Van:
immediate ll
women. WW
rammed s;

 

ABSTRACT

THE INFLUENCE OF ACQUIRING THE FAMILY ELDERCARE ROLE
ON FEMALE EMPLOYMENT ADAPTATION

By

Susan Tesch Franklin

Demographic and social trends precipitate concern about work and family
interactions, but eldercare’s inﬂuence on employment is relatively unknown. Extending
previous cross-sectional research about family care’s effect on employment, this study
examined the inﬂuence of acquiring the eldercare role for a physically disabled relative
on the employment adaptation of women at role inception and three months later. The
impact of caregiver/care recipient characteristics, family support, and caregiver
involvement on three employment adaptation types, Type I (maintain, but alter work),
Type II (leave of absence), and Type III (quit/ retire) was investigated. Analyses included
correlations, oneway ANOVA, paired and independent t—tests, contingency tables,
McNemar procedures, multiple and logistic regressions, and discriminant analysis.

Varied adaptations occurred at both time periods. A major ﬁnding is the
immediate inﬂuence of acquiring the eldercare role on the employment adaptation of
women. Women reporting full-time or part-time employment prior to the caregiver role
recounted substantial changes in that status at Time 1; they did not wait until Time 2.

Caregivers reported using less employment adaptive behaviors at Time 2 than at Time 1.

 

 

Type III

probabilit

 

Fl
lht caregi'
l6) includ
fessionals,
the mm

ism168, an]

Diverse variables inﬂuenced adaptation at Time 1. Caregiver/ care recipient
characteristics were more inﬂuential with Type I and Type II adaptations while family
support and caregiver involvement variables were better predictors of Type III
adaptation. At Time 2, no variables anticipated Type I or Type II adaptations, but
household income and hours of supervision offered good explanatory power for Time 2,
Type III adaptation. The use of Time 1, Type I adaptation did not inﬂuence the
probability of Time 2, Type II or Type III adaptations.

Findings indicate the need to: (a) recognize the immediate inﬂuence of assuming
the caregiver role, (b) increase workplace awareness of potential work/family conﬂicts,
(c) include work/family care content in the educational preparation of family pro-
fessionals, ((1) provide socio-emotional services for families, (e) help policy makers alter
the traditional family image, (f) champion national policies supportive of work/family

issues, and (g) consider ethnicity and relationship quality in future family/work research.

 

 

 

Copl'righ
SUSAN '.
1992

Capyright by
SUSAN 'I'ESCH FRANKLIN
1992

Dedicated
in loving memory of my father
Carl E. Tesch

March 27, 1917 August 27, I992

guidance

availabili

 

 

my disse

Rosema:

Their su

data set

I W'Orke

SOfIWaI

made [l

Sch001

Stepha]

ACKNOWLEDGMENTS

Thanks to Dr. Barbara Ames, my advisor and committee chairperson, for the
guidance and support she has provided throughout my graduate program. Her
availability, encouragement (through good times and bad), and insightful comments about
my dissertation made a difficult task easier.

Thanks to the members of my graduate committee (Dr. Sharon King, Dr.
Rosemary Walker, and Dr. June Youatt) for their interest and support of my progress.
Their suggestions regarding my dissertation gave my work added dimension.

Thanks to Dr. Charles Given and to Dr. Barbara Given for granting access to the
data set I used. The Givens also provided my ﬁrst exposure to research during the time
I worked with them as a graduate research assistant.

Special thanks to my dear husband, Greg. He not only gave me strong
encouragement and support, but also kept my computer going, installed the needed
software, learned to be a great cook, and ignored the dust! His quiet conﬁdence in me
made the stressful years of graduate school manageable.

Special thanks also go to my daughter, Stephanie. Most mothers returning to
school do not have a college senior willing to share an apartment, provide campus
orientation, study together, and think the whole process is fun! My campus year with

Stephanie will always be a special memory.

vi

vii

Thanks to my son, Jeff, who teased me about all the computer time, but also gave
my tired neck a quick rub and told me to drive safely on my many trips to East Lansing.

Thanks to my mother, Evelyn Tesch, and my mother—in-law, Emalee Franklin;
they (and my father) never doubted my ability to accomplish this goal. Also, the support
of my brother, sisters, and their families bolstered me when I needed it most.

This research was supported by grant ”Caregiver Responses to Managing Elderly
Patients at Home” (NIA #2 R01 A0065 84), C. W. Given, Ph.D., Principal Investiga-

tOI'.

TABLE OF CONTENTS

LIST OF TABLES ...................................... xi
LIST OF FIGURES ...................................... xiv
CHAPTER I: Introduction ................................... 1
Scope of the Problem ................................... 2
Statement of the Problem ................................. 8
Purpose of the Study .................................... 9
Relationship of this Study to Grant #2 R01 AGO6584 .............. 10
Signiﬁcance and Generalizability ........................... 11
Conceptual Framework ................................. 13
Human Ecology Theory .............................. l3

Choice and Exchange Theory ........................... 15
Assumptions ..................................... 15
Assumptions of the Ecological Perspective ................ 15
Assumptions of Choice and Exchange Theory .............. l7
Assumptions of this Study .......................... 17

Limitations ......................................... 18
CHAPTER II: Review of Literature ............................ 20
Inﬂuence on Employees and Families ........................ 20
Inﬂuence on the Employer ............................... 25
Inﬂuence on Society ................................... 28
Conceptual Framework ................................. 29
Human Ecological Approach ........................... 29

Choice and Exchange Theory ........................... 34
Summary .......................................... 36
CHAPTER III: Methods ................................... 37
Research Objectives ................................... 41
Research Design ..................................... 41
Research Questions and Hypotheses ......................... 43
Research Variables .................................... 48
Dependent Variables ................................ 49

Type I Employment Adaptation (maintain, but alter work) ....... 49

Arrive Late/ Leave Early ......................... 49

viii

ix

Miss Work without Pay ......................... 49

Take Sick/Personal Days ........................ 49

Change Work Hours ........................... 50

Refuse I ob or Promotion ........................ 50

Kept from Job Hunt/Better Job ..................... 50

Other Work Effects ............................ 51

Type 11 Employment Adaptation (leave of absence) ........... 52

Type 111 Employment Adaptation (quit/retire) .............. 52
Independent Variables ............................... 52
Caregiver Relationship to Care Recipient ................. 52
Caregiver Age ................................. 53
Caregiver Marital Status ........................... 53
Caregiver Education .............................. 5 3
Caregiver Occupation ............................. 54
Caregiver Household Income ........................ 54
Employment Status (prior to caregiving) .................. 55
Current Employment Status ......................... 55
Co-Residence with Care Recipient ..................... 56

Care Recipient Age .............................. 56

Care Recipient Gender ............................ 56

Family Network ................................ 57

Family Help (amount) ............................. 57

Family Assistance (frequency) ........................ 57
Caregiver Total Involvement ......................... 58

Hours of Physical Care ............................ 60

Hours of Supervision ............................. 60
Instrumentation ...................................... 60
Sampling Procedure ................................... 66
Data Collection Techniques ............................... 70
Analysis of Data ..................................... 71
CHAPTER IV: Results .................................... 93
Research Question 1 ................................... 94
Research Question 2 ................................... 99
Research Question 3 .................................. 101
Research Question 4 .................................. 110
Research Question 5 .................................. 118
Research Question 6 .................................. 118
Research Question 7 .................................. 121
Research Question 8 .................................. 126
Research Question 9 .................................. 128
Research Question 10 ................................. 128
Research Question 11 ................................. 132

Research Question 12 ................................. 137

X

CHAPTER V: Discussion ................................. 149
Summary of Findings ................................. 150
Time 1 ........................................ 152

Type I (maintain, but alter work) Employment Adaptation ...... 152

Type II (leave of absence) Employment Adaptation .......... 153

Type III (quit/ retire) Employment Adaptation ............. 154

Time 2 ........................................ 155

All Adaptation Types ............................ 155
Compatibility with the Conceptual Model ................... 155
Interaction of the Variable Groups ....................... 159
Conclusions ....................................... 163
Implications ....................................... 170
Implications for Practice ............................. 170
Implications for Policy .............................. 173
Implications for Further Research ....................... 175

LIST OF REFERENCES ................................. 177
APPENDD( A: Instrument ................................ 188
APPENDD( B: Correlation Matrix ............................ 203
APPENDDI C: Means for Interval Variables ..................... 205
APPENDIX D: UCRIHS Approval Letter ....................... 206

 

Table I

Table 2

Table 3

 

Table 4
Table 5
Table 6

Table 7

Table 9

Table 10
Table 11

Table 15

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11

Table 12

LIST OF TABLES

Probability of Time 1, Type I Inﬂuencing Time 2, Type II and
Type 111 Employment Adaptations

Correlations of Selected Caregiver/ Care Recipient Characteris-
tics with Employment Adaptation

Time 1 Oneway AN OVA by Type I for Selected Care-
giver/Care Recipient Characteristics

Time 2 Oneway ANOVA by Type I for Selected Care-
giver/Care Recipient Characteristics

Contingency Tables by Type II and Type III for Selected
Caregiver/Care Recipient Characteristics

Time 1 Regression of Signiﬁcant Caregiver/Care Recipient
Characteristics and Type I Adaptation

Canonical Discriminant Functions for Selected Signiﬁcant
Caregiver/ Care Recipient Characteristics and Time 1, Type II
Adaptation

Discriminant Function Classiﬁcation for Selected Signiﬁcant
Caregiver/ Care Recipient Characteristics and Time 1, Type II
Adaptation

Canonical Discriminant Function Coefﬁcients for Selected
Signiﬁcant Caregiver/Care Recipient Characteristics and Time
1, Type II Adaptation

Independent t-test for Caregiver Household Income and Time
1, Type 111 Employment Adaptation

Independent t-test for Caregiver Education and Time 2, Type
III Employment Adaptation

Correlations of Family Support Variables with Employment
Adaptation

xi

100

102

103

104

105

112

113

114

114

115

115

119

 

Table 13.

Table 14

 

Table l5

Table 16

Table 17
Table l8
Table 19
Table 20
Table 21

Table 22

 

Table 13

Table 14

Table 15

Table 16

Table 17

Table 18

Table 19

Table 20

Table 21

Table 22

Table 23

Table 24

Table 25

Table 26

xii

Regression for Family Support Variables and Time 1, Type I
Employment Adaptation

Canonical Discriminant Functions for Family Support Vari-
ables and Time 1, Type II and Type III Adaptations

Discriminant Function Classiﬁcation Results for Family
Support Variables and Time 1, Type II and Type III Adapta-
tions

Canonical Discriminant Function Coefﬁcients for Family
Support Variables and Time 1, Type II and Type HI Employ-
ment Adaptations

Correlations of Caregiver Involvement with Employment
Adaptation

Regression for Caregiver Involvement and Time 1, Type I
Employment Adaptation

Canonical Discriminant Functions for Caregiver Involvement
Variables and Time 1, Type II and Type III Adaptations

Discriminant Function Classiﬁcation for Caregiver Involvement
Variables and Time 1, Type II and Type III Adaptations

Independent t-test for Hours of Supervision and Time 1, Type
111 Employment Adaptation

Canonical Discriminant Function Coefﬁcients for Caregiver
Involvement Variables and Time 1, Type II and Type III
Employment Adaptations

Regression of Selected Signiﬁcant Variables and Time 1, Type
I Adaptation

Backward Regression of Selected Signiﬁcant Variables and
Time 1, Type I Adaptation

Canonical Discriminant Functions for Selected Signiﬁcant
Variables and Time 1, Type II and Type III Adaptations

Discriminant Function Classiﬁcation for Selected Signiﬁcant
Variables and Time 1, Type II and Type III Adaptations

122

123

124

126

129

133

134

134

135

136

138

139

140

141

 

Table 2'

 

Table 2i

Table 29

Table 3(

Table 27

Table 28

Table 29

Table 30

xiii

Canonical Discriminant Functions for Selected Signiﬁcant
Variables and Time 2, Type III Adaptation

Discriminant Function Classiﬁcation for Selected Signiﬁcant
Variables and Time 2, Type HI Adaptation

Canonical Discriminant Function Coefﬁcients for Selected
Signiﬁcant Variables and Time 1, Type II and Type III
Adaptations

Canonical Discriminant Function Coefﬁcients for Selected
Signiﬁcant Variables and Time 2, Type III Adaptation

143

143

145

146

Figure 1

Figure I
Figure 3
Figure 4
Figure 5

Figure (

Figure 7

Figure 8

Figure 9

Figure 1

Figure 2
Figure 3
Figure 4
Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

LIST OF FIGURES

Preliminary Conceptual Model for Family Eldercare and
Employment Adaptations

Independent and Dependent Variables
Caregiver Characteristics

Time 2 Employment Status Changes (N = 1 l9)
Frequencies for Employment Adaptations

Disruption of Normal Work and Daily Activities because of
Caregiving Activities

Means for Caregiver Involvement Variables

Transitional Conceptual Model for Family Eldercare &
Employment Adaptations

Concluding Conceptual Model for Family Eldercare &
Employment Adaptations

xiv

38

40
68
95
97

127

130

158

162

CHAPTER I

Introduction

The expanded pool of employees with eldercare obligations solicits increasing
concern in corporate America (Scharlach, Lowe, & Schneider, 1991). The relationship
between caregiving responsibility and work quality and productivity generates the most
employer interest. The escalating numbers of employed caregivers of elderly family
members prompt predictions that eldercare will be the beneﬁt issue of the 1990’s
(Denton, Love, & Slate, 1990; Friedman, 1986; Kola & Dunkle, 1988).

The eldercare phenomenon parallels the emergence of childcare as a beneﬁt issue
thirty years ago. Warshaw, Barr, Rayman, Schachter, and Lucas (1986) state that in the
sixties, companies were forced to consider employer-supported childcare due to the inﬂux
of women into the work force. They further maintain that the workplace has paid little
attention to the issue of employees with the responsibility for incapacitated or disabled
family members. Today, childcare remains a problem; however, the eldercare concern
may rival or surpass the childcare phenomenon. What makes the issue of eldercare and
the workplace so signiﬁcant today?

An examination of converging demographic and social trends provides insight into
the situation. The trends underscore the importance of investigating the inﬂuence of
acquiring the family eldercare role on female employment adaptation. Demographic

changes, the shifting role of women, and the phenomenon of family caregiving merge

  
 

to cballe

increasin;
persons i
The pop
Waldo.
of Amer
be 17.3‘
ﬁlming

2

to challenge corporate ability to maintain a competitive edge through fully productive

employees.

Scope of the Problem

Changing population demographics activate this issue. The eldest cohort is
increasing rapidly while the youngest cohort is decreasing. The percentage of elderly
persons in the United States will swell from approximately 12 % in 1990 to 21% by 2030.
The population 19 years or less, now 29% of the total, will decline to 23.5% by 2030
(Waldo, Sonnefeld, McKusick, & Amett III, 1989). Haber (1989) illustrates the aging
of America at a slightly slower pace by projecting the percentage of persons over 65 to
be 17.3% in 2020 and 21.7% in 2050. Within that group, persons over 85 are the fastest
growing segment and are frequently dependent on long-term care, usually provided by
their families.

In the past century, many changes contributed to the population shift toward
longevity, increased numbers of elderly persons, and declining numbers of younger
people. Hooyman and Ryan (1987) refer to changes in mortality and fertility rates and
altered migration patterns as possible reasons for the dramatic increase in the absolute
number and proportion of older people. Improved preventive medical care and the actual
care and technology connected with acute illness episodes are other explanations for
longer life expectancies. This, however, is not without cost; those improvements in
preventive and acute medical care increase the possibility of chronic disease. The
incidence of chronic illness increases with age and the presence of chronic diseases

escalate the likelihood of dependency on others. The National Council on Aging

3
estimates 6.6 million people over 65 need some physical assistance. They predict that

number will increase to nine million by 2000 and 19 million by 2040 (RNA, #21, 1989).
Thus, more elderly persons and the increased incidence of chronic disease create
challenges for families.

Brody and Brody (1989) maintain that thirty years of research disprove the wide-
spread myth that families do not provide eldercare. In fact, they contend that families
always have been and continue to be the "main provider of long-term health and social
support to the aged" (p. 259). Female family members usually meet the caregiving
challenge. In the Travelers survey, 63% of primary care providers were female (The
Travelers Companies, 1988). Stone, Cafferata, and Sangl (1987) report that 72% of the
caregivers identiﬁed through the 1982 National Long-Tenn Care Survey and Informal
Caregivers Survey were female. A progress report for Caregiver II (core study for this
research) indicates that 76.3% of the caregivers were women (Pohl, Given, & Given,
1991). Other research also documents that women provide the most eldercare (Brody,
Kleban, Johnsen, Hoffman, & Schoonover, 1987; Brody & Schoonover, 1986; Horowitz,
1985).

Women working outside the home in record numbers complicate this issue. In
1950, about 33% of women, 16-64 years of age and from the civilian, non-institutional-
ized population, were in the labor force, compared to 64% in 1985 (U .8. Bureau of the
Census, 1986). The Bureau (1986) notes a decline (91% to 85%) for men during the
same period and predicts that women will comprise 46% of the labor force in 1995. The

annual growth rate for employed women between the years 1975-1990 was 2.8%;

however.

 

(Fulleno:

VT
eipzuidir.T
employed
rate of ind
54 worked
1991). Fu

continue to

 

wOmen beti
Pres
carEgll'ing ‘
Depmmem
1!}ch SUl
force, and
characleristi
employees, i
and (d) On ll
A n3
Retired PeTS<
WETC WQmEn‘.
llallof [be We

4
however, that percentage is predicted to taper to 1.6% during the years 1990-2005

(Fullerton, 1991).

Winﬁeld (1988) maintains that women have been the dominant factor in an
expanding labor force. Brody (1990) concurs and reports that the proportion of
employed women rose from 24% in 1930 to 70% in 1985. She claims the most rapid
rate of increase is with middle-aged women. In 1990, 71% of women from age 45 to
54 worked outside the home, representing an increase from 54.6% in 1975 (Fullerton,
1991). Fullerton predicts that the percentage of employed women in that age group will
continue to climb, increasing to 81.5% by 2005. Creedon (1988) estimates that 75% of
women between 45 and 60 will be in the work force by the year 2000.

Presumably, the number of women coping with the dual roles of work and
caregiving will continue an upward climb. Scharlach (1989), using 1986 United States
Department of Labor information, asserts that the number of employed caregivers will
increase substantially due to the increased number of women workers, an aging work
force, and increased numbers of elders needing assistance. He also notes general
characteristics of the employed caregiver. They are: (a) usually older than other
employees, (b) generally in their 40’s or 50’s, (c) equally represented in job categories,
and (d) on the job longer.

A national survey of caregivers, conducted for the American Association of
Retired Persons and the Travelers Companies Foundation, found that 75 % of caregivers
were women, with an average age of 45 years (The Travelers Companies, 1988). Over
half of the women were employed, with about forty-two percent reporting full-time work

and 13% claiming part-time employment. The remaining women were either not

 

employee

considere.

 

years old.
Wl
the 45-64
that group
for wome:
With chilc
Ho
care, She

homemakz:

lime, SCH

 

Slgnlllcani
“Thin fOr

Wm

5
employed or homemakers (27%) or retired (16%). Sixty-three percent of the women

considered themselves to be primary caregivers of a disabled family member at least 50
years old.

Wisensale and Allison (1988) add to this ﬁnding when they report that women in
the 45—64 age group are most likely to provide care to a disabled parent or husband. Of
that group, 65% of women aged 45-54 work; however, that number decreases to 42%
for women aged 55-64. Caregiving is a life cycle event for many women, beginning
with childcare and ending with eldercare.

Horowitz (1985) notes a gender related division of labor for providing family
care. She reports that the caregiving daughter usually holds primary responsibility for
homemaking, child-rearing, emotional support for family members, and often works full-
time. Several years later, the ﬁndings remain consistent. Finley (1989), too, describes
signiﬁcant gender differences in caregiving. She also notes that, in spite of more role
conﬂict for women than for men, women are more likely to provide eldercare. Society
seldom places the same multiple demands on men that it does on women.

When both men and women provide care, the hours involved by gender varies.
Anastas, Gibeau, and Larson (1990) report the mean hours of care for female caregivers
is 10.9 per week while the average for their male counterparts drops to 5.8 hours per
week. Results of the Travelers survey indicate even more disparity; 16.1 hours for
women versus 5.3 hours for men (The Travelers Companies, 1988). High care hours
have the potential to negatively inﬂuence the ability to maintain employment responsibili-

ties.

6

Brody (1985) refers to the "woman in the middle,” a woman surrounded by
multiple pressures. This woman has principal responsibility for the care of a dependent
parent (usually the mother), experiences competing demands on time and energy, suffers
strains as a result of parent care, possibly has the ”empty nest” reﬁlled, and often works
full or part-time. Sidel (1986, p. 167) echoes the dilemma of women in the middle and
notes "younger women are increasingly torn among their responsibilities toward their
own families, their responsibilities toward the older persons, and these days a job as
well." A study by Young and Kahana (1989) illustrates employment as a signiﬁcant
predictor of adverse reaction during the caregiving period.

The Older Women’s League publication, Failing America ’s Caregivers: A Status
Report on Women Who Care (1989) states that 1.8 million women are simultaneously
caring for children and elderly relatives. Half of these women are in the work force; 20
percent of the women have a parent living in their home. Obviously, substantial numbers
of women provide care to an elderly family member while also attempting to maintain
employment and manage additional family responsibilities. The number of eldercare
hours provided varies with the characteristics of the care recipient, the involvement of
other family members and friends, and the utilization of formal services. Scharlach
(1989) reports that 20-28% of employees provide care at any one time and give 610
hours per week of aid to impaired elders. Results of the Travelers survey, where 20%
of employees over age 30 provided an average of 10.2 care hours per week, support that
statistic (The Travelers Companies, 1988). Gibeau (1988) notes the primary caregiver
provides 12.1 hours of care per week; a slightly higher mean than the preceding

references.

 

 

ll}
some car:
caregivers
job! In ar
Connectic
spent an a

Bit
ry insight

First. they

 

aid and d
conﬂicted
decreasing
TOUTS for 1
lladi’uonal
[0 {huge Wt
Wk Cle
rehoned '1,
Th
disCOl'ey-ed
one and a
Alﬁqough
illiimncC

7

While reports of means are useful, they also obscure the intense involvement
some caregivers contribute. For example, in the Travelers survey, eight percent of all
caregivers spent more than 35 hours per week providing care; almost a second full-time
job! In another example of high care hours, Wisensale and Allison (1986) refer to a
Connecticut survey where most caregivers were females between 40-59 years of age who
spent an average of 82 hours monthly (20.5 hours/week) providing care.

Brody, Kleban, Johnsen, Hoffman, and Schoonover (1987) provide complementa-
ry insight into the impact of caregiving hours per week on the employed caregiver.
First, they refer to the persevering worker as one who provides 12.7 hours per week of
aid and does not consider quitting her job to do so. The second employee example, a
conﬂicted worker, provides 23.3 care hours per week and is considering either
decreasing work hours or stopping employment. Of interest, the number of caregiving
hours for the conﬂicted worker is just under the 24.1 care hours per week that the
traditional non-working caregiver provides! Finally, the third employee category refers
to those who quit work to provide care; these caregivers give about 38 hours of help per
week. Clearly, much care hour variation exists from the mean of 10-12 hours per week
reported in some studies.

This variance partially relates to the dependencies of the elder. Scharlach (1989)
discovered that employees who cared for cognitively impaired elders spent approximately
one and a half times as many hours in caregiving as did other employee caregivers.
Although ethnicity appears to create diversity, there is limited literature reporting ethnic
differences in caregiving. Most inferences arise from case experience and indirect

sources of data (Sakauye, 1988). White-Means and Thornton (1990) examined data from

 

the 1982 .‘
to study A
hmdmm
descent to
of English
of care per

Tn;
or ethnic
restrictions
experience
139011 that
demands.

Ho
necessam}
daugluers .
Caregiver g

 

8
the 1982 National Long-Term Care Survey and National Survey of Informal Caregivers

to study Americans with different ethnic backgrounds in relation to caregiving. They
found that the mean hours of care per day ranged from 2.45 for caregivers of German
descent to 4.20 for those of African origin (17.15 to 29.40 hours per week). Caregivers
of English extraction provided 2.85 hours while those of Irish lineage gave 2.94 hours
of care per day (19.95 to 20.58 hours per week) to their ailing elders.

The hours of care provided create costs for caregivers, regardless of employment
or ethnic status. Horowitz (1985) and Buglass (1989) report that emotional strain,
restrictions on time and freedom, and economic hardships generate the burdens
experienced by some caregivers. Hooyman and Ryan (1987) reinforce this ﬁnding and
report that many women are physically and emotionally exhausted by multiple role
demands.

However, choosing to leave the work force in order to provide care does not
necessarily decrease burden. Barnes, Given, and Given (1991) report that adult
daughters who ended employment in order to give care evidenced the greatest nwd for
caregiver support and were at increased risk for depression. Clearly, multiple issues

surround the caregiving experience.

Statement of the Problem

The growing population of employed female family caregivers and the possible
negative interactions between caregiving and employment responsibilities stimulate
increasing concern. The long-held notion that work and family denote two separate

worlds has been under intense bombardment and is weakening under the pressures of a

TWO°g€lit

and life

 

lbce, 19
family, b

interactio

Purpose

9
two-gender work force (Winﬁeld, 1988). The increased ﬂexibility of former rigid role

and life patterns heightens anxiety about the interaction of work and family (Frone &
Rice, 1987). Managers are beginning to see the close connection between work and
family, but its inﬂuence is unknown. The present study addressed this family and work

interaction.

Purpose of the Study

The purpose of the study was to investigate the inﬂuence of acquiring the role of
caregiver for a physically disabled elderly relative on the employment adaptation of
women. Three types of adaptation were considered in the study. Type I adaptation
involves maintaining, but altering one’s work role and includes: (a) arriving late/leaving
early, (b) missing work without pay, (c) taking sick/personal days, ((1) changing work
hours, (e) refusing jobs or promotions, (t) not looking for jobs or better jobs, and (g)
other ways that caregiving may affect work. Type II adaptation is taking a leave of
absence and Type III adaptation is quitting or retiring from employment in order to
provide care. Unless speciﬁed otherwise, the study objectives encompassed each type
of adaptation.

The study included examination of (a) the immediate and delayed (after three
months) employment adaptation for the female employee; (b) the inﬂuence of selected
caregiver and care recipient characteristics (caregiver age, relationship to care recipient,
marital status, education, occupation, household income, employment status, co-residence
status, and care recipient age and gender) on employment adaptation at Time 1 and Time

2; (c) the inﬂuence of family support (number in network, amount of help, and frequency

 

of familj
Time 2;

of daily

 

 

 

activities

 

Managing
researchers
Mugging
The extensi.
SCH'iceS N

The
The ﬁrs: l
Caregmng‘
Provide Ca-

meet their

the Wes 0

10

of family assistance with eldercare) on caregiver employment adaptation at Time 1 and
Time 2; and (d) the inﬂuence of caregiver involvement (total assistance with activities
of daily living (ADL), instrumental activities of daily living (IADL), and health care
activities (HCA) and the hours of both physical care and supervision) on employment
adaptation at Time 1 and Time 2; and (e) the probability of the caregiver displaying Type

H or Type HI adaptation at Time 2 given the level of Type I adaptation at Time 1.

Relationship of this Study to Grant #2 R01 AGO6584

This study utilized data from the core research effort, "Caregiver Responses to
Managing Elderly Patients at Home" (Caregiver II). Caregiver H, conducted by
researchers at Michigan State University, is an extension of ”Caregiver Responses to
Managing Elderly Patients at Home" (Caregiver I). The original study began in 1986;
the extension (Caregiver II) continues through 1992 and is funded by Health and Human
Services National Institute on Aging.

The core research that forms the basis for the present study focused on two goals.
The ﬁrst was to identify an inception cohort of family members who were new to
caregiving, to follow them for 18 months, and to compare those who continued to
provide care with those who ceased their caregiving role and chose other alternatives to
meet their elderly relative’s needs. The second goal was to describe, over 18 months,
the types of care provided to all patients whose caregivers either ceased or persisted in

family caregiving (Pohl, Given, and Given, 1991).

 

Signiﬁcan

As
increase (1.
job perior:
ﬁnancial b
morivate er
Will persis
liability ar,

“&y

 

lou‘hich ti...
FUpoauor

adaquately ~

 

l 1
Significance and Generalizability

As noted, the number of employed female family caregivers will continue to
increase due to current and projected demographic and social trends. The employees’
job performance, time lost from work, and the ability to remain employed inﬂuence the
ﬁnancial bottom line for employers (Scharlach, 1989). Consequently, the desire to
motivate employees, decrease absenteeism and tardiness, and attract productive workers
will persist in generating employer concern. Increased productivity and ﬁnancial
viability are employers’ desired outcomes.

Warshaw, Barr, Rayman, Schachter, and Lucas (1986) maintain that the ”extent
to which the competing demands of work and family care may actually inhibit labor force
participation or affect either the quality of family care or job performance has not been
adequately explored” (p. 2). Also, Brody, Kleban, Johnsen, Hoffman, and Schoonover
(1987) claim that little research exists regarding patterns of women and work in relation
to eldercare. Most previous research uses the context of childcare (Friedman, 1988).
Kola and Dunkle (1988) also note the lack of adequate research focusing on the
interaction between family and work roles. They, too, maintain that most existing
research about family and work relates to childcare. The present study adds to that body
of knowledge by examining the inﬂuence of acquiring the family caregiver role on female
employment adaptation, both at the inception of the eldercare function and three months
later.

Much previous research involving family caregivers studied caregiver burden
(Archbold, 1983; Barusch & Spaid, 1989; Brody, 1985; Fitting, Rabins, Lucas, &

Eastham, 1986; George & Gwyther, 1986; Gilhooly, 1984; Given, King, Collins, &

12
Given, 1988; Given, Stommel, Collins, King, & Given, 1990; Horowitz, 1985;

Montgomery, 1989; Stommel, Given, & Given, 1990). A gradual shift in research
emphasis from caregiver burden to the inﬂuence of eldercare on employment began about
six years ago. While that body of literature continues to grow and provide valuable
insight into the complicated interaction of work and caregiving, much is still unknown.
Some people choose to stop work in order to provide care; others elect to halt informal
caregiving and institutionalize the elderly relative in order to continue employment. In
between those extreme adaptations, a range of alternative behaviors, such as decreased
work hours, choosing a less demanding job, or arranging to share the care with formal
and informal caregivers, exists. It is important to identify key events that inﬂuence these
decisions.

In contrast to the present study, most previous researchers investigating the
consequences of caregiving studied well-established caregivers and used a cross-sectional
design. However, some longitudinal studies exist and they present a complex picture of
change over time (Given, Stommel, & Lin, 1991; Robinson & Thumher, 1979; Stoller
& Pugliesi, 1989b; Zarit, Todd, & Zarit, 1986). Neal, Chapman, Ingersoll-Dayton,
Emlen, and Boise (1990) urge that future research on employees and caregiving
responsibilities considers variation within and across groups of caregivers as well as how
caregiving demands (and inﬂuence on employment adaptation) change over time. Stone
and Short (1990) reference a growing number of federal, state, and private initiatives that
target beneﬁts to employed caregivers. They, too, recommend the empiric investigation

of the interrelationship with caregiving and employment.

Conceptual F
Maudie.
The cr
general theori
unmet
relational con‘
Adaptation of
FTSPOCtit-e.
ent'ironment i
analytically 5i
biological en.
In the
Changing ne
disequiitbnu,
Eldercare. T
cmplol'lllent
The 2
it“ famineE
(Blister, 19g
“menses fr

other
The family c

13

Conceptual Framework
Way—”meant

The conceptual framework and theoretical base for the study ﬂow from two
general theories; namely, the human ecological and choice and exchange theories. The
use of these theories allows the examination of family caregiving and employment in a
relational context. Caregiving represents a dynamic, not static process of interaction and
adaptation of family members with their environment, a basic tenet of the ecological
perspective. The individual (caregiver) or unit (family) in interaction with the
environment constitutes an ecosystem. The family ecosystem environment includes the
analytically separated, but interrelated human-built, social-culture, and natural physical-
biological environments (Bubolz & Sontag, in press).

In the context of their environments, individuals and families adapt to meet
changing needs. Adaptation allows individuals and families to respond to the
disequilibrium that may occur when the family system is disturbed by the need to provide
eldercare. The primary caregiver endeavors to maintain a balance between home and
employment obligations. The family as a whole responds to the altered situation.

The adaptation (modiﬁcation of behavior, feelings, ideas, and the environment),
that families make to meet changed member demands is a key ecological concept
(Bristor, 1990). A primary need of one person (in the present study, eldercare) triggers
responses from other relatives. The employment adaptation that caregivers may make
in order to satisfy eldercare commitments is the dependent variable for the study.

Other ecological concepts important to this study include the family and resources.

The family can be a resource to the employed caregiver. Matthews and Rosner (1988)

 

report that 5
their parent
of the prin
resources to
tics also ma
the present 1
age, relatio
income. am
of CO’TCSldl

Characterist

Thr
the Priman
are involve
and Three r
Care, Term
of the car e
mi eco

Co:
geographic
family rel
themselves
and flimit

l
Thrill-jam;11

14

report that sister dyads used routine involvement and back-up support in order to meet
their parent care obligations. The responses of the family, either the direct involvement
of the primary caregiver or the family assistance other members provide, denote
resources for the elderly relative needing help. Caregiver and care recipient characteris-
tics also may be resources that inﬂuence the caregiving and employment interaction. In
the present study, selected caregiver/care recipient characteristics include the caregiver’s
age, relationship to the care recipient, marital status, education, occupation, household
income, and employment status; the care recipient’s age and gender; and the dyad feature
of co-residence. Family support, caregiver involvement, and caregiver/care recipient
characteristics are the independent variables in the study.

Throughout the process of eldercare and the probable employment adaptation by
the primary caregiver, other meaningful ecological concepts (time, space, and energy)
are involved. In the present study, the period between the inception of the caregiver role
and three months later represents an actual example of the calendar time spent providing
care. Temporal orientations, such as past experiences, present needs, and future desires
of the caregiver and family system inﬂuenced the decisions (made in the context of the
family ecosystem environment) that led to the acquisition of the eldercare role.

Concurrently, space, such as the residence the caregiver dyad may share or the
geographic distance between members in the family network, impacts eldercare and
family relations. Also, conceptual space, as with family members ”distancing"
themselves from direct care involvement or the bounded, but interrelated systems of work
and family inﬂuence the care situation. Likewise, energy, or the ebb and ﬂow of

individual and family internal and external resources, enables the adaptation and response

to employmen

support may

employment a

Choice and E.

The in
Competing det
discussing the
numbers of ch
(or least loss )
choices regarc
decrease Worl
dim Choices
Additional ele
calegiying 103

Finally
dsisranCe to t

DTlmay-y 93mg

15

to employment and family needs that regulate system equilibrium. For example, family
support may allow the caregiver to recharge personal energy levels and minimize

employment adaptation.

W

The interaction and adaptation families make as they struggle to manage the
competing demands of work and eldercare involve many decisions or choices. When
discussing the choice and exchange theory, Nye (1979) maintains that one makes inﬁnite
numbers of choices in order to decrease costs and maximize rewards for the most proﬁt
(or least loss). He views choice as the most signiﬁcant aspect of the theory. Some
choices regarding caregiving and work are obvious and individual, such as choosing to
decrease work hours in order to provide care. Caregivers also make less obvious or
direct choices, for example, using vacation time to meet eldercare obligations.
Additional elements of preference for the family caregiver may include the amount of
caregiving load assumed and the time allotted for competing work and family roles.

Finally, other family members choose to provide either physical or emotional
assistance to the caregiver or the care recipient or to distance themselves from the care
situation. Thus, decisions regarding eldercare that the family makes may inﬂuence the

primary caregiver’s adaptation to employment.

 

Assimpm
Assumptio.
The follov
l. The
2. The

neer
3. All

TCSC

em-
5. 1:3r

me-

hu;

in

16

Annotations

Assumptions of the Ecological Perspective

The following assumptions are from Bubolz and Sontag (in press).

1.

2.

10.

ll.

12.

The family and its interaction with the environment is considered an ecosystem.*
The family manages the bio-physical, psycho-social, economic, and nurturance
needs and functions of its members.*

All humans are interdependent with one another and with environmental
resources.*

The properties, structure, and processes involved with the family and its
environment must be considered as interdependent and analyzed as a system.
Families are interdependent with other forms of life and the non-living environ-
ment.

Adaptation to their environments is a continuous process in families.

All parts of the environment (natural biological-physical, social—cultural, and
human built) are interrelated and inﬂuence each other.

Families are part of and interact with many environments.

Families are energy transformation systems that use energy for survival,
interaction, and adaptation.

Family interactions are guided by physical and biological laws of nature and
human derived rules.

Environments provide limitations and opportunities for families.

Families have varied amounts of control related to environmental interactions.

13. D1

’=key

Assump

The fol

h.)

«4351”

ft.)

13.

17

Decision-making is the central control process that families use to attain individual

and family goals.

* = key assumptions

Assumptions of Choice and Exchange Theory

The following assumptions are from Nye (1979).

1.

Human behavior is rational in spite of the occasional use of insufﬁcient informa-
tion and incorrect forecasting of the future.

Humans at any level (from individual to nations) act to decrease their costs and
increase their rewards.

Humans are able to expect increased rewards and decreased costs from responsive
governmental or private institutions.

Humans are able to presume general reciprocity between themselves and society.
Humans recognize their choices inﬂuence the rewards and costs of others in the

groups to which they belong.

Assumptions of this Study

1.

The ability of an elderly individual to adapt to increased functional health problems
relates to the amount and kind of family resources available.

Families make choices to aid their elderly relative based on a consideration of
alternatives, costs, and beneﬁts.

Families prefer to help their elderly member live independently and avoid

institutionalization.

4. The.
adapt

i Thes
ande

6. 'The

shua

Lhnhathor

Tin
them to de
ha150 see
the Origin;
is not the
lisearcltt

II
the on'gi‘
“EVE of
We at.
enigma
lite Emm
Unknowl

SimPath

18

4. The degree of caregiver involvement in eldercare inﬂuences employment
adaptation.

5. The size and quality of the caregiver’s family network inﬂuence the care situation
and employment adaptation.

6. The assistance families provide to the primary caregiver inﬂuences the care

situation and employment adaptation.

Limitations

The core study purpose is to identify an inception cohort of caregivers and follow
them to determine differences between those who cease caregiving and those who persist.
It also seeks to describe the type of care provided to all patients regardless of whether
the original caregiver stopped or continued to provide care. Because the core study focus
is not the employment adaptation of female caregivers, it is impossible for the present
research to provide an exhaustive investigation of the caregiving/work interaction.

The use of secondary data limits the variables available for analysis to those from
the original research. In some instances, the same variable is not measured on each
wave of data, thus making comparisons of two time frames difﬁcult or impossible.
While the employment variables for this study are valuable, additional components would
enhance the usefulness of the results. For example, the general work environment and
the employment policies and practices affecting eldercare for caregivers in this study are
unknown. Perhaps some caregivers in the sample are employed by companies

sympathetic to their employees’ family care needs while others are not. Objective

measures of
self-report in

Also.
thus, the ma;
be measured
ployod careg
1990).

As \1
family supp
measures. 1
arbount and
their own e1 ‘
the Elationg‘

Final
to erhillcity,
ushhhl does
late is COnsi.
ofcahegll’er

hhhhlllace a“

19

measures of the actual employment adaptations used (employment records, rather than
self-report information) would increase the value of the present research.

Also, the analysis did not include comparison groups of employed non-caregivers;
thus, the magnitude of the inﬂuence of caregiving on employment adaptation could not
be measured. Systematic research on the differences between employed and nonem-
ployed caregivers needs to be done (Neal, Chapman, Ingersoll-Dayton, Emlen, & Boise,
1990).

As with employment adaptation data, the caregiver self-report responses for
family support or eldercare involvement offered subjective, rather than objective
measures. Individuals vary in their perceptions of reality; thus, their judgments of the
amount and frequency of family help may reﬂect those biases. Similarly, how they rate
their own eldercare involvement may be inﬂuenced by their perception of the quality of
the relationship with the care recipient or by their deﬁnition of the role.

Finally, little research on eldercare and employment considers variables related
to ethnicity. Thus, the analysis of caregiving determinants and labor market decisions
usually does not use an ethnic approach (White-Means & Thornton, 1990). Although
race is considered in the core study, preliminary information indicates the vast majority
of caregiver dyads in the study are white. Ethnicity and its relationship to eldercare and

workplace adaptation should be considered in the future.

Rese
the inﬂuem
followed b

exchange ll

Inﬂuence

Far

researcher

CHAPTER II

Review of Literature

Research and related literature are organized in ﬁve categories. Reviews include
the inﬂuence of eldercare on employees and families, on employers, and on society,
followed by an examination of the human ecological approach and the choice and

exchange theory.

Inﬂuence on Employees and Families

Family caregiving’s effect on employees has stimulated increased attention from
researchers. While studies generally support the notion that caregivers adapt employment
to manage caregiving obligations, the reported amount of inﬂuence varies. Anastas,
Gibeau, and Larson’s (1990) survey of employed caregivers was the ﬁrst of its kind to
look at working families and eldercare with a national rather than a regional perspective.
Although not a large sample (N =409), results did support similar local or regional
surveys. The respondents were rather evenly divided between male and female workers.

In respOnse to questions about work adaptation because of caregiving, the
caregivers in the study reported that they: (a) used vacation time—64% , (b) changed
schedules—33%, (c) used personal leaves—32%, (d) missed work meetings—18%, (e)

missed outside conferences-14%, (i) missed overtime—13%, (g) called in sick—8%,

20

(l1) refused mt
their jobs—59
Anothr
Persons and l
were employ
due to Careg
beneﬁts as a
Other
and Boyd (19
(M1557).
hm only 49
respollsibilr'r
0f the other
al'erage nor
than fOI thc
Fen
emPIOyee S.
m“ ('2
mlaﬁonshjl
dlingerences
“leaving
of Hugged

WDmen Ve

21
(h) refused more responsible jobs—7%, (i) were unable to seek new jobs—6%, (i) left

their jobs—5%, (k) refused job offers-4%, and (1) took leaves of absence—3%.

Another study of caregivers (N=754) conducted by the Association for Retired
Persons and The Travelers Companies Foundation (1988) found that 53% of caregivers
were employed and that 38% of that group either lost time from work or came in late
due to caregiving responsibilities. Twenty percent of the employed caregivers lost
beneﬁts as a result of the employment adaptation that they made.

Other studies support the competing demands of work and caregiving. Scharlach
and Boyd (1989) compared employed caregivers (N =34 1) with non-caregiving employees
(N = 1557). They found that 73.1% of caregivers reported interference with their jobs,
but only 49.1% of the other employees described conﬂicts with work and family
responsibilities. The researchers also learned that 37% of the caregivers versus 23.2%
of the other employees missed work due to family responsibilities. In addition, the
average number of missed work days was one and one half times higher for caregivers
than for those without care responsibility.

Fernandez (1990) reports that eight percent of the people surveyed in his 1988
employee survey (N =26,000) recounted eldercare responsibilities while slightly over two
percent (2.4%) claimed both elder and childcare obligations. Results about the
relationship between caregiving responsibility and lost productive time show gender
differences for caregivers of elderly persons, children, or both, and for those without
caregiving responsibilities. In three of four categories, women had a higher percentage
of missed work time. For example, for those with eldercare responsibilities, 45% of

women versus 28% of men missed work; 33% of the women and 25% of men arrived

late; and 54%
percentages of
= 81%; men
The bi
eldercare 0in
work (womer
left early (wo
Noreen the g
Nor surpnsir
work adjustn
grouping,
Schar
for Either Cg
lmponam v;
missed an a
missed by c
work Chang
cognitively
WW '1.
versus 11%
“I“ 53.19

Comm Erer

22

late; and 54% of the women versus 46% of men left early. However, nearly equal
percentages of men and women reported dealing with family problems at work (women
= 81%; men = 82%).

The biggest gender differences came when caregivers claimed both child and
eldercare obligations. Again, women reported greater disruption, especially with missed
work (women = 55%; men = 28%), arrived late (women = 41%; men = 27%), and
left early (women = 63%; men = 47%). Like with eldercare, there was little difference
between the genders in dealing with care issues at work (women = 87%; men = 83%).
Not surprisingly, those without care responsibility had little gender disparity in each
work adjustment category, although women reported more disruption than men in each
grouping.

Scharlach’s (1989) study of TransAmerica Life employees who were caregivers
for either cognitively or physically impaired elderly persons (N =332) also suggests that
important work-related adaptation occurs. Caregivers of a cognitively impaired person
missed an average of 3.4 hours of work in the previous month compared to 2.4 hours
missed by caretakers of the physically impaired. Also, the percentage of those reporting
work changes in the previous two months was substantially more for the caregiver of the
cognitively impaired. For example, 41.8% of caregivers for the cognitively impaired
reported ”left early" versus 28.8% of other caregivers; "extended a break" was 23.5%

versus 11%; "took a day off (with pay) " was 45.5% versus 28.2%; "used vacation time”
was 53.1% versus 36.8%; "too tired to work" was 42.4% versus 26.3%; and

"considered quitting" was 14.7% versus 5.8%.

I
employr
when th
Roberts

ofthec

in three
Suggest
translat

1933).

for fen
Female
Comte;
non.“
in Poor

EOOd h,

23

Another report involving TransAmerica employees (N =34l) indicates that
employment adaptation was greatest when the care recipient was more impaired and
when the caregiver felt that support for the role was inadequate (Scharlach, Sobel, &
Roberts, 1991). That study emphasizes the importance of family roles and characteristics
of the caregiver and the care recipient in predicting employment outcomes.

Employment outcomes do make a difference! TransAmerica Life found that one
in three TransAmerica employees providing eldercare missed work. Survey results
suggest that Transamerica employees lose more than 1600 work days annually,
translating into a yearly loss of $250,000 in salaries and beneﬁts (”1,600 Days Go,”
1988).

In another study (N =491), Stone and Short (1990) found an increased probability
for female employees to adapt work schedules to meet caregiving responsibilities.
Female caregivers were 11.8 % more likely to adjust their work schedules than their male
counterparts. Also, white caregivers were 15.1% more likely to adjust schedules than
non-white caretakers. Caregiver health complicated the work/ family conﬂict. Caregivers
in poor health were more likely to adjust their work schedules than those who reported
good health.

Stone and Short (1990) also found that the higher the nwd of the elder, the more
likely were caregivers to adjust their work schedules. For example, in their study,
caregivers helping elders with behavioral problems were 18% more likely to alter their
work schedules. Surprisingly, the severity of the care recipient’s activity of daily living
dependencies did not inﬂuence caregiver work accommodation. However, 22-28% of

caregivers worked with some employment adaptation. That percentage contrasts sharply

with the
Survey

respons'r

Chapmz
work p
moons

higher

“8833
than it
“Wk 1
emplo

been

men

24

with the American Association of Retired Persons (1987) Caregivers in the Workplace
Survey (N =322) where only ﬁve percent of caregivers reported their caregiving
responsibilities regularly interfered with work duties.

In a study (N =7801) of absenteeism and stress among employed caregivers, Neal,
Chapman, Ingersoll-Dayton, Emlen, and Boise (1990) conclude that interruptions during
work provide the only measure of time loss obviously associated with caregiving
responsibilities. However, in that study, interruption rates were two to three times
higher for the caregiver employee than for workers without dependent responsibility.
Again, women reported the most difﬁculty with work interruptions and, as expected, full-
time employees had more problems than did the part-time group. The study also
suggests that the personal and home-related aspects of the caregiver’s life suffered more
than work-related components. Orodenker (1990) supports the notion of caregiving and
work conﬂict and the resulting inﬂuence on employment adaptations. She reports that
employed caregivers experienced more stress after they altered their work schedules
because of their inability to balance competing role demands.

The stress resulting from employment and caregiving conﬂicts may lead to the
decision to quit work, especially for women. Data from the Family Survival Project
suggest that caregiving is a stronger deterrent to adult daughters employment than to any
other caregiver group (Enright & Friss, 1987). A later account notes that 22% of
unemployed caregiving daughters in the study of brain-impaired adults quit their position
in order to provide care (Enright, 1991).

Anastas, Gibeau, and Larson (1990) also recount that women are more likely than

men to describe high degrees of work and family conﬂict and to consider ending

 

emplc
pron]
we ;
likel)
hour:
stud}
prov
worn
likel
malr

sens

Inn:

for ‘

“Urn.

25
employment. In their study (N =436), 75% of those who pondered quitting in order to

provide care were women. Personal characteristics of the caregiver and the amount of
care given are important factors in the decision equation. Relinquishing a job is more
likely to be the choice of a woman over 50 who is also providing a large number of care
hours per week (Brody, Kleban, Johnsen, Hoffman, & Schoonover, 1987). In their
study of adult caregiving daughters (N = 150), 45% of the women who quit work
provided 40 hours of care per week for their dependent mothers. These also were
women with the lowest education, income, and occupational status and who were more
likely to have a "job" versus a "career." Ferree (1987) supports that ﬁnding and
maintains that "working-class women are not committed to careers in the middle-class
sense, and they are likely to change jobs, refuse promotions, and restrict hours in order
to carry out their domestic responsibilities” (p. 292).

Finally, preliminary results from the analysis of data from the present study
sample showed that caregivers made some employment adaptation in order to provide
care (Franklin, 1991). At Time 1, their involvement with family eldercare was
signiﬁcantly correlated with missing work days without pay, decreasing total work hours,
taking a leave of absence, or quitting work. However, there were no signiﬁcant

correlations with adaptation in employment hours (arriving late/leaving early).

Influence on the Employer
Employee caregiving responsibilities inﬂuence the workplace in differing ways
for employers. Key costs to employers include productivity losses due to increased

numbers of personal phone calls made or received by employees and more time off due

 

to illness. '.
the employe
e1dercare('
valued emp‘

Corr
intense cor.
implication
to attract a
inﬂuence c
beneﬁts (N
have recog
Travelers]
Lowe, & S
1983); Ae
(Femande;

The
[0 Urese a
Schneideu
per year fo;
ml for Cac

Cos
mlSSed 0r ‘

26

to illness. The elevated employer health beneﬁt expenses stemming from a decline in
the employee’s health that began with caregiving responsibilities increase the cost of
eldercare ("Eldercare Beneﬁts," 1988). Fernandez (1990) adds the ultimate loss of a
valued employee to the employer’s costs.

Companies with a large, middle-aged, female work force and that also experience
intense competition for certain job skills often become acutely aware of eldercare
implications. Without accommodation to family nwds, some companies lose their ability
to attract and retain workers (Friedman & Gray, 1989). Employers who see the
inﬂuence of family caregiving on productivity and quality tend to support eldercare
beneﬁts (Neal, Chapman, Ingersoll-Dayton, Emlen, & Boise; 1990). Organizations that
have recognized and responded to eldercare’s signiﬁcance for their company include
Travelers Insurance Companies, Bank of America, IBM, and Phillip Morris (Scharlach,
Lowe, & Schneider, 1991); Southwestern Bell and Remington Products (Kola & Dunkle,
1988); Aetna Life and Casualty, Marriott (Buglass, 1989); and Chase Manhattan
(Fernandez, 1990).

The desire for increased productivity and cost savings provides a strong impetus
to these and other companies concerned about eldercare. Scharlach, Lowe, and
Schneider (1991) estimate that a company without formal eldercare programs loses $2500
per year for each employee with eldercare responsibilities. Merck and Company estimate
that for each dollar spent on family care, it gets three returned (Fernandez, 1990).

Costs to the employer may be directly related to the employed caregiver and
missed or poor quality work or indirectly associated by the low morale that may

ultimately permeate employees who must ﬁll the gap left by another worker’s absence

in poor p
relationshij
mmﬂeon
annoyed a
anoneenr
generate 1:
end as pe
billion do]
work due
Replacem.
SOme ofn

bUdget
In
employer:

27

or poor performance. While an element of common sense exists regarding the
relationship of predictable attendance and productivity, clear outcomes of stress and
morale on productivity are more obscure (Friedman, 1988). A co-worker may be
annoyed at having to work harder to cover a caregiver’s absence, even though the
absenteeism is explained. Employee peer resentment and unhappiness may result and
generate less total productivity. Also, caregiver absences that begin as temporary may
end as permanent, thus creating additional employer costs. It is estimated that 100
billion dollars are lost annually when trained and experienced workers are absent from
work due to their own or family illness (Bureau of National Affairs, #21, 1989).
Replacement of workers, whether for a temporary need or a permanent change, creates
some of those costs. Orientation and training expenses profoundly affect an employer’s
budget.

In spite of costs stemming from employees with eldercare responsibilities, some
employers do not see a problem. A survey of 96 New York companies indicated that
over 50% of the companies did not see employee caregivers as presenting work-related
problems (Warshaw, Barr, Rayman, Schachter, & Lucas, 1986). However, the
researchers ultimately concluded those issues may be hidden and, instead, more
acceptable problems brought to the workplace.

A similar exploratory study by Kola and Dunkle (1988) found that more than half
of the surveyed organizations indicated that caregivers’ needs seldom came to the
workplace. However, the remaining respondents indicated that many of their employees
needed time off for eldercare that was manifested in lateness, absenteeism, and

emergency hours and unscheduled time off. The same employment issues also were

 

 

 

nnpo:

inthe

atnor

ofhu

Conrr
Care}
famn
olrh
is an
the e
“arrn

‘0 UM

Liven

28

important for the companies in the New York study who did see eldercare as a problem
in the workplace (Warshaw, Barr, Rayman, Schlachter, & Lucas, 1986).

Even businesses that recognize eldercare’s importance have done little to take
action. A survey of 101 companies initiated by Personnel Journal indicated that 67%
of human resource executives felt that eldercare should be considered, but that only about
10% actively studied the problem (Magnus, 1988). More signiﬁcant, the vast majority
(89%) of the executives said they have no plans to address the issue. Only three of the

101 companies have a beneﬁt that supports eldercare.

Influence on Society

Jacobs (1987, p. 18) deﬁnes social cost as the ". ..measure of the total resource
commitment made by all members of the society in the undertaking of any activity. "
Caregiver employment adaptation inﬂuences personal resources available, as well as
family, co-worker, employer, and societal reactions. Graaff (1990) advises clariﬁcation
of the scope of society when discussing social cost. For example, in this study, society
is the United States, not the world community, although as ecological theory suggests,
the effects of the interactions of family caregiving and employment cannot be conﬁned
to this country. Diverse repercussions affect multiple levels, from the micro (individual)
to the macro (world) and the reverse.

The costs of employed caregivers on society are often hidden. Some burdens of
working caregivers, such as the inﬂuence of caregiving on health, schedules, and esteem,
can affect attendance and productivity. If employment adaptation necessary to provide

family care leads to lost wages, a decrease in buying power and tax contribution results.

 

Also, a
health 1
stemmi
also suf
Short (
gross r
suppor-
beneﬁt

COHSllll

29

Also, a possible increase in public assistance (more caregiver unemployment) and higher
health care costs (increased caregiver morbidity) may occur (Horowitz, 1985). Costs
stemming from the morbidity of the caregiver may escalate, particularly if the caregiver
also suffers from chronic ill health that worsens with the strain of caregiving. Stone and
Short (1990) support Horowitz’ ﬁndings. They regard societal costs as a decrease in the
gross national product, foregone income tax revenue, and increased expenditures to
support caregivers with decreased incomes, fringe beneﬁts, and future retirement
beneﬁts. These costs ultimately result in a higher price of goods and services for the
consumer.

Of course, the decreased expense of informal care provided by families versus the
price of formal care available in institutions somewhat offsets this cost. A study by Hu,
Huang, and Cartwright (1986) indicates average annual nursing home charges to be
$22,458, while the yearly homecare cost mean was $11,735. Feldstein (1988) estimates
the average annual nursing home expenditure to be $27,000. Either institutional cost
ﬁgure suggests that it is cheaper to provide care at home. Limited private coverage for
nursing home care adds to the cost for society. Government, usually in the form of
Medicaid, covers the expense when families become impoverished. When government

pays, society pays in the form of higher taxes.

Conceptual Framework
1 i A r a
Human ecology theory assumes that phenomena must be examined in the

wholeness of interaction and interdependence rather than with simple cause-and-effect

nhnm
mane
study,
be fai.

rxnne

ulnn
This 11
used
intern
Ship e
cowc
”16 ex
Wmn
men
elihEr

nmnr
hUma
Dim-*1

em")!

510k)

30
relationships (Andrews, Bubolz, & Paolucci, 1980). This allows analysis and

interventions at several levels, depending on the researcher’s area of interest. In this
study, the interaction of work and family roles provides the focus. This interaction may
be fairly simple, with minimal repercussions, or more complex, with effects that
permeate through multiple levels and environments.

Bronfenbrenner (1979) proposed a nested arrangement of concentric environmen-
tal structures; the microsystem, the mesosystem, the exosystem, and the macrosystem.
This idea clariﬁes the importance of work and family interactions. While the employed
caregiver functions at the micro level within either the home or work environments, the
interrelationship between work and family comprises the mesosystem. This interrelation-
ship extends to affect events in the work setting that, in turn, inﬂuence the reactions of
co-workers. This indirect result, involving co-workers (peers or supervisors) illustrates
the exosystem. Finally, the macro or largest system includes society as a whole. Thus,
what occurs at the micro level ultimately extends to involve the macrosystem. The
reverse of this process also is true. For example, public policy related to the support of
either the family caregiver or the frail elderly person changes the care situation and its
ramiﬁcations.

The environments of a family ecosystem (the natural biological-physical, the
human-built, and the social-cultural) as described by Bubolz and Sontag (in press) also
provide the context for the varied interactions that families experience. These
environments are analytically separate, but interrelated and impact each other throughout
family caregiver and employment interaction and adaptation. Examples of the natural

biological-physical environment include the climate, atmosphere, soil, trees, or what is

commonly
made to t1
goals or n
houses (w
hospitals (
01 equipr
presence

ramiﬁcatr'
lllSllllltlor
commum

“1 all em

31

commonly called ”nature," while the human-bth environment involves modiﬁcations
made to the natural setting to enable survival (food or shelter) or to meet other human
goals or nwds (family involvement, employment). These human alterations might be
houses (with adjusted living quarters to meet eldercare needs), factories, businesses, or
hospitals (to meet employment nwds), or material objects (household goods, belongings,
or equipment). The third environmental component, social-cultural, includes the
presence of humans and their interactions (families, friends, co-workers), cultural
ramiﬁcations (societal norms, family values and support), social and economic
institutions (regulatory systems and employment policies), and provides the basis for
communication, order, and moral rules. Bubolz and Sontag (in press) assert that quality
in all environments is necessary to maintain life.

Employed female family caregivers make many decisions to maintain their lives
and that of their elderly family member. These decisions, involving employment and
eldercare responsibilities plus other personal and family obligations, are made in the
context of the family ecosystem environment. According to Bristor (1990), the decision-
making process used by humans begins with inputs (resources and information),
continues through transactions (information processing), and ends with outputs (decisions,
judgments, and actions). Characteristics of the interrelated family ecosystem environ-
ments affect those choices or outcomes about employment and family.

Kantor and Lehr’s (1975) dimensions of family process also contribute to the
understanding of eldercare decisions. They (p. 36) describe dimensions as being
"physical and conceptual ﬁelds of interactional activity" and refer to access dimensions

of time, space, and energy, and target dimensions of affect, power, and meaning. These

 

 

 

dimension
for specif
interfaces
lhese inn

A:
includesl
of elderc
has for r
individua
example.

£13111 de‘

ProceSs

32

dimensions provide a context for decision-making through which family members’ needs
for speciﬁc goals are accomplished. The access and target dimensions have many
interfaces and thus have a multitude of combinations that expand their range and power.
These interfaces contribute to both the well-being and the stress of the family.

As with Kantor and Lehr’s access and target dimensions, family eldercare
includes both quantitative and qualitative components. Examples are the amount and type
of eldercare actually provided (quantitative) and the meaning that caring for one’s own
has for the individual members involved with that care (qualitative). The power that
individuals exert in decision-making involving eldercare offers another qualitative
example. While the present study cannot address qualitative aspects of eldercare to any
great degree, the quantitative or physical aspects of caring offer much opportunity for
analyses related to caregiving’s inﬂuence on employment adaptation.

The inﬂuence of eldercare on employment adaptation is likely to vary due to the
changing needs of the elderly person and altered family circumstances. Thus, the
primary caregiver and the family make many decisions to address those issues. As in
other events or developmental changes that occur during the family life cycle, the family
system experiences periods of equilibrium and disequilibrium with eldercare. In time,
most caregivers and their families adjust personal and work roles, but the eventual
symmetry or equilibrium shifts with the care recipient’s next period of change. The
adaptation process provides needed stabilization for the family and the caregiver with
multiple employment and family obligations.

Clearly, the norm of family care in society (an output of the decision-making

process) involves many changing decisions, circumstances, and interactions over the life

oourse.
sustena
rnemh<
and er

caregi

halant
cultur
from
with
the it.
those

emp]

Only

Cmp

Gide

Inee

mm

Emp

33

course. Bubolz and Sontag (in press) state that "a family carries out physical-biological
sustenance, economic maintenance, and psychosocial and nurturance functions for its
members. " In order to manage family eldercare responsibilities, decisions about choices
and exchanges permeate family relationships and interactions and inﬂuence the primary
caregiver’s employment adaptation.

Employment adaptation, stemming from caregiver decisions that attempt to
balance work and family obligations, varies with the corporate work site culture. A
culture sympathetic to caregiving concerns provides a very different work environment
from one where family problems must be "kept at home” and not allowed to interfere
with employment responsibilities. However, regardless of the workplace environment,
the interactive work and caregiving situations cannot be separated. Ripple effects of
those interactions permeate all levels: individual employees and their family members,
employers and co—workers, and society.

The employee’s adaptation to competing family and work roles may inﬂuence not
only the quality of care provided to the elderly relative and relationships within the
family as a whole, but also work quality, productivity, and the ability to remain
employed. Likewise, the involvement of the caregiver and family assistance in meeting
eldercare obligations may inﬂuence how the caregiver ultimately adapts employment to
meet care demands.

As ecological theory suggests, the consequences or outcomes of caregiver
decisions involve multiple levels. In addition to individual, family, and employer
inﬂuences, society ultimately suffers from an extreme employment adaptation, the

employee decision to quit work. The repercussions and ramiﬁcations, although not

immediate
purchasing
of public
employed
possible c

Bl
0f multi¥
present st
inﬂuence
the role t]
0f family
flamewo]
the Social
f“Inland

”Scythe

34

immediately apparent, may be declines in the gross national product (decreased
purchasing power), foregone income tax revenue (decreased income), and increased use
of public assistance programs (inadequate income to meet nwds). As the number of
employed family caregivers increases because of projected demographic changes, these
possible outcomes gain signiﬁcance for society.

Bubolz and Sontag (in press) maintain that the ecoperspective allows examination
of multi-level functions and systems in relation to each other and over time. The
present study acknowledges the dynamic nature of caregiving and seeks to determine the
inﬂuence that eldercare responsibility has on employment adaptation from inception of
the role through a three-month period. The ecoperspective permits the interactive nature
of family processes and relationships to be captured and studied. Also, the ecological
framework provides a more comprehensive approach to analyzing human behavior than
the social systems model (Bubolz, Eicher, & Sontag, 1979). The social systems model
frequently minimizes interrelationships between natural and social systems. This

researcher recognizes the strong interaction between work and family.

Ex h Th 0
The choice and exchange theory is versatile in that its concepts can be applied to
a wide variety of situations ranging from one-to-one encounters to group interactions
(Nye, 1979; Rook, 1987). In this study, the caregiver and employment adaptation
represented individual choices and exchanges, while the caregiver assistance from the
family denoted group interaction and involvement related to eldercare. The exchanges

may be speciﬁc, as when limited to a few individuals, or very diverse, as those that

llll’Oll
provir
relatit
qualit

work

PICPC
u'me.
child
perio
influc

l0 mg

35
involve individuals and society (Nye, 1979). Rook (1987) suggests that the theory

provides the potential to analyze not only quantitative, but qualitative aspects of
relationships. The present study offers a quantitative approach, but also includes
qualitative potential by the examination of open-ended responses about other effects on
work that may occur due to eldercare responsibility.

Exchanges with parents and children vary over the life course. From a
preponderance of parental ”giving" to very young children, the pendulum shifts over
time. At the end of the parental life cycle, the primary giving may switch to the adult
child as the eldercare role evolves. These " giving" behaviors and activities are usually
performed for positive reasons; however, a negative perspective (or least cost) may also
inﬂuence the interaction. For example, adult children may provide parent care in order
to meet societal norms and avoid censure from others.

Eldercare by an employed caregiver involves decisions that inﬂuence not only the
care situation, but also the workplace. When the employee chooses to arrive late, leave
early, or be absent for extended periods, the decision impacts other employees left to do
the job. Co-workers who pick up the slack may resent that necessity, even when they
are aware of the reason. When employees elect to reduce hours or quit work, they gain
more time to provide care, but sacriﬁce employment beneﬁts, the opportunity for
advancement, and the socialization and satisfaction employment can provide.

Employment and other roles (unless carried to an extreme) may enhance caregiver
well-being by buffering the stresses stemming from the care situation, linking external
resources, and increasing caregiver feelings of self-conﬁdence and worth (Stoller &

Pugliesi, 1989a). Quitting work may increase the caregiver’s problems. Further, when

family
have g;
in orde
elderca
(‘0um
have 3
Chang.
elderc;
mal’ be
Dim-id.

deClSlC

 

 

36

caregivers reduce work hours, take extended leaves, or quit employment in order to
provide care, management faces recruitment and training costs for new workers hired to

ﬁll the void.

Summary

During the past six years, studies began to address the competing demands of
family eldercare and employment. Researchers, through cross-sectional investigations,
have garnered impressive data about the employment adaptation many caregivers make
in order to meet their family obligations. While some employers are oblivious to the
eldercare impact on their employees and the resulting inﬂuence that it may have on the
company’s bottom line, that trend is changing. Companies that take a proactive approach
have assessed their work force and altered policies and procedures to address the
changing needs of employees with eldercare responsibilities. While the impact of the
eldercare phenomenon is less clear for society than for the workplace, its ramiﬁcations
may be projected. The use of the human ecological and the choice and exchange theories
provides a versatile framework to study eldercare and the workplace and related

decisions, interactions, and adaptation over time.

mEnt.

land [

CHAPTER III

Methods

The study explored the inﬂuence of acquiring the family eldercare role on the
employment adaptation of women. The data utilized were from a longitudinal caregiver
study at Michigan State University, ”Caregiver Responses to Managing Elderly Patients
at Home” (NIA #2 R01 AGO6584), referred to in this paper as the core study. This
chapter includes discussion of the conceptual model, the research design, objectives,
questions, hypotheses, and variables for the study. The instrumentation, sampling
procedure (core study and this research), data collection techniques, and the methods of
data analysis used conclude this section.

Figure 1 displays a preliminary conceptual model for family eldercare and its
inﬂuence on employment adaptation over a three-month time period. In this model,
selected caregiver and care recipient characteristics, family support, and caregiver
involvement inﬂuence the employment adaptations that female family caregivers make.
These employment adaptations, in turn, inﬂuence family support and caregiver involve-
ment. While family caregiving and decisions made related to eldercare inﬂuence society

(and the reverse), that effect is not included in this model.

37

‘- a r-- —

v-‘fJ‘

 

38

30:833.. EoE>oEEm uca Sacco—em 2:5“. .3. .235. 3:33:60 EaSE=oE .p 2:2“.

 

    

          
    

:. __ _
mcozmamu<
EmE>oEEm

:oaazm
2E8

 

 

E __ _
wcozmamg
EmESEEw

 

a 9:5 2:85 8::
: mg: 8385

O
D

    

EmEm>_o>c_
520050

82256920
Emﬁamm Emu ..mZmeQ

 

 

 

 

v EmEm>_o>c_
$2960

 

 

 

 

 

all!

(Cr

in

he

pt

39
The study involved three blocks of independent variables. The ﬁrst, caregiver

and care recipient characteristics, included caregiver age, relationship to the care
recipient, marital status, education, occupation, household income, employment status
(current full-time or part-time) co-residence status (now and before caregiving), and the
care recipient gender and age. The second, family support, encompassed the size of the
family network and the amount and frequency of family assistance provided to either the
caregiver or the care recipient. Caregiver involvement, the third block, included the total
help provided with activities of daily living (ADL), instrumental activities of daily living
(IADL), and health care activities (HCA), and also the time demands (in hours) of
physical care and supervision of the elderly relative.

Three types of employment adaptation comprised the dependent variables. Type
I employment adaptation was deﬁned as maintaining, but altering one’s work role.
Possible modiﬁcations included: (a) arriving late/leaving early, (b) missing work without
pay, (c) taking sick/personal days, (d) changing work hours, (e) refusing jobs or
promotions, (f) not looking for jobs or better jobs, and (g) other ways that caregiving
was affecting work. Type II adaptation was interpreted as taking a leave of absence, and
Type III adaptation meant quitting or retiring from employment in order to provide care.

Figure 2 displays the independent and dependent variables for the study.

40

333:5 .coaconoa use Essence—E. .« 2:2“.

 

 

co__m.amu<
z. 3:

co__m_amo< _
:_ mat; _

lllllllr.

 

__ we:

co._m_amo<

 

_ mg;

3998,“. emoumo a 9:6 8.8.2 85.» O
EzomEo n 00 1 I _

 

 

2 9:5 8.68:." r. 1
8.2285 .0 9:0:

 

 

 

.cmEm>_0>c_
Ezaemo

 

 

_l 9.8 .mann. .o everl—

_cme>_0>c_ 00 .93

_ QEI 2.83 5:03de _
. Baa:
_l 99.. 3.8mm anE< 1— l _>=Emmm

_l {0362 2.83 L

j 880 5 L
1 m2 mo L

858m w .5on
8:85me .00

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cat”:
3.9m EmESEEw 00
088: l 8.6.6.0920
_l _ 00 ll— .cmaﬁom mao

 

1 8.8380 8 L . $28.8
—l c2838 8 l_
_l 3292.5: 8 l—
_l 958.66,. 00 l—
7 mo< 8 L l

Hagga—

 

 

 

 

 

 

 

 

rol

of

r.)

Re

41

Research Obj ectives

The overall purpose of the study was to investigate the inﬂuence of acquiring the

role of caregiver for a physically disabled elderly relative on the employment adaptation

of women. Unless speciﬁed otherwise, the following speciﬁc objectives encompassed

each type of adaptation.

1.

To determine the immediate and delayed (after three months) employment adaptation
for the female employee.

To determine the inﬂuence of selected caregiver and care recipient characteristics
(caregiver age, relationship to care recipient, marital status, education, occupation,
household income, employment status, co-residence status, care recipient age and
gender) on caregiver employment adaptation at Time 1 and Time 2.

To determine the inﬂuence of family support (number in network, amount of help,
and frequency of family assistance with eldercare) on caregiver employment
adaptation at Time 1 and Time 2.

To determine the inﬂuence of caregiver involvement (assistance with ADL, IADL,
and HCA and the hours of both physical care and supervision) on employment
adaptation at Time 1 and Time 2.

To assess the probability that the caregiver displays Type II or Type III adaptation

at Time 2 given the level of Type 1 adaptation at Time 1.

Research Design

This research, using a longitudinal design (panel study), examined how selected

caregiver and care recipient characteristics, family support, and caregiver involvement

 

inf’.
Shut
the

1198

oft

and

ma}

holi
rela-
melt
llll-‘o
COnc‘

in M

42

inﬂuenced employment adaptation. Currently, cross-sectional designs predominate in the
study of caregiving’s inﬂuence on employment. This makes it impossible to represent
the varied decisions that caregivers make over time in response to changing care recipient
needs. Longitudinal studies provide a mechanism to determine critical periods when a
caregiver adapts employment in order to provide care. Of course, the reverse, to stop
caregiving in order to maintain employment, also may result. The longitudinal approach
of the present study enhanced the understanding of the dynamics and interaction of work
and family care.

Care recipient, caregiver, and family needs change and result in adaptation by the
caregiver in order to meet the new care requirements. Employment responsibilities also
may shift, thus creating the environment for additional adaptation by the employed family
caregiver. The choice of a longitudinal design considered the ecological concept of
adaptation and acknowledged that caregiving is a dynamic, not a static process. The
study encompassed two waves of data collection; inception of the eldercare role (Time
1) and at three months (Time 2).

Human ecology theory mandates that humans and their environments be viewed
holistically. While it is impossible for any research to examine all possible variables
related to a situation, a study can deal with multiple issues. In the present study, these
include selected caregiver and care recipient characteristics, family support, caregiver
involvement, and possible employment adaptations. The study was non-experimental and
conducted in the natural setting. Female employed family caregivers (N =236), living

in Michigan, comprised the units studied.

Res

prir

43

Research Questions and Hypotheses

All research questions and hypotheses referred to the employed female who had

primary family caregiving responsibility for a physically disabled elderly relative.

1.

Does employment adaptation occur, and if so, does the adaptation type differ
between Time 1 (inception) and Time 2 (three months)?

a. What Type 1 adaptation (maintain, but alter work) occurs?
arrive late/leave early
miss work without pay
take sick/personal days
change work hours
refuse job/promotion
kept from job hunt/better job
other work effects
b. What Type II adaptation (leave of absence) occurs?
c. What Type III adaptation (quit/retire) occurs?

Hla: A variety of Type I employment adaptations occur at both Time 1 and
Time 2.

Hlb: There are fewer Type I adaptive behaviors at Time 1 than at Time 2.

ch: There is less Type II adaptation at Time 1 than at Time 2.

Hld: There is no difference between Type III adaptation at Time 1 and Type III

adaptation at Time 2.

2.

44

What is the probability that Time 1, Type I employment adaptation (maintain,
but alter work) influences Time 2, Type 11 (leave of absence) and Type III
(quit/retire) employment adaptations?

H2a: The use of Time 1, Type I adaptation inﬂuences the use of Time 2, Type

II and Type III adaptations.

What is the relationship of selected caregiver and care recipient characteristics
to each type of employment adaptation at Time 1 and Time 2?

caregiver age

relationship to the care recipient

caregiver marital status

caregiver education

caregiver occupation

caregiver household income

employment status (current part-time or full-time)
co-residence status (now and before)

care recipient age

care recipient gender

'-‘"-"':-="!° ﬁne? 9'?

H3a: Caregiver age relates to each type of employment adaptation.

H3b: Employment adaptation differs for spouse and daughter/daughter-in-law
caregivers.

H3c: Employment adaptation differs by relationship to care recipient.

H3d: Caregiver education relates to employment adaptation.

H3c: Employment adaptation differs with caregiver occupation.

H3f: Caregiver household income relates to employment adaptation.

H3g: Employment adaptation differs with full—time and part-time employment

status.

45

H3h: Employment adaptation differs if caregiver and care recipient reside
together.
H3i: Care recipient age relates to employment adaptation.

H3j: Care recipient gender makes no difference in employment adaptation.

How much variation in each type of employment adaptation at Time 1 and

Time 2 do selected significant caregiver and care recipient characteristics

explain?

a caregiver age

b relationship to the care recipient

c caregiver education

(1. caregiver household income

e co-residence status (now)

f care recipient gender

H4a: Caregiver age explains more variation in employment adaptation at Time
1 and Time 2 than do other caregiver/care recipient characteristics.

H4b: Care recipient age and care recipient gender explain little variation in

employment adaptation at Time 1 and Time 2.

Was there a reported change in the amount of help from family members
between Time 1 and Time 2?

H5a: The majority of caregivers will report that the level of family help

remained "about the same" at Time 2.

6.

46

What is the relationship of family support to each type of employment

adaptation at Time 1 and Time 2?

a. number in network

b. amount of help provided

c. frequency of assistance given

H6a: The number in the family network has no relationship to employment
adaptation at Time 1 and Time 2.

H6b: The amount of help provided by the family relates to employment
adaptation at both Time 1 and Time 2.

H6c: The frequency of family help relates to employment adaptation at both

Time 1 and Time 2.

How much variation in each type of employment adaptation at Time 1 and
Time 2 does family support explain?

a. number in network

b. amount of help provided

c. frequency of assistance given

H7a: The amount of help and frequency of assistance provided explains more

variation in employment adaptation at both Time 1 and Time 2 than does

the number in the network.

Do caregiving responsibilities disrupt normal work and daily activities at both
time periods?
H8a Caregiving responsibilities disrupt normal work and daily activities at both

time periods.

9.

10.

ll.

47

Does the level of reported caregiver involvement change from Time 1 to Time
2?
H9a: The majority of caregivers will report that they provide ”about the same"

amount of care compared to three months ago.

What is the relationship of caregiver involvement (activities and time) to each

type of employment adaptation at Time 1 and Time 2?

a. total involvement (ADL, IADL, and HCA)

b. hours of physical care

c. hours of supervision

H10a: Total involvement (ADL, IADL, and HCA) relates to employment
adaptation.

H10b: Hours of physical care relate to employment adaptation.

H10c: Hours of supervision relate to employment adaptation.

How much variation in each type of employment adaptation at Time 1 and
Time 2 is explained by caregiver involvement (activities and time)?

a. total involvement (ADL, IADL, and HCA)

b. hours of physical care

c. hours of supervision

H1 la: Hours of physical care explain more variation in employment adaptation

than total involvement or hours of supervision at both Time 1 and Time 2.

 

 

48
12. Which identified key variables from questions 4, 7, and 11 are the best

predictors of each type of employment adaptation at Time 1 and Time 2?

(TypeI = caregiver age, relationship to care recipient, caregiver gender, amount

of family help, hours of physical care; Type II = caregiver education, co-

residence now, amount of family help, hours of physical care; Type 111 =
caregiver household income, family network, frequency of family help, total
caregiver involvement, hours of physical care)

H12a: The hours of physical care are the best predictors of each type of
employment adaptation at Time 1 and Time 2.

H12b: The amount of family help at both Time 1 and Time 2 is less effective in
predicting employment adaptation than are the other variables in the
equation.

H12c: Caregiver and care recipient characteristics explain more variation in

employment adaptation than does caregiver involvement.

Research Variables

This section contains the conceptual and operational deﬁnitions of the dependent
and independent variables. Unless otherwise stated, the deﬁnitions applied to both Time
1 and Time 2. Questions for Time 1 alluded to the previous month, while Time 2
questions referred to the prior three months. The portions of the instrument used for
both the study and the operational deﬁnitions are in Appendix A. In addition to speciﬁc
work adaptations noted at both Time 1 and Time 2, caregivers were asked at Time 2 if

their employment status had changed in the past three months and if so, how it had

 

changed

the type

Depender
EWID

Arrive L;
hours as ;
01' leave e

bliss “’0!-

Co

to DTOVide
Work Witht

Take Siclg;

COn

PETSOnal da;

ope:
days or take

no). At both

49
changed (open-ended response). The percentage of those indicating a change and also

the type of alteration was determined by their responses.

Winks
Type I Employment Adaptation (maintain, but alter work)
Arrive Late/ Leave Early
Conceptual deﬁnition: Whether the caregiver worked the entire shift or worked
hours as assigned.
Operational deﬁnition: The caregiver indicated if caregiving caused her to be late

or leave early (yes or no response; 1 = yes, 2 = no).

Miss Work without Pay

Conceptual deﬁnition: Whether the caregiver took time off without pay in order
to provide care.

Operational deﬁnition: The caregiver indicated if caregiving caused her to miss

work without pay (yes or no response; 1 = yes, 2 = no).

Take Sick/Personal Days

Conceptual deﬁnition: Whether the caregiver used either paid sick time or
personal days off in order to provide care.

Operational deﬁnition: The caregiver indicated if caregiving caused her to use sick
days or take personal days in order to provide care (yes or no response; 1 = yes; 2 =

no). At both Time 1 and Time 2, caregivers indicated the speciﬁc number of work days

 

 

 

lost (withir

numbers, I

Change W
Co

could mar

Op

change he
RBfUSC Jot

Co
PrOmotiOn

Kept frOm
Cor.-

liCl' job l:

looking for

50

lost (within the past three months) because of caregiving responsibilities. From these

numbers, the mean and standard deviation for each time period was determined.

Change Work Hours

Conceptual deﬁnition: Whether the caregiver changed her work schedule so she
could manage her caregiving tasks.

Operational deﬁnition: The caregiver indicated whether caregiving caused her to

change her work hours (yes or no response; 1 = yes, 2 = no).

Refuse Job or Promotion

Conceptual deﬁnition: Whether or not the caregiver refused a job offer or a
promotion so that she could manage her caregiving tasks.

Operational deﬁnition: The caregiver indicated whether caregiving caused her to

turn down a job or promotion (yes or no response; 1 = yes, 2 = no).

Kept from Job Hunt/Better Job

Conceptual deﬁnition: Whether the caregiver was unable to look for a job or a
better job because of caregiving responsibility.

Operational deﬁnition: The caregiver indicated whether caregiving kept her from

looking for a job or a better job (yes or no response; 1 = yes, 2 = no).

 

5 1
Other Work Effects

Conceptual deﬁnition: Whether the caregiver experienced any additional, ”other"
effects related to work not speciﬁcally deﬁned as Type I adaptations.

Operational deﬁnition: The caregiver indicated whether caregiving affected her
work in ways not mentioned (yes or no response; 1 = yes, 2 = no). If the response was
yes, the caregiver described how caregiving affected her work. The results of this item
and the preceding work alteration items were summed to create an index of Type I
employment adaptation (range from 0-7). The index of Type I employment adaptation
was then used to obtain a mean score for the sample. This provided a summary measure
of the Type I employment adaptation that caregivers experienced. A high score indicated
more caregiver Type I adaptation to employment than did a low score.

The open-ended responses generated with "other work effects" were grouped
according to patterns of similarity and provided a qualitative supplement to the
quantitative information about employment adaptation. Groupings from Time 1 data
included: 2 = no, 3 = emotional stress, 4 = cannot do other things, 5 = cannot
concentrate, 6 = schedule, and 7 = other. Response patterns for Time 2 varied slightly
and comprised: 2 = no, 3 = more tired, 4 = more stressed, 5 = worry causes lack of
concentration, 6 = quality of work suffers, 7 = turned down better job, 8 = loss of
beneﬁts, and 9 = other ("1 =" was not used for either Time 1 or Time 2). The
percentage of caregivers who responded yes to ”other work effects" was subdivided into
these descriptive categories.

At both Time 1 and Time 2, caregivers were asked to indicate how disruptive

caregiving was to their work and daily activities. The response choices included:

52
1 = not at all, 2 = a little, and 3 = somewhat, and 4 = a great deal. The percentage

of caregivers who responded to each choice was determined.

Type II Employment Adaptation (leave of absence)

Conceptual deﬁnition: Whether or not the caregiver took extended time off from
work to provide care.

Operational deﬁnition: The caregiver indicated whether caregiving caused her to
take a leave of absence from her job (yes or no response; 1 = yes, 2 = no). The

percentage of caregivers who answered yes was obtained from the responses.

Yype III Employment Adaptation (quit/reti re)

Conceptual deﬁnition: Whether the caregiver quit or retired from her job so that
she could provide care.

Operational deﬁnition: The caregiver indicated whether caregiving had caused her
to quit or retire from her job (yes or no response; 1 = yes, 2 = no). The percentage

of caregivers who answered yes was obtained from the responses.

Wines
Caregiver Relationship to Care Recipient
Conceptual deﬁnition: How the caregiver was related to the care recipient.
Operational deﬁnition: The caregiver indicated her relationship to the recipient.
Choices in the original questionnaire included: 1 = spouse, 2 = parent, 3 = daughter,

4 = son, 5 = daughter-in-law, 6 = son-in-law, 7 = sister, 8 = brother,

recode

Careg

Care é

[hep

C0 rel

Chm

Sch0

Fibs,

53
9 = sister-in-law, 10 = brother-in—law, 11 = granddaughter, 12 = grandson,

13 = niece, 14 = nephew, 15 = aunt, 16 = uncle, 17 = other relative,
18 = friend/companion, or 19 = other non-relative to the care recipient. These were
recoded for the present study so that 1 = spouse, 2 = daughter/daughter—in-law, and 3

= other.

Caregiver Age
Conceptual deﬁnition: The number of years the caregiver had been alive.

Operational deﬁnition: The caregiver indicated the date of her birth.

Caregiver Marital Status

Conceptual deﬁnition: The caregiver’s current marital status.

Operational deﬁnition: The caregiver indicated whether she was: 1 = single,
2 = married, 3 = divorced, 4 = widowed, or 5 = separated. These were recoded for

the present study so that 1 = married and 2 = unmarried.

Caregiver Education
Conceptual deﬁnition: The number of years that the caregiver attended school.
Operational deﬁnition: The caregiver indicated the level of education attained.
Choices in the original questionnaire included: 1 = grade school or less, 2 = some high
school, 3 = completed high school, 4 = some college, 5 = completed college, 6 =
some graduate school, or 7 = completed graduate school. These were recoded for the

present study so that 1 = some high school or less, 2 = high school, 3 = some college,

54

and 4 == college or more for descriptive purposes. Education was also recoded as a
continuous variable (6 = completed grade school or less, 9 = completed some high
school, 12 = completed high school, 14 = completed some college/technical school, 16
= completed college, 18 = completed some graduate school, 20 = completed
graduate/professional degree) for analytical reasons (correlations and multivariate

analyses).

Caregiver Occupation

Conceptual deﬁnition: The caregiver’s type of employment.

Conceptual deﬁnition: For the original questionnaire, the caregiver indicated her
occupation (an open-ended response). The responses were then coded: l = professional,
management, administrative; 2 = sales, clerical, technical; 3 = craft or operative; 4 =
laborer; 5 = farmer; 6 = homemaker; 7 = other. These were recoded for the present
study so that l = professional, management, administrative; 2 = sales, clerical,

technical; and 3 = other.

Caregiver Household Income

Conceptual deﬁnition: The household income of the caregiver.

Operational deﬁnition: The caregiver indicated the appropriate gross income
(before taxes) for her household: 1 = 0-1999; 2 = 2000-3999; 3 = 4000-5999; 4 =
6000-7999; 5 = 8000-9999; 6 = 10000-13,999; 7 = 14,000-17,999; 8= 18,000-2l,999;
9 = 22,000-25,999; 10 = 26,000-29,999; 11 = 30,000-34,999; 12 = 35,000-39,999;

13 = 40,000-44,999; 14 = 45,000-49,999; 15 = 50,000—54,999; 16 = 55,000-59,999;

55

and 17 = 60,000 and over. The present study used the same coding to allow a more

realistic report of the mean income than recoding into smaller groupings would allow.

Enployment Status (prior to caregiving)

Conceptual deﬁnition: The amount of paid work the caregiver was doing prior to
caregiving responsibility.

Operational deﬁnition: The caregiver indicated whether she was employed outside
the home prior to caregiving (yes or no response; 1 = yes, 2 = no). If the response was
yes, the caregiver next indicated whether the employment was full-time or part-time
(l = full-time, 2 = part-time). Caregivers who responded yes (employed prior to

caregiving) comprised the sample (N =236) for the present study.

Current Employment Status

Conceptual deﬁnition: The current employment status of the caregiver.

Operational deﬁnition: The caregiver indicated whether she was currently
employed full-time, part-time, self-employed, retired, leave of absence, unemployed, not
employed for pay, or quit work to care. Choices were: 1 = full-time, 2 = part-time,
3 = self-employed, 4 = retired, 5 = leave of absence, 6 = unemployed, 7 = not
employed for pay, and 8 = quit work to care. The present study used 1 = full-time
(includes self-employed), 2 = part-time, 3 = leave of absence, 4 = quit/retire, and
5 = unemployed. The instrument for Time 2 did not include the question about current
status. Instead, the caregiver was asked if there had been a change in employment status

in the last three months, and if so, how it shifted. Response choices included 1 =

 

full-time,

different j

Co-Reside
Co
living quar

op

currently I;
2 = no).

We recipi.

2 = no).

Care Recipz'

Conc

Oper

Unable t0 re:

56
full-time, 2 = part-time, 3 = quit working, 4 = leave of absence, 5 = retired, 6 =

different job/changed jobs, and 7 = work more as a caregiver.

Co-Residence with Care Recipient

Conceptual deﬁnition: Whether or not the caregiver and the care recipient shared
living quarters.

Operational deﬁnition: The caregiver indicated whether the care recipient
currently lived in the same household with the caregiver (yes or no response; 1 = yes,
2 = no). If the response was yes, the caregiver was asked whether she lived with the
care recipient prior to assuming the caregiver role (yes or no response; 1 = yes,

2 = no).

Care Recipient Age
Conceptual deﬁnition: The number of years that the care recipient had been alive.
Operational deﬁnition: The care recipient (or caregiver if the care recipient was

unable to respond) indicated the date of the care recipient’s birth.

Care Recipient Gender
Conceptual deﬁnition: Whether the care recipient was male or female.
Operational deﬁnition: The care recipient (or caregiver if the care recipient was
unable to respond) indicated the care recipient’s gender to the interviewer ( 1 = male,

2 = female).

 

Fami.

mern
the r
fami
1989

men

Fan

reel

57
Family Network

Conceptual deﬁnition: The number in the caregiver’s family.

Operational deﬁnition: The caregiver identiﬁed her living adult relatives and how
far (in miles) they lived from the caregiver. From that ﬁgure, the number of network
members who lived 50 miles or less from the caregiver was determined. The mean of
the network number living within 50 miles of the caregiver provided the measure of
family members available to help with eldercare. The Given studies (Given & Given,
1989) of the core data used the 50 mile criterion as a basis for determining those network

members most likely to offer aid. The same concept was used in the present study.

Family Help (amount)

Conceptual deﬁnition: The amount of help with eldercare that the caregiver
received from her family network.

Operational deﬁnition: The caregiver indicated the amount of help that each adult
relative provided with caregiving (1 = none or very little, 2 = a little, 3 = some, 4 =
quite a bit, 5 = a great deal). The mean score for the amount of help given by network
members who lived within 50 miles of the caregiver provided the measure of family help
with eldercare. At Time 2, the caregiver indicated whether in the past three months,

overall help from the family increased (I), stayed the same (2), or decreased (3).

Family Assistance frequency)
Conceptual deﬁnition: The caregiver’s recognition of the frequency of family help

with caregiving responsibilities.

 

provi

ofth

item:

Care

healr

felat

dail}
38th
wall

58

Operational deﬁnition: The caregiver indicated how often anyone in the family
provided assistance in the following ways:

helped with physical care

spent time keeping the care recipient company

stayed with the care recipient so that the caregiver could do something else for
a few hours

gave the caregiver emotional support

helped with transportation (caregiver or recipient)

helped with money or other material goods

checked on the caregiver to see if she was alright

provided or encouraged diversional activities for the care recipient

99‘!”

FPS“??-

Response choices for each item were: 1 = rarely or none of the time, 2 = some
of the time, 3 = most of the time, 4 = almost all of the time. The mean of the summed

items provided the measure of family assistance given.

Caregiver Total Involvement

Conceptual deﬁnition: The help with physical (ADL), instrumental (IADL), and
health care activities (HCA) of daily living that the caregiver provided for her elderly
relative.

Operational deﬁnition: The caregiver described her involvement with activities of
daily living (ADL), instrumental activities of daily living (IADL), and health care
activities (HCA) of daily living. The ADL categories included dressing, eating, bathing,
walking inside the house, toileting, and transferring in and out of bed. The possible
responses to how frequently the caregiver helped the care recipient were: 0 = never,
1 = once a week or less, 2 = several times a week, 3 = once a day, 4 = several times

daily. The sum of the individual item scores in the ADL section represented the ADL

 

involver
than dic
l
housewr
section,
ranged
scores i

more c;

OXFgen
ids, [u
Ulcers/1
medica
Katha

by are

59

involvement score. High scores indicated more caregiver involvement with ADL tasks
than did low scores.

Caregiver involvement with IADL activities included cooking/preparing meals,
housework, shopping, laundry, transportation, and money management. As in the ADL
section, the possible responses to how frequently the caregiver helped the care recipient
ranged from 0 = never to 4 = several times daily. The sum of the individual item
scores in the IADL section represented the IADL score. Here, too, high scores indicated
more caregiver involvement with IADL duties than did low scores.

Caregiver involvement with HCA functions (urinary catheter/catheter care,
oxygen administration, IV, Hickman, Broviac catheter care/dressing, IV medications/ﬂu-
ids, tube or IV feedings, injections, special exercises/physical therapy, care of
ulcers/bedsores, slo'n care, colostomy care, care of post-operative incision, oral
medications, nasogastric tube and care, incontinence of urine, incontinence of stool,
tracheostomy/tracheostomy care, respirator/care of respirator, suctioning) was measured
by asking the caregiver if the care recipient required any of the activities.

For the indicated activities, the frequency of assistance provided by the caregiver
was assessed. Scores for each indicated item were: 0 = never, 1 = once a week or less,
2 = several times a week, 3 = once a day, 4 = several times daily. The sum of the
scores for pertinent items in the HCA section provided the health care activity
involvement score. As with ADL and IADL scores, the higher the score, the greater the
caregiver involvement in those activities. The sum of the scores from ADL, IADL, and
HCA sections represented the caregiver’s total involvement with those activities.

Although the total involvement score was the measure, the sum of each section enabled

 

the d1
1 and

montl

Hours

theca

Ueatm.

PTOVid;

Hours .

hours).

Safest) 1

number

60

the determination of internal ﬂuctuation with ADL, IADL, and HCA functions at Time
1 and Time 2. At Time 2, the caregiver indicated the amount of care compared to three
months ago. Response choices were: 1 = less care, 2 = about the same, and 3 = more

eare.

Hours of Physical Care

Conceptual deﬁnition: The time the caregiver spent providing physical care for
the care recipient.

Operational deﬁnition: The caregiver indicated, in a usual day, how many hours
(0-24) were spent providing physical assistance (activities such as eating, dressing,
treatments, etc.) for her relative. A mean score for the number of hours per day spent

providing care provided the measure.

Hours of Supervision
Conceptual deﬁnition: The time the caregiver spent supervising the care recipient.
Operational deﬁnition: The caregiver indicated, in a usual day (excluding night
hours), how many hours were spent supervising (being with the recipient to ensure
safety) her relative. Hours could range from 0 to 20 per day. A mean score for the

number of hours per day spent supervising care provided the measure.

Instrumentation
Telephone interviews and self-report questionnaires provided the data for the core

study, "Caregiver Responses to Managing Elderly Patients at Home. " The portions of

 

instrur
core 3
conten

sound

61
the instrument to be used in the present research included the following sections from the
telﬁphoﬂe interview: (3) W1. 0)) W. (C)
mm. (d) We: (of the family). and (e) CaregixcLlnxoLvernent
and (O W.

The research team of Given, Given, Collins, King, and Stommel developed the
instruments (with the exception of the ADL and IADL items taken from OARS) for the
core study. Experts and peers evaluated the measures and concluded they had face,
content, and construct validity. All tools used in the core study are psychometrically
sound research instruments that have been tested on over 600 caregivers during the ﬁrst
caregiver study, "Caregiver’s Response to Managing Elderly Patients at Home”
(Caregiver I). Examples from the instruments are in Appendix A.

The dependent variables, employment adaptations (Type I, Type II, and Type III),
were described at two intervals; one at inception of the caregiver role (Time 1) and the
second at three months (Time 2). The measure used was the W
section of the core instrument. This survey assessed how many times in the last month
(or three months for Time 2) the employee adapted employment in order to meet
caregiving responsibilities. Employment adaptation, with the exception of "other work
effects” (Type I), which required an open-ended response, was self-reported as yes or
no (1 = yes; 2 = no) and included:

a. Type I employment adaptation (maintain, but alter work)

arrive late/leave early
miss work without pay
change work hours

take sick/personal days
refuse job or promotion

 

ker

b. T):
c. Ty]

At b0tl

62

kept from job hunt/better job

other work effects
b. Type II employment adaptation (leave of absence)
c. Type III employment adaptation (quit/retire)

At both Time 1 and Time 2, the caregiver indicated how many work days were

missed in order to provide care (open-ended response) and also how disruptive caregiving

had been on normal activities. Response choices for disruption of activities included:

1 = not at all

2 = a little

3 = somewhat
4 = a great deal

The screening instrument and the Time 1 telephone questionnaire for the core

study contributed the caregiver and care recipient sociodemographic information for the

present study. The data included:

1.

2.

Care recipient gender (1 = male; 2 = female).
Caregiver gender (1 = male; 2 = female).

Care recipient birth date _month, _day, _year.
Caregiver birth date _month, _day, _year.

Caregiver marital status (1 = single, never married, 2 = married/remarried, 3 =
divorced, 4 = widowed, 5 = separated). This was recoded for the present study
(1 = married, 2 = unmarried).

Caregiver relationship to care recipient (l = spouse, 2 = parent, 3/4 =
daughter/son, 5/6 = daughter/son-in-law, 7/ 8 = sister/brother, 9/10 = sister/bro-
ther-in-law, 11/12 = granddaughter/son, 13/ 14 = niece/nephew, 15/ 16 =
aunt/uncle, 17 = other relative, 18 = friend/companion, 19 = other non-relative).
This was recoded for the present study (1 = spouse, 2 = daughter/daughter-in-
law, 3 = other).

Caregiver highest level of education (1 = completed grade school or less, 2 =
completed some high school, 3 = completed high school, 4 = completed some
college/technical training, 5 = completed college, 6 = completed some
graduate/professional school, 7 = completed graduate/professional degree). This

 

was r
or 1e
colic,
conti
prese
scho
scho
corn]

3. Care
cleri
horn
man

9. Live
YES,
10. H01
5,9<
17'

30,1
15

family. '

their diSt

10.

63

was recoded for the present study for descriptive purposes (1 = some high school
or less, 2 = high school, 3 = some college/technical training, 4 = completed
college/technical training or more). The core study values were also recoded to
continuous variables for correlations and multivariate analyses performed in the
present study (6 = completed grade school or less, 9 = completed some high
school, 12 = completed high school, 14 = completed some college/technical
school, 16 = completed college, 18 = completed some graduate school, 20 =
completed graduate/professional degree.

Caregiver occupation -(l = professional, management, administrative; 2 = sales,
clerical, technical, 3 = craft or operative, 4 = laborer; 5 = merchant; 6 =
homemaker; 7 = other. This was recoded for the present study (1 = professional,
management, administrative; 2 = sales, clerical, technical; 3 = other).

Live in the same household, now (1 = yes, 2 = no); if the response was yes, the
caregiver was asked about co-residence before assuming the caregiver role (1 =
yes, 2 = no).

Household gross income before taxes (1 = 0-1,999; 2 = 2,000-3,999; 3 = 4,000-
5,999; 4 = 6,000-7,999; 5 = 8,000-9,999; 6 = 10,000-13,999; 7 = 14,000-
17,999; 8 = l8,000-21,999; 9 = 22,000-25,9999; 10 = 26,000-29,999; 11 =
30,000-34,999; 12 = 35,000-39,999; 13 = 40,000-44,999; 14 = 45,000-49,999;
15 = 50,000-54,999; 16 = 55,000-59,999; 17 = 60,000 and over).

The Family Network instrument assessed information about the caregiver’s

family. The caregiver listed family members (spouses, children, siblings, and in-laws),

their distance (in miles) from the caregiver, and then indicated how much help with

eldercare that each network member provided. Response choices were:

1 = none or very little
2 = a little

3 = some

4 = quite a bit

5 = a great deal

At Time 2, the caregiver indicated whether overall help from the family increased,

decreased, or stayed ”about the same" during the past three months.

 

The la
caregivers pe
The present

Reliability fo

roar-”9999:.”
:-
(D

Ath-d
II H H H

s?

(Time 1) am
“8 used to
items that l-Z
Mimics of
Mimics 0f
dres:
eatin
bathr

Walk
lone.

64
The last instrument used for the study, the W scale, appraised the

caregiver’s perception of the frequency of assistance received from family and friends.
The present study used only the portion dealing with the family of the caregiver.
Reliability for the W scale was alpha .69. Items included:

helped with physical care

spent time keeping your _ company

stayed with _ so that you could do something else for a few hours

given you emotional support or encouragement

helped with transportation; for either you or care recipient

helped you with money or other material goods

checked on yoo to be sure that you were all right

provided or encouraged diversional activities for (care recipient) such as cards
or Scrabble, etc.

5"?" rm 99 9‘9

Caregiver response possibilities were:

rarely or none of the time
some of the time

most of the time

almost all of the time

ammu—
II II II II

The independent variable, caregiver involvement, was measured at inception
(Time 1) and at three months (Time 2). The Involvement portion of the core instrument
was used to determine the caregiver’s responsibility for eldercare. This section contained
items that tapped the degree of assistance with physical, instrumental, and health care
activities of daily living that the caregiver provided for the care recipient. Physical
activities of daily living (ADL) included:

dressing

eating

bathing

walking inside the house

toileting
transferring

 

lnstrun

Possib?

activiti

The he

Possibr.

65
Instrumental activities of daily living (IADL) included:

cooking

housework
shopping

laundry
transportation
money management

Possible responses for how often the caregiver provided help with both ADL and IADL
activities were:

0 = never

1 = once week/less

several times/week
once a day

2
3
4 several trmes/ day

The health care activities (HCA) included:

urinary catheter/catheter care

oxygen administration

IV/Hickman or Broviac catheter care/dressing
IV medications/ﬂuids

tube or IV feedings

injections (pain meds or insulin)
special exercises/physical therapy
care of ulcers/bedsores

skin care (special cleansing lotions)
colostomy/colostomy care

care of post-operative wound/incision
oral medications

nasogastric tube and care
incontinence of urine

incontinence of stool
tracheostomy/tracheostomy care
respirator/ care of respirator
suctioning

Possible responses for how often the caregiver provided help with HCA were:

never
once week/less

0
1
2 several times/ week

 

C
s:

3
4

Relia‘.
activities of
impaired pa1

An :
mu
mregiver v
number of
bathing, d:
night hour
Physical a:
per day; fr
“Ether s

°°mpared

Sampling
Tl,

medical c'

eldercare

tion beCaL

(dream!

66

3 = once a day
4 = several times/day

Reliability for the Involvement scale was alpha 0.86 for caregiver involvement in
activities of daily living, 0.82 for instrumental activities of daily living (for physically
impaired patients), 0.52 for health care activities (based on 11 HCA).

An additional assessment of caregiver involvement was obtained through Ibo
W. The hours of both physical care and supervision spent by the
caregiver were measured at inception and three months. The caregiver estimated the
number of hours per day spent providing physical assistance (activities such as eating,
bathing, dressing, treatments) and also calculated the number of hours per day (excluding
night hours) consumed supervising the care recipient. Supervision time did not include
physical assistance. Response possibilities for physical care ranged from 0 to 24 hours
per day; for supervision, from 0 to 20 hours per day. At Time 2, the caregiver indicated
whether she was providing more care, less care, or about the same amount of care

compared to three months ago.

Sampling Procedure

The sample for the core study consisted of 630 family caregivers of elderly
relatives. The majority of care recipients had a sudden, dramatic health crisis with a
medical diagnosis involving the circulatory system. The recruitment criteria for the
eldercare dyad follows. Patients (a) were aged 55 or older, (b) had a recent hospitaliza-
tion because of some event or disease progression that qualiﬁed them for skilled home

care, and (c) had increased dependencies (within the past 30 days) in at least one physical

 

aetivity o
(lADL) 0
member 1
had shift:
The care
in the st
1
criteria.

elderca]

(SD

67

activity of daily living (ADL) or at least two instrumental activities of daily living
(IADL) or one equipment and/or health care activities. The caregiver was (a) the family
member providing the most care and (b) during the 30 days prior to intake into the study
had shifted assistance with one or more IADL activities to one or more ADL activities.
The caregiving dyads were recruited by 27 sites (hospitals and visiting nurse agencies)
in the state of Michigan, including one in the upper peninsula.

The sample for the present study (N =236) met the following additional caregiver
criteria. The caregivers were (a) female and (b) employed prior to assuming the
eldercare role. The care recipient criteria remained as described in the preceding
paragraph. At Time 2, the number of subjects in the core study dropped to 312. The
decreased sample size was due to a variety of reasons. In some cases, the primary
caregiver changed or the original caregiver decided not to continue with the study; in
other instances, the care recipient either no longer required care, was hospitalized or
placed in a nursing home, and still other care recipients died. A sort and match by
identiﬁcation number of Time 1 and Time 2 subjects for the present study indicated the
original sample fell to 119 over the three-month period.

Most caregivers were married (75.4%); the mean household income was $35,812
(SD = $17,426). Caregivers included daughters or daughters-in-law (60.6%), wives
(29.7%), and other relatives (9.7%). The caregivers averaged about 51 (50.67; range
= 19.59 - 75.96; SD = 11.97) years of age, while the mean care recipient age was
approximately 73 (73.36; range = 53.11 to 93.30; SD = 9.16) years. Most care

recipients were female (57.2%). Figure 3 contains other caregiver characteristics.

 

. - ii .. i--. i it iiHi \CT.
#iAL. Z ..... 32.: mUU 27.2 Z _ 2.2: D :75... Z 323.331....9Sula \

 

68

Emasmm

3 i EmESEEm iv
mocmuamm

mamo c:

-O

”v ﬁdgw mc:H
c_m©cmco tag

me;

m

3.3.8.0830 325930 .n 952“.

$me 952E 9 mac 6me mg 6c 22: c “.202

.86 .m
com<_mo:m_o\mm_mm ..
559368205 _m

Al 8283000 iv

90869.60 863800 .U
cosmosnm coo<mmm=oo 2:8 .0
506m :9; umsmagoo .9
$0300ch :9: mEom .m

 

.....

.....

 

 

 

 

 

 

IIIIIIIIIIIIIIIIIIIIIIII

 

.....

I ttttttttttttttttt

:3

 

 

 

UM “a” ”at

 

 

 

 

 

 

IIIIIIIIIIIIIIIIIIII

I l

i c2683 iv

 

 

 

 

 

 

 

I I I I I I I I I I I I I I I I O I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I l

 

 

 

minz m oEPQ 8th P oEEU cmmuz mcsaocmoddg

 

 

 

 

Or

ON

0m

0?

OD

O©

Oh

Omw

mm

Be
Home',\
obtained
dmmu

T.
employer
gmeraliz
recipient,
instrume.
identify .
acuvities
transfem‘
619-) is t
aCtlvilieg
“€113 With

Thong): m

69

Because the core study, "Caregiver Responses to Managing Elderly Patients at
Home” , was conducted at Michigan State University, UCRIHS review and approval was
obtained by the research team. This researcher also secured the review and endorsement
of that committee.

The sample was non random; thus, results cannot be generalized to apply to all
employed female caregivers of elderly relatives. Statements about the results can be
generalized only for the caregivers in the present study. However, the patient (care
recipient) criteria used for the sample (needing increased assistance in physical or
instrumental activities of daily living or health care activities) present a deﬁned way to
identify the beginning of a new caregiver relationship. Having to help with physical
activities of daily living (dressing, eating, bathing, walking inside the house, toileting,
transferring) or health care activities (catheters, colostomies, oxygen, IV lines, injections,
etc.) is believed to be especially burdensome to caregivers. Involvement with those
activities may be more likely to cause conﬂict with work and family roles than would
help with instrumental activities (cooking, housework, shopping, laundry, transportation,
money management).

Using this approach to deﬁning a new relationship allowed the study of may
versus well-seasoned caregivers. The information thus obtained provides valuable insight
into the inﬂuence of acquiring the caregiver role on employed female caregivers.
Although preliminary, ﬁndings identify areas for future investigation. Thus, results may
be the impetus to further study of the issue with a larger cross section of caregivers for

a longer period of time and, if possible, a random sample.

 

Data Collt
As
Michigan
National 1
and for p
data. In
an intere
analysis
a tremer
Elderly
and its e

H) rece

he no

 

 

70
Data Collection Techniques

As indicated, the present study used data from longitudinal caregiver studies at
Michigan State University, funded by the United States Health and Human Services
National Institute on Aging. Data used by persons not part of the ﬁrst investigative team
and for purposes different from the original research intent are considered secondary
data. In such cases, the principal investigator shares data with an outsider who may have
an interest related to the initial research goal. According to Babbie (1989), secondary
analysis makes it possible for individuals to pursue particular research interests without
a tremendous investment of time and money. The ”Caregiver Responses to Managing
Elderly Patients at Home" (Caregiver I) was funded through 1986-1989 with $389,552
and its extension, "Caregiver Responses to Managing Elderly Patients at home (Caregiver
II) received continued funding of $996,669 through 1992.

The data from the core study provided a rich store of information, much of which
has not been analyzed (including the present study’s dependent variable, employment
adaptation). Thus, the present study analyses heightened the importance of the research
and the value of the monies already invested.

Much data for the core study were obtained through telephone interviews by
trained interviewers at the study’s inception (recruitment for the study began in
September 1989) and then at three-month intervals for 18 months. Data for the present
study (from telephone interviews) were from inception (Time 1) and three months (Time
2). Telephone interviews were used because previous research by the Michigan State
University caregiver research team suggested that the quality of the data did not suffer,

the caregivers were receptive to the approach, and the method was cost effective.

 

Caregis
suppler

inform;

Analys

screeni
from v
Used ir
vaﬁgd
the pn
Utilize
instm,
(the d
treate

Sard)

f0r ']
T10”
dE\r:

X113]

71

Caregivers completed self-report questionnaires after each telephone interview to
supplement the information obtained by the interviewer. The present study did not use

information from the self-report questionnaires.

Analysis of Data

The data ﬁle for the core study was constructed by including variables from the
screening and telephone interviews and the self-administered questionnaires. Only data
from variables in the screening and telephone interviews (for Time 1 and Time 2) were
used in the present study analyses. The actual number of cases for both time periods
varied on some items due to respondent omissions. Also, the number of total cases for
the present study dropped from Time 1 (N=236) to Time 2 (N =ll9). The variables
utilized include a mix of nominal, ordinal, and ratio measures. In some cases, the survey
instrument uses Likert-type response scales. While these constitute an ordinal measure
(the differences between responses may not be equal), for statistical purposes they were
treated as interval data. This represents a typical approach. According to Grosof and
Sardy (1985), most statistical analyses consider such data as interval if there are more
than ten items with ﬁve or more choices for each.

The approach to data analyses, using the SPSS system, follows. First, the data
for Time 1 and Time 2 were inspected for patterns or irregularities (gaps or outliers) and
normal distribution. Frequency tables, descriptive statistics such as means, standard
deviations, and relative and cumulative frequencies were used to inspect the data.

Inspection of the frequency tables revealed how scores clustered for some variables.

72

This process enabled the recategorization of certain variables (caregiver marital
status, relationship to the care recipient, education, employment status, occupation) for
the present study. Since the sample size for the present study was smaller than the core
study, the value results for those variables sometimes included very small percentages.
Combining two or more values with small percentages improved the usefulness of results
for the present study.

For example, frequencies for family relationship, using core study categories,
generated percentages that were relatively high for spouse (29.7) and daughter (53.4)
afﬁliations. However, other family relationship divisions had values with much smaller
percentages (.4, 1.7, and 3.4). A larger percentage of caregivers (7.2) were daughters-
in-law, but that was still small in comparison with spouse and daughter percentages.
Thus, caregivers who were daughters-in-law were grouped with daughter and the other
low-percentage relationship categories were combined to form a new grouping, other.
The resulting three classiﬁcations, spouses, daughter/daughter-in-law, and other, showed
more equivalent representation in the present study sample.

Skewness checks for less than + 2 or - 2 and kurtosis inspection for Type I
(maintain but alter work) adaptation, family support, and total caregiver involvement
showed normality of distribution for the variables. Descriptive statistics proﬁled each
type of employment adaptation (including Type I components), caregiver and care
recipient characteristics, family support, and caregiver involvement at Time 1 and
Time 2.

Next, the mix of items that produced the summary score for total caregiver

involvement (ADL, IADL, and HCA) at Time 1 and Time 2 was examined. Even

73

though the summary score changed little from Time 1 to Time 2, the factors comprising
each score were different. This inspection of the item mix revealed internal shifts
impossible to note by review of a total score. However, in spite of internal shifts, total
caregiver involvement was used as the measure for the present study. This researcher
recognized that caregivers provided a mixture of the components (ADL, IADL, HCA)
and that their sum (total involvement) would generate a more complete measure than
independent analyses alone.

With the exception of analysis methods that linked Time 1 with Time 2 (paired
t-tests, McNemar tests, and logistic regression), the same techniques (correlations,
oneway ANOVA, contingency tables, independent t-tests, multiple regression, and
discriminant analysis) were used to study both time periods. In the study, Type I
employment adaptation (an index of maintain, but alter work adaptations) was treated as
a continuous variable, while Type 11 (leave of absence) and Type III (quit/retire)
adaptations were considered categorical variables. Signiﬁcant ﬁndings for each question
required a p < .05 unless the question dictated a descriptive response. Speciﬁc data

analysis procedures for each research question follow.

Question 1: Does employment adaptation occur, and if so, does the adaptation type
differ between Time 1 (inception) and Time 2 (three months)?

Data for Type I (maintain, but alter work), Type H (leave of absence), and Type
III (quit/retire) employment adaptation were summarized for each time period, using

descriptive statistics (means, standard deviations, relative, and cumulative frequencies)

74

appropriate for the variable under consideration. The mean, standard deviation, and
range of work days missed (past three months) at Time 1 and Time 2 were calculated.

To assess Time 1 and Time 2 changes in Type II and Type III adaptations
(categorical variables), the differences in the frequencies of each were noted. Willett
(1988) claims that difference scores have been incorrectly perceived as unreliable and
invalid and that they are, in fact, a reasonable measure of change. Because Type I
adaptation is a continuous variable, the mean differences between Time 1 and Time 2
could be decided.

Next, to check the signiﬁcance of the differences between the time periods,
additional analyses followed, using Time 2, Type I, II, or III employment adaptation as
the dependent variable and Time 1, Type I, II, or III as the independent variable. A
paired t-test was used to determine the signiﬁcance of differences between Time 1 and
Time 2, Type I adaptation; McNemar tests were utilized to examine the signiﬁcance of
differences between Time 1 and 2, Type II and Type III employment adaptations.

The paired t-test was used to assess differences between Type I adaptation at Time
1 and Time 2, because the Type I sample means came from the same research subjects,
but were obtained at two time periods. Craft (1990) notes that when some factors that
inﬂuence scores in one group also sway scores in the other group, the signiﬁcance of the
difference between the means can be tested with a matched group t-test. The t-test
indicates whether scores have changed as a group from Time 1 to Time 2. However,
Craft (1990) warns that comparing only means may produce an erroneous conclusion

about score change, especially if the difference between the means is insigniﬁcant.

75

The McNemar test for two correlated dichotomous variables (Norusis, 1990b) was
used to ascertain whether the changes in proportions were the same for pairs of
dichotomous variables (Time 1 and Time 2, Type II and Type 111). According to Norusis
(1990b), the pairs of variables being tested must have the same two values; in the present
study, 1 = yes and 2 = no for Time 1 and Time 2, Type 11 (leave of absence) and Type

III (quit/retire) employment adaptations.

Question 2: What is the probability that Time 1, Type I employment adaptation
(maintain, but alter work) influences Time 2, Type II (leave of
absence), and Type III (quit/retire) employment adaptations?

Logistic regression, a technique used with dichotomous variables that allows the
direct estimate of the probability of an event occurring (Norusis, 1990c), was used.

Hosmer and Lemeshow (1989) state that having a dependent variable that is binary

(dichotomous) differentiates a logistic regression model from the linear regression model.

The guiding principle in logistic regression is to compare observed values of the response

variable to predicted values obtained from models with and without the variable in

question (Hosmer & Lemeshow, 1989).

In the present study, the coefﬁcient for Time 1, Type I adaptation (alter, but
maintain work) forecasted the probability of the caregiver displaying Type 11 (leave of
absence) or Type III (quit/retire) employment adaptations at Time 2, given the level of

Type I adaptation at Time 1. The variables for Type H and Type III adaptations were

recoded to dummy variables (1 = yes, 0 = no) for the logistic analyses.

76

The regression equation is:

Y: 1

where Y = the probability of Type II or Type III adaptation at Time 2
Bo & B1 = coefﬁcients estimated from the data
X = the independent variable, Time 1, Type I adaptation
e = base of natural logarithms, approximately 2.718

Norusis (1990c) notes that in logistic regression, the parameters of the model are
established using the maximum-likelihood method (coefﬁcients that make observed results
most likely are chosen). Hosmer and Lemeshow (1989) claim that in a regression
analysis when the outcome is dichotomous, the following apply: 1) The conditional mean
of the regression equation must be formulated to be bounded between zero and one, 2)
The binomial, not the normal, distribution describes the distribution of the errors and will
be the statistical distribution upon which the analysis is based, and 3) The principles that

guide an analysis using linear regression also guide logistic regression.

Question 3: What is the relationship of selected caregiver and care recipient
characteristics to each type of employment adaptation at Time 1
and Time 2?

Pearson correlations, oneway analysis of variance (ANOVA), and contingency
tables were used to demonstrate the association between selected caregiver and care
recipient characteristics and each employment adaptation type. In this question,
employment adaptation (each type) was the dependent variable, while selected caregiver

and care recipient characteristics comprised the independent variable. Some caregiver

and care recipient characteristics used to answer Question 3 were categorical (caregiver

77

marital status, relationship to the care recipient, occupation, current employment status,
co-residence with the care recipient now and before, care recipient gender), while others
were continuous (caregiver age, household income, education, and care recipient age).

Correlations require continuous dependent and independent variables. They were
used to show the relationship of Type I (maintain, but alter work) employment adaptation
to caregiver and care recipient characteristic continuous variables. Correlations provide
an index of the degree of relationship between two interval-scaled variables. The
Pearson r is the measure of the strength and direction of the linear relationship between
two variables (Craft, 1990). The Pearson r varies in direction and strength from -l.00,
a strong negative relationship to +1.00, a strong positive relationship. A zero (0) value
indicates there is no relationship between the variables. Craft (1990) states that
assumptions for correlations include: 1) interval level of measurement, 2) linear
relationship, and 3) homoscedasticity (the variances in Y at each X should be similar).

Oneway ANOVA was used to examine the variation in the sample and to test the
signiﬁcance of mean group differences for Type I employment adaptation and the
categorical variables. Oneway ANOVA utilizes a continuous dependent variable and
categorical independent variables. Oneway ANOVA helps the researcher determine
whether the differences between two or more means are greater than would be expected
by sampling error alone. For the categorical variables with more than two groupings
(family relationship, caregiver occupation), Tukey’s HSD (honestly signiﬁcant
differences) test also was done. This multiple range test provided pairwise comparisons

to determine how the groups differed.

78

ANOVA assumptions include: 1) interval measure dependent variable, 2)
probability sampling, 3) normality of the dependent variable in the populations studied,
and 4) homogeneity of variance (Craft, 1990). Craft claims that assumptions may be
violated somewhat since ANOVA is such a robust and powerful statistical procedure.

The ANOVA formula is:

Y=u+a+e

where Y = variable being tested
u = the grand mean
a = the effect of the time period
e = residual error

Contingency tables (the cross-tabulations of frequencies for the combination of
categories of two variables, Glass & Hopkins, 1984) were used to demonstrate the
relationship between Type II (leave of absence) and Type III (quit/retire) employment
adaptations and the categorical caregiver and care recipient characteristics. In
contingency tables, the values of the dependent variable are contingent upon the value
of the independent variables (Babbie, 1989).

The contingency coefﬁcient (C) examines the association between two sets of
variables when only nominal information about one or both sets is available (Touliatos
& Comptom, 1988). In this study, some variables used to answer Question 3 were
nominal. Chi-square was used for the analysis of the contingency tables. The higher the
chi-square value, the less likely it is that the value is due to sampling error alone.
Grosof and Sardy (1985) note that chi-square can be used with all levels of data and

requires only minimal assumptions. These are: l) the sample is a simple random sample,

79

2) the samples (if more than one) are independent, and 3) for each characteristic of
interest, each subject can be assigned to one class or category that is mutually exclusive.

Finally, in order to study the relationship between Type II and Type III
employment adaptations with the continuous dependent variables, Type II and Type III

were recoded to dummy variables (1 = yes, 0 = no) and correlations were done.

Question 4: How much variation in each type of employment adaptation at Time
1 and Time 2 do selected significant caregiver and care recipient
characteristics explain?

Before multivariate analyses (multiple regression or discriminant analyses) were
used, bivariate associations among the variables and each adaptation type for Time 1 and
Time 2 were examined. This process allowed the elimination from multivariate analyses,
potential explanatory factors that did not have signiﬁcant associations with the adaptation
types.

Norusis (1990b) claims that one regression procedure encompassing all variables
is an inappropriate strategy to use. A long list of variables makes interpretation difﬁcult,
because each variable in the equation partials out the other variables, producing a smaller
co-efﬁcient for each and less effect in the regression equation. Norusis (1990b)
maintains that the observed increase in R2 from the addition of more independent
variables does not necessarily produce a better model. Conversely, using irrelevant
variables increases the standard error of all estimates without improving prediction. This
concept also applies to discriminant analysis.

Both multiple regression and discriminant analysis offer a fuller explanation of

the dependent variable (more than one independent variable enters the equation) than

80

univariate analysis, since few phenomena have only one explanation. This approach is
consistent with the ecological framework of the study.

Employment status, although a caregiver characteristic, was not used in the
analyses for Type I, Type II, or Type III adaptations. This was done in order that
caregiver and care recipient characteristics demonstrating signiﬁcant relationships with
the adaptation types could be assessed for their unique inﬂuence on employment
adaptation. Including employment status could diminish the impact of other caregiver
and care recipient characteristic variables on employment adaptation. Lewis-Beck (1989)
claims that multiple regression describes the precise effect of a particular independent
variable, because distorting inﬂuences from other independent variables are removed.
However, Craft (1990) cautions that the relative importance of any given variable
depends on which variables are included in the analysis.

Stepwise multiple regression provided data to determine the inﬂuence of selected
caregiver and care recipient characteristics (caregiver age, education, household income,
relationship to care recipient, care recipient gender) for Time 1, Type I (maintain, but
alter work) adaptation, while two-group discriminant analysis (Wilks’s method) was used
to determine whether selected caregiver and care recipient characteristics (caregiver
education, co-residence now) could differentiate those caregivers who used Time 1, Type
11 (leave of absence) adaptation from those who did not. Time 2 data did not suggest the
nwd for additional analyses with those variables and Type I and Type II adaptations.
An independent t-test provided data to assess the signiﬁcance of caregiver household
income with Time 1, Type III (quit/ retire) adaptation and that of caregiver education with

Time 2, Type III adaptation.

81

Certain of the caregiver and care recipient characteristics (caregiver relationship
to care recipient, co-residence now, care recipient gender) were recoded to dummy
variables, using the rule "When a noninterval variable has G categories, use G - 1
dummy variables to represent it" (Lewis-Beck, p. 68, 1989). Next, a stepwise regression
approach was used so that variables offering a contribution at the p < .05 level to the
explanation of Type I adaptation could be determined. In stepwise regression, the
independent variable with the largest probability of F is removed if the value is p > .10;
the equation is recomputed omitting the removed variable, and the process is repeated
until no variables can be removed. Next, the independent variable not in the equation
with the smallest probability of F is entered if this value is smaller than .05. All
variables in the equation are again examined for removal. The process continues until
no variables in the equation need removal and no variables oot in the equation meet
eligibility for entry requirements (Norusis, 1990b). With each variable entered into the
equation, all variables are partialed out so the effect of the latest can be known (Glass
& Hopkins, 1984). In the ﬁnal step, the variables not meeting entry requirements are
listed with the beta for each noted i_f the variable was entered at that point in the

regression.

82

The regression equation is:

Y = a + le + bX2 + bX3 +...bX6 + e

where Y = dependent variable, employment adaptation
a = the intercept between employment adaptation and the caregiver and care
recipient variables
b = the increase or decrease in Y for a one-unit change in X
X, = caregiver age
X2 = relationship to the care recipient
X3 = caregiver education
X4 = caregiver household income
X5 = co-residence status now
X6 = care recipient gender
e = residual error

Lewis-Beck (1989) offers the following regression assumptions: 1) no speciﬁca-
tion error (the relationship is linear, no relevant independent variables have been
excluded, no irrelevant independent variables have been included), 2) no measurement
error, 3) assumptions met that concern the error term (zero mean, homoscedasticity, no
autocorrelation, the independent variable is uncorrelated with the error term, normality
of distribution), and 4) the absence of perfect multicollinearity (none of the independent
variables is perfectly correlated with either another or a linear combination of others).

For Type II employment adaptation, two-group discriminant functional analysis
(Wilks’s method) was used. This procedure, closely related to multiple regression, is
used to identify relationships when the dependent variable is dichotomous (Norusis,
1990c). Norusis (1990c) notes the two sets of coefﬁcients (discriminant and regression)
are always proportional with two-group discriminant analysis. A desired outcome of
discriminant analysis, as in regression, is the identiﬁcation of the independent variables
most important for group separation (good ”predictor" variables) and which are

extraneous. Discriminant analysis, not multiple regression, was appropriate for analyzing

83

Type 11 (leave of absence) adaptation, because only a dichotomous response (yes or no)
was available.

Discriminant analysis generates standardized and unstandardized discriminant
function coefﬁcients. The unstandardized units are the ”multipliers of the variables when
they are expressed in the original units" (Norusis, 1990c, p. B-l4). Standardizing the
units to a mean of 0 and a standard deviation of 1 (as with multiple regression) produces
the standardized discriminant function characteristic (Norusis, 1990c).

The interpretation of the coefﬁcients is similar to multiple regression. The
magnitude of the unstandardized coefﬁcients do not provide a good index of their relative
importance to the equation if the variables differ in their measuring units (Norusis,
1990c). However, as with the beta coefﬁcients of multiple regression, the standardized
canonical discriminant function coefﬁcients show the relative importance (but not the
actual inﬂuence) of the predictor variables (Craft, 1990). According to Klecka (1980),
the larger the magnitude of the standardized canonical discriminant function coefﬁcient
(regardless of the sign), the greater the variable contribution to the equation. However,
the relative importance of any given variable depends on the variables included in the
equation.

Also, Klecka (1980) cautions that if two variables are highly correlated, they
share their contribution to the score (even if that joint contribution to the score is very
important), resulting in smaller standardized coefﬁcients than if only one of the variables
is used. He warns that the coefﬁcients of two variables might be large, but with opposite
signs, thus partially canceling the contribution of one variable by the reverse contribution

of the other variable.

84

Grosof and Sardy (1985) offer the following assumptions for discriminant
analysis: 1) at least two groups, 2) at least two cases per group, 3) any number of
discriminating variables, but must be less than the number of cases minus two, 4)
continuous interval or ratio discriminating variables (some authors relax this assumption),
5) independent discriminating variables; no one is a linear combination of the others, nor
are any pair or subset highly intercorrelated, 6) each group is drawn from a population
whose distribution on the discriminating variables is multivariate normal, and 7) the
covariance matrices (intercorrelations) for each group must be approximately equal.
Grosof and Sardy (1985) claim that assumptions six and seven are fairly robust, but warn
that if there are many borderline cases, the accuracy of classiﬁcation predictions can be
impaired.

The linear discriminant equation is:

D=B+BX1+BX2

where D = the dependent variable, Type II employment adaptation
B = coefﬁcients, estimated from the data
X = values of the independent variables
X1 = caregiver education
X2 = co-residence status

A stepwise approach (Wilks’s method), in which variables are entered based on
the smallest lambda for that step, was used. The stepwise approach produces an "optimal
set of discriminating variables" (Klecka, p. 53, 1980). At each step (twice the number
of independent variables), the variable that results in the smallest Wilks’s lambda is
picked for entry (Norusis, 1990c). If a discriminatory variable contributes signiﬁcantly
to the explanation, it is included in the discriminant function; otherwise, it is not (Grosof

& Sardy, 1985). Wilks’s lambda assesses discriminating power and considers both

85

differences between groups and the homogeneity within groups. Grosof and Sardy
(1985) claim the Wilks’s statistic is preferred, since it can be converted into chi-square
distributions, its values can be tested for signiﬁcance, and it can be easily interpreted.

An independent t-test (two-tailed) was used to check the signiﬁcance of mean
differences in caregiver household income with Time 1, Type III adaptation. According
to Norusis (1990b), a t-test allows determination of whether two population means are
equal. Norusis ( 1990b) maintains that the observed signiﬁcance level from the t-test is
the probability that differences at least as large as the observed difference would have
occurred if the means were really equal. In the present study, the equality of means for
the household income of those caregivers who chose to quit or retire at Time 1 and those
who did not was the item of interest.

An independent t-test (two-tailed) also was used to check the signiﬁcance of mean
differences in caregiver education with Time 2, Type HI adaptation. In the present
study, the equality of means for the educational level attained for caregivers who chose

to quit or retire at Time 2 and those who did not was the item of interest.

Question 5: Was there a reported change in the amount of help from family
members between Time 1 and Time 2?

Time 2 data from the answers for that query were summarized. Then, frequen-
cies for each possible response category were determined in order to answer this research

question.

86

Question 6: What is the relationship of family support to each type of employment
adaptation at Time 1 and Time 2?

Pearson correlations were used to display the association of family support
(number in network, amount of help provided, and frequency of assistance given) to
employment adaptation at Time 1 and Time 2. As noted, correlations require continuous
dependent and independent variables; thus, they were appropriate to show the relationship
of Type I (maintain, but alter work) employment adaptation to family support variables.
As in Question 3, Type 11 (leave of absence) and Type III (quit/retire) employment
adaptations were recoded to dummy variables before correlations with the independent
variables were determined. The interpretation and assumptions of correlations remain

as stated.

Question 7: How much variation in each type of employment adaptation at Time
1 and Time 2 does family support explain?

Unlike Question 4, signiﬁcant bivariate relationships with employment adaptation
did not determine the independent variables used for subsequent analyses to answer this
research question. All three variables were used in the regression and discriminant
equations because of the importance of family in the conceptual model for this study.
As in Question 4, stepwise multiple regression analysis showed the amount of variance
in the dependent variable, Type I (maintain, but alter work) employment adaptation,
explained by the independent variable, family support (number in network, amount of
help provided, and frequency of assistance given) at Time 1 and Time 2. The regression
assumptions and interpretation remain as stated with Question 4. The regression equation

is:

a:
II II

X
['0
11 II II II II

87

Type I employment adaptation

the intercept between employment adaptation and the family support
variables

the increase or decrease in Y for a one unit change in X

number in network

amount of help provided

frequency of assistance given

residual error

Two-group discriminant functional analysis (Wilks’s method) was used to

determine whether family support could differentiate those caregivers who used Type 11

(leave of absence) and Type III (quit/ retire) employment adaptations from those who did

not. The formerly noted assumptions and interpretation for discriminant analysis

(Question 4) prevail.

The linear discriminant equation is:

where D
B
X
X1
X2
X3

Time

D=B+BX1+BX2+BX3

the dependent variable, Type II employment adaptation
coefﬁcients, estimated from the data

values of the independent variables

number in network

amount of help provided

frequency of assistance given

2 data did not suggest the need for additional analyses with family

characteristic variables and Type 1, Type II, or Type III adaptations.

88

Question 8: Do caregiving responsibilities disrupt normal work and daily activities
at both time periods?

Data from Time 1 and Time 2 responses to this query were summarized. Then,
frequencies for each possible category were calculated in order to answer this research

question.

Question 9: Does the level of reported caregiver involvement change from Time 1
to Time 2?

Frequencies for reported changes in caregiver involvement at Time 2 were

determined in order to answer this research question.

Question 10: What is the relationship of caregiver involvement (activities and time)
to each type of employment adaptation at Time 1 and Time 2?

Pearson correlations were used to show the association of caregiver involvement
(total caregiver involvement, hours of physical care, hours of supervision) to Type I
(maintain, but alter work) employment adaptation at Time 1 and Time 2. As noted,
correlations require continuous dependent and independent variables; thus, they were
appropriate to show the relationship of Type I employment adaptation to caregiver
involvement variables. As in Questions 3 and 6, Type II and Type III employment
adaptations were recoded to dummy variables before correlating with the caregiver
involvement variables. The interpretation and assumptions of correlations remain as

stated.

89
Question 11: How much variation in each type of employment adaptation at Time

1 and Time 2 is explained by caregiver involvement (activities and
time)?

As in Question 4, bivariate associations among the variables and each adaptation
type were examined to eliminate from multivariate analyses potential explanatory factors
that did not have signiﬁcant associations. Next, stepwise multiple regression analysis
was used to determine the amount of variation in the dependent variable, Type I
(maintain, but alter work) employment adaptation, explained by the independent variable,
caregiver involvement (total caregiver involvement, hours of physical care).

Discriminant functional analyses provided data to assess the contribution of the
independent variable caregiver involvement (total caregiver involvement, hours of
physical care, and hours of supervision) to Type II and Type III adaptations. The
assumptions and interpretation for both regression and discriminant analyses remain as
stated.

The regression equation for Time 1, Type I is:

Y=a+bX1+bX2+e

where Y = employment adaptation
a = the intercept between employment adaptation and the caregiver involve-
ment variables
b = the increase or decrease in Y for a one-unit change in X
X1 = caregiver total involvement (ADL, IADL, and HCA)
X2 = hours of physical care
e = residual error

90
The linear discriminant equation for both Type H and Type III adaptation is:

where D = the dependent variable, Type II (or Type III) employment adaptation
B = coefﬁcients, estimated from the data
X = values of the independent variables
X, = caregiver total involvement (ADL, IADL, and HCA)
X2 = hours of physical care
X3 = hours of supervision

Time 2 data did not suggest the need for additional analyses with caregiver
involvement and Type I or Type II adaptations. An independent t-test (two-tailed)
provided data to assess the signiﬁcance of mean differences in hours of supervision with
Time 2, Type III (quit/retire) adaptation. In the present study, the equality of means for
the hours of supervision spent by caregivers who chose to quit or retire at Time 2 and

those who did not was the item of interest.

Question 12: Which identified key variables from questions 4, 7, and 11 are the

best predictors of each type of employment adaptation at Time 1 and
Time 2?

Type I = caregiver age, relationship to care recipient, caregiver
gender, amount of family help, hours of physical care

Type II = caregiver education, co-residence now, amount of family

help, hours of physical care

caregiver household income, family network, frequency of

family help, total caregiver involvement, hours of physical

care or caregiver education and hours of supervision)

Type III

As in Questions 4, 7, and 11, multiple regression and discriminant analysis
provided data to answer this research question and its hypotheses. The variables selected
for additional study using stepwise regression to analyze Time 1, Type I (maintain, but
alter work) employment adaptation were those signiﬁcant from the three preceding Time

1 regressions (caregiver age, caregiver relationship to the care recipient, care recipient

91

gender, amount of family help, and hours of physical care). To determine whether an
alternate regression approach would generate different outcomes for this question,
backward multiple regression also was used. With backward regression, all variables
initially enter the equation and then are systematically removed based on pre-established
criteria (Norusis, 1990b).

The signiﬁcant variables (caregiver education, co-residence now, amount of
family help, hours of physical care) from the previous Time 1 discriminant analyses were
used for further two-group discriminant analysis of Time 1, Type H (leave of absence)
adaptation. Similarly, signiﬁcant variables from the preceding Time 1 discriminant
equations (family network, frequency of family help, total caregiver involvement, hours
of physical care) or t-test analysis (caregiver household income) were used for more
study of Time 1, Type III (quit/retire) adaptation.

A review of Time 2 data did not suggest the need for additional analyses of Type
I or Type II adaptations with any of the variables. However, signiﬁcant variables from
Time 2, Type 111 independent t-test analyses (caregiver education, hours of supervision)
were used for additional investigation of Type III adaptation at Time 2.

The regression equation for Time 1, Type I adaptation is:

Y=a+bX1+bX2+..;bX5+e

where Y = employment adaptation
a = the intercept between Y and the independent variables
b = the increase or decrease in Y for a one-unit change in X
X1 = caregiver age
X2 = caregiver relationship to care recipient
X3 = care recipient gender
X4 = amount of family help
X5 = hours of physical care
e = residual error

where

where

where

92

The linear discriminant equation for Time 1, Type II adaptation is:

D=B+BX1+BX2+BX3+BX4

D = the dependent variable, Type II employment adaptation
B = coefﬁcients, estimated from the data

X = values of the independent variables

X1 = caregiver education

X2 = co-residence now

X3 = amount of family help

X4 = hours of physical care

The linear discriminant equation for Time 1, Type III adaptation is:

D=B+BX1+BX2+...BX5

D = the dependent variable, Type III employment adaptation
B = coefﬁcients, estimated from the data

X = values of the independent variables

X1 = caregiver household income

X2 = family network

X3 = frequency of family help

X4 = total caregiver involvement

X5 = hours of physical care

The linear discriminant equation for Time 2, Type III adaptation is:

D = the dependent variable, Type III employment adaptation
B = coefﬁcients, estimated from the data

X = values of the independent variables

X1 = caregiver education

X2 = hours of supervision

CHAPTER IV

Results

Study ﬁndings are presented in twelve sections that correspond to the research
questions. Supported hypotheses required a p = < .05 level of signiﬁcance. The ﬁrst
question asked whether employment adaptation occurred at Time 1 and Time 2; and, if
so, if the adaptation type (I, II, or III) differed between the time periods. The
probability of Time 1, Type I adaptation (maintain, but alter work) inﬂuencing Type 11
(leave of absence) and Type III (quit/ retire) adaptations at Time 2 received scrutiny
through the second research question.

The third, sixth, and tenth research questions queried the relationship of selected
caregiver and care recipient characteristics, family support, and caregiver involvement
with each type of employment adaptation. The fourth, seventh, and eleventh questions
examined the variation in employment adaptation explained by caregiver and care
recipient characteristics, family support, and caregiver involvement. The ﬁfth and ninth
questions (speciﬁc for Time 2) asked about changes in the amount of family help and the
measure of caregiver involvement over the past three months. The eighth question
queried whether caregiving responsibility disrupted normal work and daily activities.

Finally, the twelfth question asked about the inﬂuence of signiﬁcant variables determined

93

94

from the answers to questions four, seven, and eleven on each type of employment

adaptation. Study results for the questions and related hypotheses follow.

Research Question 1 Does employment adaptation occur, and if so, does
the adaptation type differ between Time 1 (inception)
and Time 2 (three months)?

First, frequencies or means and standard deviations for Time 1 and Time 2
employment adaptation variables were determined. Next, difference scores for each
adaptation type were calculated. Finally, a paired t-test was utilized to examine Type I
(maintain, but alter work), Time 1 and 2 differences, while the McNemar test (paired
contingency tables) was used to evaluate the signiﬁcance of differences for Type II (leave
of absence) and Type III (quit/retire) employment adaptations.

The caregivers in the study reported full-time (67.8%) or part-time (30.5%)
employment status prior to assuming the caregiver role. At Time 1, the percentage of

full-time employees dropped to 43.2% and the proportion of part-time workers declined

to 21.2%. At Time 2, some caregivers (7.2%) reported additional employment changes

(Figure 4).

95

 

....................

 

 

 

.................

....................................................

 

.....................................................................
...................................................................
.....................................................................

..............................

 

 

 

 

 

 

 

 

 

...................................................
....................................................................................
.................................................

 

 

 

 

 

Employed Employed Quit Leave of Retired
Full Time Part Time Working Absence

Figure 4. Time 2 Employment Status Changes (N=119)

Also, work days lost due to caregiving responsibilities varied between pre Time
1 and between Time 1 and Time 2. At Time 1, caregivers recounted an average of 8.53
(SD = 17.52) lost work days (in the past three months); three months later (Time 2),
caregivers reported a mean of 4.83 (SD = 15.96) for missed work days because of
eldercare obligations (during the past three months). A paired sample t-test (two-tailed)
revealed that the differences between the time periods were not signiﬁcant.

The mean for the index of Time 1, Type I adaptations (X = 1.75, SD = 1.58)
declined for Time 2, Type I adaptation (X = 1.07, SD = 1.18). Thus, the mean dif-
ference score (0.68) for Type I adaptation suggested less Type I adaptation at Time 2.
The paired t-test conﬁrmed signiﬁcant differences (two-tail p = .000) between the time

periods. The frequencies for Type II and Type III adaptations also dropped, implying

96

less use of those options during that period. Taking a leave of absence fell 11.0
percentage points (from 18.6% to 7.6%) and choosing to quit or retire from work
declined 3.1 percentage points (from 16.5% to 13.4%). The McNemar Test conﬁrmed
signiﬁcant differences (p = .002) for Type II, but not Type III adaptations between the
time periods.

Figure 5 contains the frequencies for the components comprised in the index of
Type I adaptation (arrive late/ leave early, miss work without pay, take sick/personal
days, change work hours, refuse a job/promotion, kept from looking for a job/better job,
and other work effects) and the frequencies for Type 11 (leave of absence) and Type III
(quit/retire) adaptation at Time 1 and Time 2. The results revealed that each adaptation

type occurred at both Time 1 and Time 2.

97

2.23323 EoE>oEEm .2 3.28339.“— .m 959“.

_m>m_ 00. 3 3899.6 .cmo__.c9m u .
ZO:<E<Q< .zO:<E<Q<

:_md>b __wa>H .anh<ba<o<._md>b
202E poo _ozwm .mSOI
x5>> bag: 90.. 8.6805 x53 .35 _mcoﬂmd 3mm 305:5 .>:mm 9mm:
550 So: 58. 30.. mwnﬁm mocmco 22m mxmp «to; $3 \w_m_ w>_:<

 

..................................

..................................

..................................

..................................

\\

..................................

..................................
.......

.....................

\
V

.........................................

..................................

_ .................... m m m m
m m \ m ...... m .....................
\ \
m

 

 

 

 

 

 

 

 

 

 

 

......

 

 

 

 

 

 

 

.......

 

 

 

 

.......
uuuuuuuuuuuuuuuu

 

 

 

 

 

 

 

 

 

 

 

 

 

.............. . ..... m ..... - mm ..-

.................................................... maize..- 9.

won:

 

 

 

 

 

 

 

 

 

 

 

 

 

oCHZ m. oEF. ommlz roEﬁE

 

 

 

om

98

Hla A variety of Type I employment adaptations occurs at both Time 1 and
Time 2.

The results supported this hypothesis. Caregivers reported a variety of
employment adaptations (Type I index) at both Time 1 and Time 2 (Table 3). For both
time periods, the most frequently recounted adaptation was arrive late! leave early (38.1%
at Time 1; 23.5% at Time 2), while refuse job/promotion was the least often noted

(18.6% at Time 1; 10.9% at Time 2).

Hlb There are fewer Type I adaptive behaviors at Time 1 than at Time 2.
The results did not support this hypothesis. Descriptive statistics suggested less
Type I adaptation at Time 2 (Time 1, Type I X = 1.75; Time 2, Type I X = 1.07) than
at Time 1. The paired t-test conﬁrmed signiﬁcant differences (p = .000) between
Time 1, Type I and Time 2, Type I adaptation. While signiﬁcant, the differences were

not in the expected direction.

ch There is less Type II adaptation at Time 1 than at Time 2.
The results did not support this hypothesis. Descriptive statistics suggested more
Type II adaptation at Time 1 than at Time 2 (Time 1 = 18.6%; Time 2 = 7.6%). The

McNemar test conﬁrmed signiﬁcant differences (p = .002), but, as with Type I

adaptation, not in the expected direction.

 

 

99

Hld There is no difference between Type III employment adaptation at Time
1 and Type III adaptation at Time 2.

The results supported this hypothesis. Descriptive statistics suggested more Type
III adaptation at Time 1 than at Time 2 (Time 1 = 16.5%; Time 2 = 13.4%).

However, the McNemar test did not conﬁrm signiﬁcant differences between the time

periods.

Research Question 2 What is the probability that Time 1, Type I employment
adaptation (maintain, but alter work) influences Time 2,
Type II (leave of absence) and Type III (quit/retire)
employment adaptations?

Logistic regression determined the probability that Type I adaptation at Time 1
inﬂuences Time 2, Type II and Type III employment adaptations. The results of the
procedure helped establish whether the model that included the variable in question (Time
1, Type I adaptation) told more about the outcome variables (Time 2, Type II or Type
III adaptations) than a model that did not include the variable (model with only the
constant).

For Time 2, Type II and Type III adaptations, the -2 Log Likelihood (Type II,

p = .999; Type III, p = .744) was not signiﬁcant. The addition of Time 1, Type I
adaptation did not indicate signiﬁcant improvement, as indicated by the chi-square
statistic, to the model with only the constant term for either Time 2, Type II (p = .618)

or Type III (p = .812) employment adaptations. A partial correlation with the variables

in both models was zero. Table 1 includes the logistic regression results for the

inﬂuence of Time 1, TypeI employment adaptation on Time 2, Type II and Type III adaptations.

100

Table 1. Probability of Time 1, Type I Influencing Time 2, Type II and Type 111
Employment Adaptations

 

 

Variable b Wald DF Sig
Time 2,
Type II
(N =98)
Time 1, .113 .252 1 .616
Type I
Constant -2.527 17.403 1 .000
Time 2,
Type III
(N =99)
Time 1, .043 .057 1 .811
Type I
Constant -1.733 14.359 1 .000

 

Results from logistic regression provided the information to answer the following

hypothesis.

H211 The use of Time 1, Type I adaptation inﬂuences the use of Time 2, Type II
and Type III adaptations.

The results did not support this hypothesis. The logistic model that included Time

1 , Type I adaptation told no more about Time 2, Type II or Type III adaptations than a
model with only the constant. The insigniﬁcant results indicated that Time 1, Type I
employment adaptation did not inﬂuence the use of either Time 2, Type II or Type HI

adaptations.

101

Research Question 3 What is the relationship of selected caregiver and care
recipient characteristics to each type of employment
adaptation at Time 1 and Time 2?

caregiver age

relationship to the care recipient

caregiver marital status

caregiver education

caregiver occupation

caregiver household income

employment status (current part-time or full—time)
co-residence status (now and before)

care recipient age

care recipient gender

'-"=".=‘<z° rep-cues?!»

Pearson correlations (for all adaptation types and continuous variables), oneway
ANOVA (for Type I and categorical variables), and contingency tables (for Types II and
III and categorical variables) were used to answer this research question and the related
hypotheses. Variables signiﬁcantly correlated with Type I (maintain, but alter work)
adaptation at Time 1 included caregiver age, household income, and education. No
caregiver/care recipient characteristic variable displayed a signiﬁcant relationship with
Type I adaptation at Time 2.

For Type 11 (leave of absence) adaptation, only one variable (caregiver education)
was signiﬁcant at Time 1; none were signiﬁcant for Time 2. Caregiver household income
correlated signiﬁcantly with Time 1, Type III (quit/ retire) adaptation, while caregiver
education showed a signiﬁcant relationship with Time 2, Type III adaptation. Table 2
displays the Time 1 and Time 2 correlation results for selected caregiver/care recipient

characteristics and employment adaptation.

102

Table 2. Correlations of Selected Caregiver/Care Recipient Characteristics with
Employment Adaptation

 

. __ . Caregiver . . .
Time 1 (N —236) Caregiver Household Caregiver Care Recrptent

 

Time 2 (N = 119) Age Income Education Age
Type I
Time 1 -.344*** .205" .133* -.025
Time 2 .001 -.043 .096 -.001
Type II
Time 1 .011 -.148 -.182* -.l30
Time 2 .090 -.055 -.125 .036
Type 111
Time 1 -.113 -.157* -.082 -.034
Time 2 -. 127 ~.084 -.222* -.032

 

p (2-tailed) * <.05; ** <.01; *** <.001

At Time 1, oneway ANOVA results indicated signiﬁcant differences for four of
the seven categorical variables and Type I adaptation. These included caregiver
relationship to care recipient, current employment status, co-residence (now) with the co-
recipient, and care recipient gender. There were no signiﬁcant differences noted with
Type I and any of the caregiver/care recipient variables at Time 2. Tables 3 (Time 1)
and 4 (Time 2) present the oneway ANOVA outcomes for selected caregiver/care

recipient characteristics and Type I employment adaptation.

103
Table 3. Time 1 Oneway ANOVA by Type I for Selected Caregiver/Care Recipient

 

 

Characteristics
Variable (N =236) SS DF MS F Sig F
Relationship to 52.460 2 26.230 11.482 .000*
Care Recipient+
Marital Status .172 1 .172 .069 .793
Caregiver Occupa- .964 2 .482 .194 .824
tion+ + (n= 190)
Current FT/PT 9.439 1 9.544 4.130 .044*
Employment
Status (n= 151)
Co-residence 12.282 1 12.282 5.020 .026*
(now)
Co-residence .922 1 .922 .392 .532
(before)
(n = 161)
Care Recipient 37.139 1 37.139 15.870 .000*
Gender

 

Note: * Signiﬁcant differences between groups.
+ Tukey’s HSD revealed daughters/daughters-in-law signiﬁcantly different
from other groups at .05 level.
+ + Tukey’s HSD revealed no signiﬁcant differences between groups at .05
level.

104

Table 4. Time 2 Oneway ANOVA by Type I for Selected Caregiver/Care
Recipient Characteristics

 

 

Variable (N = 119) SS DF MS F Sig F
Relationship to 1.916 2 .958 .688 .505
Care Recipient“ +

Marital Status .051 1 .051 .037 .848
Caregiver Occupa- .431 2 .215 . 155 .857

tion (n=89)+ +

Co-residence .253 1 .25 3 . 169 .682
(before) (n = 83)

Care Recipient 2.363 1 2.363 1.716 .193
Gender

 

Note: ‘H Tukey’s HSD revealed no signiﬁcant differences between groups at .05
level.

Contingency tables indicated signiﬁcant results for two categorical caregiver/care
recipient characteristics (caregiver current employment status, p = .000; co-residence
now, p = .010) and Time 1, Type II adaptation and one (caregiver current employment
status, p = .000) with Time 1, Type III adaptation. There were no signiﬁcant Time 2,
Type II or Type III relationships. Table 5 presents the contingency table results for Time
1 and Time 2, Type II and Type III employment adaptations. The speciﬁc approach and
discussion for each hypothesis about caregiver/care recipient characteristics and

employment adaptation follow.

105

Table 5. Contingency Tables by Type II and Type III for Selected Caregiver/Care
Recipient Characteristics

 

 

Time 1 Time 2
Variable (N =236) (N = 119)
x2 x2
Type II
Family Relationship .263 .791
(1‘1, n=177; T2 n=98)
Marital Status .900 .759
(Tl, n=177; T2, n=98)
Caregiver Occupation .224 .227
(TI, n=l67; T2, n=85)
Current Employment Status .000 NA
(T1, n=177)
Co-residence (now) .010 NA
(Tl, n=177)
Co-residence (before) .611 .713
(T1, n=112; T2, n=66)
Care Recipient Gender .884 .168
(T1, n=177; T2, n=98)
Type III
Family Relationship .146 .829
(T1, n=233; T2, n=98)
Marital Status .529 .456
(Tl, n=233; T2, n=99)
Caregiver Occupation .374 .931
(Tl, n=187; T2, n=85)
Current Employment Status .000 NA
(1‘1, n=233) ‘
Co-residence (now) .087 NA
(Tl, n=233)
Co-residence (before) .444 .072
(T1, n=158; T2, n=67)
Care Recipient Gender .388 .776

¥ (Tl, n=233; T2, n=99)
Note: NA At Time 2, caregivers did not indicate their current employment status or co-
residence now.
D < .05

106

H3a Caregiver age relates inversely to each type of employment adaptation.

Time 1 results supported this hypothesis for Type I adaptation, but not for Type
II or Type HI adaptations. Correlations were used to assess the relationship of caregiver
age with each type of employment adaptation (Table 2). There was a signiﬁcant negative
relationship with Time 1, Type I adaptation (r = -.344, p = .000) and caregiver age,
but not with Type II (r = .011) or Type III (r = .118) employment adaptations.

Time 2 outcomes did not support this hypothesis. At Time 2, there were no
signiﬁcant inverse relationships with caregiver age and Type I (r = .001), Type 11 (r =

.090), or Type 111 (r = -.127) employment adaptations.

H3b Employment adaptation differs by caregiver relationship to care recipient.
The results supported the hypothesis for Time 1, but not Time 2, Type I
adaptation. The results did not sustain the hypothesis for Type II and Type III
adaptations at either time period. Analysis using oneway ANOVA (Type I) and
contingency tables (Type II and Type 111) provided results to examine the differences
between caregiver/care recipient relationships (spouse, daughter/daughter—in—law, other)
and employment adaptation. Signiﬁcant group differences (p = .000) were revealed for
Time 1, Type I adaptation (Table 3). Additional testing with Tukey’s Honestly
Signiﬁcant Differences method showed daughters/daughters-in-law to be different from
spouse and other caregivers for Type I adaptation. However, this signiﬁcant difference
did not persist at Time 2 (Table 4). Chi-square analysis of contingency table results
(Table 5) did not indicate a signiﬁcant association for spouse, daughter/daughter—in-law,

Or other family caregivers for either Time 1 or Time 2, Type II or Type III adaptations.

107

H3e Employment adaptation differs for married and unmarried caregivers.

The results did not support this hypothesis at Time 1 or Time 2 (Tables 3-5).
Oneway ANOVA (Type I) and contingency tables (Type II and Type III) provided results
to determine whether there was a difference in any type of employment adaptation
between married and unmarried caregivers. There were no signiﬁcant differences

between married and unmarried caregivers for Type I, II, or HI employment adaptations.

H3d Caregiver education relates inversely to employment adaptation.

The correlation outcomes supported this hypothesis only for Time 1, Type 11 (r
= -.182, p = .015) and Time 2, Type III (r = -.222, p = .027) adaptations. While
there was an inverse relationship between education and Type II adaptation at Time 2‘ (r
= —.082) and education and Type III adaptation at Time 1 (r = -.125), the relationships
were not signiﬁcant. At Time 1, there was a signiﬁcant relationship with education and
Type I adaptation, but not in the direction expected (r = .133, p =.042). At Time 2,
the relationship of education with Type I adaptation also was positive, but insigniﬁcant

(r = .096). Thus, the outcomes (Table 2) offered limited support for the hypothesis.

H3e Employment adaptation differs with caregiver occupation.

The results did not support this hypothesis. Occupation made no signiﬁcant
difference in the use of Type I, Type II, or Type III adaptations at either Time 1 or Time
2 (Tables 3-5). Oneway ANOVA (Type I) and contingency tables (Type II and Type III)

provided outcomes to assess differences between employment adaptation and caregiver

108

occupation (professional, management, administrative; sales, clerical, technical; or

other).

H3f Caregiver household income relates inversely to employment adaptation.

The correlation outcomes (Table 2) supported the hypothesis for Type III
adaptation (r = -.157, p = .027) at Time 1. However, it did not support the hypothesis
for Time 1, Type I or Type II adaptations or for any adaptation type at Time 2.

At Time 1, income had a signiﬁcant (but positive, not inverse) relationship
(r = .205, p = .003) with Type I adaptation, and an inverse (but not signiﬁcant)
relationship with Type II (r = -. 148). While there were inverse relationships with
caregiver household income and each adaptation type at Time 2 (Type I, r = -.043; Type

11, r = -.055; Type 111, r = -.084), they were not signiﬁcant.

H3g Employment adaptation differs with full-time and part-time employment
status.

The results supported this hypothesis for Time 1 and all adaptation types.
Employment status was signiﬁcantly different for Time 1, Type I (p = .044), Type II
(p = .000), and Type III (p = .000) adaptations. Oneway ANOVA (Type I) and
contingency tables (Type II and Type III) provided the techniques (Tables 3-5) to
determine whether there were differences and associations between full-time and part-
time employment status and employment adaptation. At Time 2, caregivers were not

asked to indicate their current employment status.

109

H3h Employment adaptation differs if caregiver and care recipient reside together.

The results partially supported this hypothesis for Time 1. There were signiﬁcant
differences with Time 1, Type I (p = .026) and Type II (p = .010) adaptations between
caregivers who currently lived with their care recipient and those who did not.
However, there was no signiﬁcant difference (p = .088) between the groups with Time
1, Type III adaptation. Oneway AN OVA (Type I) and contingency tables (Type II and
Type III) contributed the results (Tables 3-5) to determine whether there were differences
and associations between caregiver/care recipient co-residence and residence apart and
employment adaptation. Caregivers were not asked about current co-residence status
at Time 2. At both Time 1 and Time 2 there were no signiﬁcant differences based on

co-residence before caregiving with any adaptation type.

H31 Care recipient age relates positively to caregiver employment adaptation.
The results (Table 2) did not support this hypothesis for Time 1 or Time 2.
There were insigniﬁcant negative correlations with care recipient age and Type I (r = -
.025), Type II (r = -. 130), and Type 111 (r = -.034) adaptations at Time 1. Likewise,
at Time 2, there was an insigniﬁcant negative relationship with care recipient age and
'Type I (r = -.001) and Type III (r = -.032) adaptations. At Time 2, the relationship
of care recipient age to Type II employment adaptation was positive, but insigniﬁcant

(r = .036).

110

H3] Care recipient gender makes no difference in employment adaptation.

This hypothesis was partially supported at Time 1 and completely supported at
Time 2. Gender was not signiﬁcantly associated with Time 1, Type II or Type III
adaptations; however, gender did make a signiﬁcant difference (p = .000) in Time 1,
Type I adaptation. At Time 2, there were no signiﬁcant differences with gender and the
three employment adaptation types. Oneway ANOVA (Type I) and contingency tables

(Type II and Type 111) provided results (Tables 3-5) to verify this hypothesis.

Research Question 4 How much variation in each type of employment adapta-
tion at Time 1 and Time 2 do selected signiﬁcant care-
giver and care recipient characteristics explain?

caregiver age

relationship to the care recipient
caregiver education

caregiver household income
co—residence status (now)

care recipient gender

W99.“ 9‘!”

Bivariate associations among the variables and each adaptation type were
examined to eliminate from multivariate analyses potential explanatory factors that did
not have signiﬁcant associations. Stepwise multiple regression provided results to answer
this research question and the related hypotheses for Time 1, Type I (maintain, but alter
work) adaptation, while two-group discriminant analysis (Wilks’s method) was used to
determine whether selected caregiver and care recipient characteristics could differentiate
those caregivers who used Time 1, Type 11 (leave of absence) from those who did not.
An independent t-test provided results to assess the signiﬁcance of the only caregiver/care

recipient variable displaying a signiﬁcant relationship with Time 1, Type III adaptation.

111

Results from preliminary analyses of Time 2 data (correlations, oneway AN OVA,
contingency tables) did not suggest the need for additional regression or discriminant
analyses for either Type I or Type II adaptations. However, as with Time 1 data, an
independent t-test was used to evaluate the importance of the only caregiver/care
recipient characteristic variable showing a signiﬁcant relationship to Time 2, Type III
employment adaptation.

Employment status, in spite of signiﬁcant differences between part-time and full-
time employment and each adaptation type, was not used for any additional analysis.
This was done in order that the other signiﬁcant caregiver/care recipient characteristics
could be assessed for their unique and relative inﬂuence on employment adaptation.

Variables with signiﬁcant correlations (caregiver age, education, and household
income) or oneway ANOVA results (relationship to care recipient, care recipient gender,
and co-residence now) were used in the Time 1 regression analysis. Variables eliminated
included caregiver marital status, occupation, co-residence status (before), and care
recipient age. Three variables, caregiver age, caregiver relationship to care recipient, and
care recipient gender explained 19% of the total variance (R2 = .191, p = .000) for
Type I employment adaptation at Time 1.

Table 6 displays the stepwise regression results for Time 1, Type I adaptation and

selected caregiver/care recipient characteristics.

1 12
Table 6. Time 1 Regression of Significant Caregiver/Care Recipient Characteris-

 

 

tics and Type I Adaptation

Variable b B T Sig t
Constant 3.475 5.45 1 .000
Caregiver age -.051 -.364 -4.368 .000
Relationship to 1. 196 .366 3.166 .002
care recipient
(daughter)
Relationship to 1.568 .454 2.964 .003
care recipient
(Spouse)
Care recipient -.685 -.213 -2.195 .029
gender
R2 .191
F 12.139
p-value .000
N 210

 

Note: Variables in equation = p < .05

As with Type I adaptation, only variables with signiﬁcant relationships
(correlations or contingency tables) with Type II adaptation were used in the discriminant
analysis. For Time 1, Type II, this included signiﬁcant correlation (caregiver education)
and signiﬁcant contingency table (co-residence now). results.

In discriminant analysis, as in multiple regression, the emphasis is on analyzing
variables together, not one at a time, so that important relationships can be noted.

Norusis (1990c) claims that the linear combination of the variables, summarized in a
single index, serves as the basis for allocating cases to groups. The unstandardized

discriminant function coefﬁcients are the "multipliers of the variables when they are

113
expressed in the original units” (Norusis, 1990c, p. B-14). The standardized canonical

discriminant function coefﬁcients show the relative contribution of the variables to the
equation. According to Klecka (1980), the larger the magnitude (regardless of the sign),
the greater the variable contribution.

Chi-square analysis of the canonical discriminant function for Time 1, Type II
employment adaptation and signiﬁcant caregiver/care recipient characteristics (caregiver
education and co-residence now) indicated signiﬁcance for the overall model effectiveness
(p = .004). Discriminant analysis correctly grouped almost 59% (58.76) of the cases.
Although a statistically signiﬁcant result, practical signiﬁcance in the present study is
considered to be 60% or above.

Table 7 contains the canonical discriminant functions, Table 8 includes classiﬁca-
tion results, and Table 9 incorporates the unstandardized and standardized canonical
discriminant function characteristics for signiﬁcant selected caregiver/care recipient
characteristics and Time 1, Type II employment adaptation.

Table 7. Canonical Discriminant Functions for Selected Significant Caregiver/ Care
Recipient Characteristics and Time 1, Type II Adaptation

 

 

Eigen- Canonical Wilks’s .
value Correlation Lambda X2 DP Slg'
Type II .065 .247 .939 10.964 2 .004

(N=177)

114

Table 8. Discriminant Function Classification for Selected Significant Care-
giver/Care Recipient Characteristics and Time 1, Type II Adaptation

 

 

Predicted Predicted
Time 1, Type II
(N = 177)
Actual Group Cases LOA No LOA
LOA 44 (100%) 31 (70.5%) 13 (29.5%)
No LOA 133 (100%) 60 (45.1%) 73 (54.9%)
Percent of cases 58.76%

correctly grouped

 

Table 9. Canonical Discriminant Function Coefficients for Selected Significant
Caregiver/ Care Recipient Characteristics and Time 1, Type II Adaptation

 

 

Unstandardized Standardized
Type II (N = 177)
Caregiver education -.266 -.636
Co-residence now 1.459 .694

 

For Time 1, Type III adaptation, only one variable displayed a signiﬁcant
relationship (caregiver household income). Because there was only one variable with a
signiﬁcant relationship to Time 1, Type III adaptation, an independent t-test (two-tailed),
not discriminant analysis, was used to Check the signiﬁcance of mean differences in
caregiver household income with Time 1, Type III adaptation. The independent t-test
conﬁrmed signiﬁcant differences (two-tail p = .023) related to the income of those

caregivers who chose to quit or retire at Time 1 (less income) and those who did not.

115

Similarly, at Time 2, just one variable (caregiver education) showed a signiﬁcant
relationship with Type III adaptation. An independent t-test conﬁrmed signiﬁcant
differences (two-tail p = .027) related to the education of those caregivers who elected
to quit or retire at Time 2 (less education) and those who did not. Table 10 contains the
independent t-test outcomes for Time 1, Type III adaptation and Table 11 includes results
for Time 2, Type HI adaptation.

Table 10. Independent t-test for Caregiver Household Income and Time 1, Type 111
Employment Adaptation

   

Type 111 Number of Cases Mean SD

 

 

Time 1 38 530263.16 316199.66
Quit/Retire

Time 1 171 837318.71 817387.01
No Quit/Retire

Pooled Variance t value 2.29 DF 207 2-tail probability
Estimate .023

 

Table 11. Independent t-test for Caregiver Education and Time 2, Type III
Employment Adaptation

 

 

 

_ r m
Type III Number of Cases Mean SD
Time 1 16 12.19 1.83
Quit/Retire
Time 1 83 13.63 2.43
No Quit/Retire
Pooled Variance t value 2.25 DF 97 2-tail probability

Estimate .027

116

In summary, some caregiver/care recipient characteristic variables offered
signiﬁcant explaining power and differentiation for each adaptation type. Three variables
(caregiver age, family relationship, care recipient gender) predicted Time 1, Type I
adaptation (R2 = .191), two variables (caregiver education, co-residence now)
contributed to the discriminant function of Time 1, Type II adaptation (59% correct
grouping), and one variable differentiated Type III adaptation at both Time 1 (caregiver
household income) and Time 2 (caregiver education).

The absence of signiﬁcant relationships for caregiver and care recipient
characteristics with Time 2, Type I and Type II adaptations implies that they offer scant
explanatory power for those adaptation types at that time. The speciﬁc approach and
discussion for each hypothesis about caregiver/care recipient characteristics and

employment adaptation follow.

H4a Caregiver age explains more variation in employment adaptation at Time 1
and Time 2 than do other caregiver/care recipient characteristics.

The results did not support this hypothesis for either Time 1 or Time 2. While
caregiver age did explain variation in Time 1, Type I adaptation (Table 6), it was one
of three variables (caregiver age, relationship with care recipient, and care recipient
gender) to do so. Of those variables, family relationship (daughter/daughter-in-law,
ﬂ = .366; spouse, 3 = .454) had a slightly higher relative contribution to the equation
than did caregiver age (B = .364).

The other caregiver/care recipient characteristic variables (caregiver income,

education, and co-residence with the care recipient) showing signiﬁcant relationships with

l 17
Type I adaptation did not meet criteria for the ﬁnal regression equation (Table 6). While

this suggests that caregiver age offered relatively more explaining power for Type I
adaptation than those variables, it cannot be said that caregiver age explains more
variation than any of the caregiver/care recipient characteristics for Time 1, Type I
adaptation.

Likewise, the results did not support this hypothesis for Time 1, Type II or Type
III adaptations. Caregiver age did not show a signiﬁcant correlation with Type II or Type
III adaptations at Time 1 and was not used for additional analyses. At Time 2,
correlations revealed no signiﬁcant relationships with Type I, Type H, or Type III

adaptations. Thus, additional analyses were not done.

H4b Care recipient age and care recipient gender explain little variation in
employment adaptation at Time 1 and Time 2.

The results partially supported this hypothesis. Care recipient age had no
signiﬁcant relationship with Time 1, Type I adaptation; however, care recipient gender
was one of three signiﬁcant variables in ﬁnal regression equation (Table 6). At Time 2,
neither care recipient age or gender showed a signiﬁcant relationship with Type I
adaptation; thus, they were not used in subsequent regression analysis. This suggests
support for the hypothesis for Type I adaptation at Time 2.

The outcomes also intimate support for the hypothesis for Type II and Type III
adaptations at Time 1 and Time 2. Neither care recipient age nor gender showed
signiﬁcance at either time period (implying scant relationship with either adaptation type)

and were not used for subsequent Type II and Type III discriminant analysis.

118
Research Question 5 Was there a reported change in the amount of

help from family members between Time 1 and

Time 2?
Frequencies provided results to respond to this question and the following
hypothesis. Some caregivers reported changes (36.1%) in the amount of family help

while others did not (63.9%).

HSa The majority of caregivers will report that the level of family help remained
”about the same” at Time 2.

The results supported this hypothesis. Most caregivers (63.9%) who remained in
the study at Time 2 noted that the amount of help from their family remained about the
same three months later. However, for some caregivers, the amount of help increased

(14.3%), while for others it decreased (21.8%).

Research Question 6 What is the relationship of family support to
each type of employment adaptation at Time 1
and Time 2?

a. number in network
b. amount of help provided
c. frequency of assistance given

Caregivers reported an average of two relatives living within 50 miles (X = 2.27,

SD

2.10). The mean for the amount of help provided at Time 1 was 2.93
(SD = 3.52, range 0 - 19), while the frequency of help (Time 1, X = 1.74, SD = .48;
Time 2, X = 1.66, SD =.57) ranged from rarely or none of the time (1) to almost all
of the time (4).

Pearson correlations provided outcomes that indicated a signiﬁcant relationship

with the amount of family help (r = .223, p < .001) and Time 1, Type I adaptation, but

119
not with Time 1, Type II or Type III adaptations. However, no other family variable

displayed a signiﬁcant relationship to any employment adaptation type at Time 1. At
Time 2, the family variables available for study (number in the family network,
frequency of family help) did not show signiﬁcant relationships with any adaptation type.
Table 12 contains correlation results for the family support variables and employment

adaptation.

Table 12. Correlations of Family Support Variables with Fanployment Adaptation

 

 

 

 

Time 1 (N =236) Family Members Frequency of Amount of
Time 2 (N = 1 19) within 50 Miles Family Aid Family Help
Type I
Time 1 .041 -.031 .223***
Time 2 -.098 .138 NA
Type 11
Time 1 -.086 -.048 -.096
Time 2 -.O29 .039 NA
Type 111
Time 1 .120 -.066 .032
Time 2 .188 -.140 NA

 

p (2-tailed) * <.05; ** <.01; *** <.001
Note: NA Time 2 data do not include comparable information for amount of family

help.

The speciﬁc approach and discussion for each hypothesis about family support and

employment adaptation follow.

120

H6a The number in the family network has no relationship to employment
adaptation at Time 1 and Time 2.

The results supported this hypothesis. There were no signiﬁcant relationships
between the number of caregiver family members living within a 50 mile radius of the
caregiver and Type I (r = .041), Type 11 (r = -.086), and Type III (r = .120)
adaptations at Time 1 or Time 2 (Type I, r = -.098; Type II, r = —.029; Type III,

r = .188).

H6b The amount of help provided by the family relates inversely to employment
adaptation at both Time 1 and Time 2.

The results did not support this hypothesis. At Time 1, there was an inverse, but
not signiﬁcant relationship between Type II (r = -.096) adaptation and the amount of
family help. There was a signiﬁcant, positive relationship with Type I (r = .223,
p = < .001) and an insigniﬁcant correlation with Type III adaptation (r = .032). At
Time 2, caregivers were not asked to indicate the amount of help from each network

member.

H6c The frequency of family help relates inversely to employment adaptation at
both Time 1 and Time 2.

The results did not support this hypothesis for either time. At Time 1, there were
inverse, but insigniﬁcant, relationships (Type I, r = -.031; Type 11, r = -.048; Type III
r = -.066) between the frequency of family help and each employment adaptation type.
At Time 2, there were positive, but insigniﬁcant relationships between Type I (r = . 138)

and Type 11 (r = .039) employment adaptations and frequency of help. While Time 2,

121
Type III adaptation had an inverse relationship (r = -. 140) with frequency of family

help, it was not signiﬁcant.

Research Question 7 How much variation in each type of employment adapta-
tion at Time 1 and Time 2 does family support explain?

a. number in network

b. amount of help provided

c. frequency of assistance given

As in Question 4, stepwise multiple regression provided results to answer this

research question and its hypotheses for Type I adaptation. While only one variable
(amount of family help) had a signiﬁcant correlation with Time 1, Type I adaptation, all
three family variables were used in the regression equation because of the importance of
family in the conceptual model for this study. Amount of family help explained only ﬁve
percent (R2 = .047, p = .001) of the variance in Time 1, Type I adaptation. The other
variables (frequency of family help and number in family network) did not meet the

criteria for the regression equation. Table 13 contains the Time 1 regression results for

the family variables.

122

Table 13. Regression for Family Support Variables and Time 1, Type I Employ-
ment Adaptation

 

 

 

Variable b is r Sig ;

 

Constant 1.460 10.919 .000
Amount of family .098 .217 3.351 .001
help

R2 .047

F 1 1.227

p-value .001

N 229

 

Note: Variables in equation = p < .05

No family variables showed signiﬁcant correlations with Time 1, Type II or Type
III adaptations, but, as with the regression analysis, they were used for discriminant
analyses due to their importance in the conceptual model of the present study. Two-
group discriminant analysis (Wilks’s method) provided results for Types II and III
employment adaptation. Chi-square analyses of the canonical discriminant function did
not show signiﬁcance (Type II, p = .194; Type HI, p = .102) for the overall

effectiveness of either model (Table 14).

123

Table 14. Canonical Discriminant Functions for Family Support Variables and
Time 1, Type II and Type III Adaptations

 

Eigen— Canonical Wilks’s x2

 

value Correlation Lambda DP Slg'
Type 11
Time 1 .010 .100 .990 1.685 1 .194
(N = 171)
Type III
Time 1 .021 .142 .980 4.559 2 .102
(N =226)

 

The absence of an effective model explains the poor classiﬁcation results,
especially for Time 1, Type II adaptation. The discriminant function correctly grouped
only 55% (54.80%) of the cases for Type II adaptation. For Time 1, Type III, correct
classiﬁcation results increased slightly, to almost 57% (56.64%). Table 15 contains the
discriminant analysis classiﬁcation outcomes for family support variables and Type II and

Type III adaptations.

124

Table 15. Discriminant Function Classification Results for Family Support
Variables and Time 1, Type II and Type III Adaptations

   

 

 

Predicted Predicted

 

 

Type II

(N = 171)

Actual Group Cases LOA No LOA

LOA 44 (100%) 28 (63.6%) 16 (36.4%)

No LOA 133 (100%) 64 (48.1%) 69 (51.9%)

Percent of cases 54.80%

correctly

grouped

Type III

(N =226)

Actual Group Cases Quit/Retire No Quit or
Retire

Quit/Retire 38 (100%) 16 (42.1%) 22 (57.9%)

No Quit/Retire 188 (100%) 76 (40.4%) 112 (59.6%)

Percent of cases 56.64%
correctly

grouped

Clearly, family support offered little explanatory power for Time 1, Type I
adaptation (R2 = .047). While the discriminant analysis classiﬁcation results were better
than the 50% correct that might be expected by chance, neither Time 1, Type II (55%
correct grouping) nor Type III (57% correct grouping) adaptations indicated a strong

connection with the family support variables. Thus, the analyses did not yield either

125

statistically signiﬁcant results or the practically signiﬁcant outcomes (60%) identiﬁed by
this researcher for the present study.

At Time 2, caregivers were not asked comparable information about the amount
of family help. Instead, they were asked if the amount of help from family members
changed in the past three months (Research Question 5). Preliminary analyses of the
remaining Time 2 family variables (family network, frequency of aid) did not indicate
signiﬁcant relationships with any adaptation type. Thus, additional study was not done.
The speciﬁc approach and discussion for each hypothesis about family support and

employment adaptation follow.

H7a The amount of help and frequency of assistance provided explain more
variation in employment adaptation at both Time 1 and Time 2 than does the
number in the network.

The results partially supported this hypothesis for Time 1. The amount of family
help met criteria to enter the regression equation (the only family variable to do so);
thus, it explained more Type I variation than did the network number. However, the
frequency of family aid did not meet criteria for the regression equation; consequently,
it along with family network, explained little Type I variation.

Similar results occurred for Time 1, Type II adaptation. The amount of family
help was the only variable in the ﬁnal discriminant equation. Neither frequency of family
help nor family network met the tolerance level for the ﬁnal analysis; thus, neither

variable contributed to the equation. This ﬁnding suggested that the amount of family

help, but not the frequency of aid contributed to group differentiation.

126
The Time 1, Type III analysis provided the opposite outcome from that of Type

I or Type H investigation. Caregiver family network and the frequency of family
assistance met the criteria for the ﬁnal discriminant equation, while the amount of family
help did not. Thus, results supported the frequency of family assistance, but not the
amount of family help, to predict Time 1, Type III employment adaptation. Correlations
did not reveal signiﬁcant relationships with the family variables and any adaptation type
at Time 2; thus, additional study was not done. Table 16 contains the unstandardized
and standardized canonical discriminant function coefﬁcients for family characteristics
and Time 1, Type H and Type III adaptations.

Table 16. Canonical Discriminant Function Coefficients for Family Support
Variables and Time 1, Type H and Type HI Employment Adaptations

 

 

 

(N=171) (N=226)
Unstand- Standard- Unstand- Standard-
ardized ized ardized ized
Amount of family help .274 1.000 NA NA
Family members within 50 NA NA .420 .892
miles
Frequency of family help NA NA -1.123 -.545

 

Note: NA = Variable did not meet tolerance level to enter the equation.
Research Question 8 Do caregiving responsibilities disrupt normal work and
daily activities at both time periods?
Frequencies provided results to answer this research question and the following

hypothesis. The majority of caregivers reported disruption with usual activities (Time

127
1 = 88%; Time 2 = 87%), but some did not (Time 1 = 12%; Time 2 = 12%).

Differences in the reported disruption were not signiﬁcant.

 

 

 

 

 

 

 

35
DTime1N=236 Grime 2 N=119 Tip...
25 ifiizieiii iii???

...................

....................
....................
....................

....................

 

....................

....................
....................
...................
...................
....................
....................
....................
....................
....................
....................

....................

 

..................

 

 

...................

..........

....................

..........

....................

.........

.........................................

........................................

.........................................

.........................................

.......................................

.............................

.......................................

........................................

.......................................

.........................................

..............................

 

 

 

 

\\\\\\\\\\\\\\\\\\\\\

 

 

 

 

 

 

 

\\\\\\\\\\\\\\\

 

 

 

N

. ........ / ..........

Not at All A Little Somewhat A Great Deal

 

Figure 6. Disruption of Normal Work and Daily Activities because of
Caregiving Activities

H8a Caregiving responsibilities disrupt normal work and daily activities at both
time periods.

The results supported this hypothesis. Caregivers recounted disruption at both
time periods (Figure 6). Reports ranged from ”a little disruption" (Time 1 = 25%;

Time 2 = 21.0%) to ”a great deal of disruption" (Time 1 = 30.9%; Time 2 = 32.8%).

128

Research Question 9 Does the level of reported caregiver involvement change
from Time 1 to Time 2?

Frequencies provided results to answer this question and the following hypothesis.

Some caregivers reported changes (63 %) while others did not (38%).

H9a The majority of caregivers will report that they provide ”about the same”
amount of care compared to three months ago.

The results did not support this hypothesis. While 38% (37.8) reported ”about

the same," almost 40% (39.5%) reported less care and 23% (22.7) reported more care.

Research Question 10 What is the relationship of caregiver involvement
(activities and time) to each type of employment adapta-
tion at Time 1 and Time 2?

a. total involvement (ADL, IADL, HCA)

b. hours of physical care

0. hours of supervision

Pearson correlations were used to show the relationship between caregiver

involvement and each adaptation type. Caregiver total involvement and hours of physical
care correlated signiﬁcantly with Time 1, Type I adaptation; hours of supervision did
not. All involvement variables (total caregiver involvement, hours of physical care, and
hours of supervision) showed a signiﬁcant, positive relationship for Type H and Type HI
adaptations at Time 1. Time 2 ﬁndings demonstrated only one signiﬁcant correlation
with the three adaptation types, namely, hours of supervision with Type HI employment

adaptation. Table 17 contains correlations for Time 1 and Time 2 employment

adaptation and the caregiver involvement variables.

129
Table 17. Correlations of Caregiver Involvement with Employment Adaptation

     

Time 1 (N =226) Total Caregiver Hours of Physical Hours of

 

Time 2 (N = 119) Involvement Care Supervision
Type I
Time 1 .228*** .253*** .078
Time 2 .135 .115 .010
Type H
Time 1 .198" 250*“ .160“
Time 2 .011 .113 .085
Type IH
Time 1 .324*** 315*” .227***
Time 2 .113 -.043 343*"

 

p (2-tailed) * <.05; ** <.01; *** <.001

The means for the involvement variables varied with the time periods. The
average for Time 1 total involvement (23.06, SD = 15.16) declined at Time 2 (21.57,
SD 14.62). A paired t-test indicated the differences were signiﬁcant (two—tailed
probability = .026). Likewise, the mean for hours of physical care dropped from 3.76
(SD = 5.26) to 2.58 (SD = 4.21) at Time 2. A paired t-test indicated the differences
were signiﬁcant (two-tailed probability = .026). The average hours of supervision
increased from Time 1 (6.98, SD = 6.71) to 7.33 (SD = 7.11) at Time 2. However,
the paired t-test showed that the differences were not signiﬁcant. Figure 7 includes the

means for the caregiver involvement variables.

130

«east; 2080222.: 32:30 ..o. 2.32 .5 9.32m

30:22.6 EmoECQm n ..

3.: Ed wage a: $0 mmcmg sec omcmmv
222235 EEO .8625 LcoEm>_o>E and mmcmmv Smd mmcmmv 2&6 mmcmmv

53350: 28350: .92 <0: .._o<_ ..o<

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.................... O
3355‘ mmmmmmm\ mum-LEV mmmmmmm\ .....\\ 3355‘
mam are m if, it in. an: \
may ....... smegma-m: \ in. if \m . m
game mummy l ﬁfe El m
53323 \ ..HLJHH vmo Om . 3.3333\ m o vmmuom
.......... i Hexnxxw mvouom mummhw
ZNHOm :oiQm mum-”mm.- 33mm.-
m ............ .9
- -. xz.a;-w _vslom oevlom
mﬂmmm- W
. .......... 8
. .......... B
.ﬁﬁﬁﬁh U
fume
............. imam-m. smr
.................. «may». ow
moiuom \
OFOPHQW
mw—HZNmECD 0N

 

 

 

 

ammuz _ 958

 

 

 

131
The speciﬁc approach and discussion for each hypothesis about caregiver

involvement and employment adaptation follow.

H10a Total involvement (ADL, IADL, and HCA) relates positively to employment
adaptation.

The results supported this hypothesis for Time 1, but not for Time 2. At Time
1, total caregiver involvement showed a signiﬁcant positive relationship with each type
of employment adaptation (Type I, r = .228, p = .000; Type H, r = .198, p = .008;
Type HI, r = (.324, p = .000). However, Time 2 results revealed no signiﬁcant
relationships between total caregiver involvement and Type I (r = .135), Type H

(r = .011), or Type HI (r = .113) adaptations.

H10b Hours of physical care relate positively to employment adaptation.

The results supported this hypothesis for Time 1, but not for Time 2. Signiﬁcant
relationships were evident with each adaptation type at Time 1 (Type I, r = .253,
p = 000; Type 11 r = .250, p = 001; Type 111, r = .315, p .001). However, Time 2
results indicated no signiﬁcant relationship with Type I (r = .115), Type H (r = .113),

or Type HI (r = -.043) adaptations.

H10c Hours of supervision relate positively to employment adaptation.
The results partially supported this hypothesis for Time 1. Signiﬁcant relationships
were present with hours of supervision and Type H (r = 160, p = .035) and Type IH

(r = .227, p = .001) adaptations, but not with Type I adaptation (r = .078) at Time 1.

132
At Time 2, only Type III adaptation showed a signiﬁcant relationship (r = .343,

p = .001) with hours of supervision. Type I (1' = -010) and Type 11 (r -085)

adaptations did not have a signiﬁcant relationship with hours of supervision.

Research Question 11 How much variation in each type of employment adapta-
tion at Time 1 and Time 2 is explained by caregiver
involvement (activities and time)?

a. total involvement

b. hours of physical care

0. hours of supervision

As in question 4, bivariate associations among the variables and each adaptation

type were examined to eliminate from multivariate analyses potential explanatory factors
that did not have signiﬁcant associations. Stepwise multiple regression provided results
to answer this research question and the related hypotheses for Time 1, Type I (maintain,
but alter work) adaptation, while two-group discriminant analysis (Wilks’s Method) was
used to determine whether caregiver involvement could differentiate those caregivers who
used Time 1, Type 11 (leave of absence) and Time 1, Type HI (quit/retire) adaptations
from those who did not.

Total caregiver involvement and hours of physical care (the only variables
signiﬁcantly correlated with Time 1, Type I adaptation) were used in the regression
equation. However, only hours of physical care met the criteria for the equation and
explained 6% (R2= .064, p = .000) of the variance in Type I adaptation. At Time 2,
no caregiver involvement variable demonstrated a signiﬁcant relationship with Type I

adaptation. Table 18 contains the regression results for caregiver involvement and Type

I adaptation.

133
Table 18. Regression for Caregiver Involvement and Time 1, Type I Fanloyment

 

 

 

 

Adaptation

Variable b B T Sig t
Constant 1.494 1 1.918 .000
Hours of physical care .076 .253 3.925 .000
R2 .064

F 15.403

p-value .000

N 228

 

Variables in equation = p < .05

Each involvement variable displayed signiﬁcant relationships with Time 1, Type
H and Time 1, Type IH adaptations; thus, all were used in the discriminant analyses.
Chi-square analyses of the canonical discriminant function showed signiﬁcance for the
overall effectiveness of the model for both Type H (p = .001) and Type HI (p = .000)
adaptations. The discriminant function correctly grouped nearly 66% (65.90) of the
cases for Type H and almost 72% (71.56) for Type IH adaptations. Correlations and
contingency tables did not reveal signiﬁcant variables for Type H adaptation at Time 2;
thus additional study with discriminant analysis was not done. Table 19 contains the
canonical discriminant functions and Table 20 includes the discriminant classiﬁcation

results.

134

Table 19. Canonical Discriminant Functions for Caregiver Involvement Variables
and Time 1, Type H and Type HI Adaptations

   

Eigen-

 

value Correlation Lambda x2 DP Sig
Type 11
Time 1 .068 .252 .937 11.039 1 .001
(N = 171)
Type IH
Time 1 .137 .347 .880 28.229 2 .000
(N =223)

 

Table 20. Discriminant Rmction Classification for Caregiver Involvement Variables
and Time 1, Type H and Type HI Adaptations

 

 

 

 

 

Predicted Predicted
Type H (N = 171)
Actual Group Cases LOA No LOA
LOA 44 (100%) 16 (36.4%) 28 (63.6%)
No LOA 129 (100%) 31 (24.0%) 98 (76.0%)
Percent of cases 65.90%
correctly grouped
Type HI
(N =223)
Actual Group Cases Quit/Retire No Quit or

Retire

Quit/Retire 36 (100%) 22 (61.1%) 14 (38.9%)
No Quit/Retire 189 (100%) 50 (26.5%) 139 (73.5%)
Percent of cases 71.56%

correctly grouped

 

135

Only one variable (hours of supervision) showed a signiﬁcant relationship with
Time 2, Type HI adaptation. Because there was only one involvement variable with a
signiﬁcant relationship to Type IH adaptation, an independent t-test (two-tailed) was used
to check the signiﬁcance of mean differences in hours of supervision with Time 2, Type
IH adaptation. The independent t-test conﬁrmed signiﬁcant differences (two-tail
p = .001) related to the hours of supervision expended by those caregivers who chose
to quit or retire at Time 2 and those who did not. Table 21 presents the independent t-
test outcomes for Type III adaptation at Time 2.

Table 21. Independent t-test for Hours of Supervision and Time 1, Type HI
Employment Adaptation

 

 

 

Type HI Number of Cases Mean SD

Time 2 16 12. 19 7.61
Quit/Retire

Time 2 83 5.82 6.26

No Quit/Retire

Pooled Variance t value -3.60 DF 97 2-tail probability
Estimate .001

 

Caregiver involvement offered little explanatory power for Time 1, Type I
adaptation (R2 = .063). The Time 1 discriminant analyses classiﬁcation results were
more impressive, especially for Type III adaptation (72% correct grouping). Type H
adaptation (66% correct grouping) indicated less association with the caregiver
involvement variables. The speciﬁc approach and discussion for each hypothesis about

caregiver involvement and employment adaptation follow.

136

Hlla Hours of physical care explain more variation in employment adaptation than
total involvement or hours of supervision at both Time 1 and Time 2.

The results supported this hypothesis for Type I and Type H adaptations, but only
partially for Type HI adaptation at Time 1. Hours of physical care was the only variable
to meet criteria for the regression equation; thus, it explained more variation in Type I
adaptation than either total caregiver involvement or hours of supervision. However, even
though hours of physical care entered the equation, it contributed little (R2 = .064,
p = .000) toward explaining the variation in Type I adaptation at Time 1. Similar
results occurred with Type II adaptation. Hours of physical care was the only variable
that met tolerance levels for entry into the discriminant equation.

However, for Type HI adaptation, both hours of physical care and total caregiver
involvement met the criteria for the discriminant function, indicating that both variables
contributed to the differentiation of Type III adaptation at Time 1. Also, review of the
standardized coefﬁcients for the variables indicated similar results (Table 22).

Table 22. Canonical Discriminant Function Coefficients for Caregiver Involvement
Variables and Time 1, Type H and Type HI Employment Adaptations

 

 

Type II Type HI
(N=171) (N=223)
Unstand- Standard- Unstand- Standard-
ardized ized ardized ized
Hours of physical care .208 1.000 .107 .541
Caregiver total in- NA NA .386 .564

volvement

 

NA Variable did not meet equation tolerance level

137

The hypothesis that hours of physical care would explain more variation in
employment adaptation than total involvement and hours of supervision was not supported
by Time 2 results. No involvement variable displayed a relationship with Type I or Type
H adaptations. Hours of supervision was the only involvement variable with a signiﬁcant
relationship to Type IH adaptation. Thus, hours of supervision explained more variation

in Type IH adaptation than did hours of physical care.

Research Question 12 Which identified key variables from questions 4, 7, and
11 are the best predictors of employment adaptation at
Time 1 and Time 2?

Type I = caregiver age, relationship to care recipient,
caregiver gender, amount of family help,
hours of physical care

Type II = caregiver education, co-residence now,
amount of family help, hours of physical care
Type 111 = caregiver household income, family network,

frequency of family help, total caregiver
involvement, hours of physical care or care-
giver education and hours of supervision)

As in Questions 4, 7, and 11, stepwise multiple regression and discriminant
analysis provided results to answer this research question and its hypotheses. The
variables selected for the stepwise regression to analyze Time 1, Type I adaptation were
those signiﬁcant from the three preceding regressions (caregiver age and relationship to
the care recipient, care recipient gender, amount of family help, hours of physical care).
Three variables (caregiver age, hours of physical care, amount of family help) explained
19% (R2 = .186), p = .000) of the variance in Type I adaptation. Table 23 contains

the regression outcomes.

138
Table 23. Regression of Selected Significant Variables and Time 1, Type I

 

 

Adaptation
Variable b [3 T Sig t
Constant 3.215 7.047 .000
Caregiver age -.037 -.283 -4.549 .000
Hours of physical .064 .212 3.478 .001
care
Amount of family .075 .160 2.570 .011
help
R2 .186
F 16.892
p—value .000
N 210

 

Variables in equation = p < .05

Backward regression, using the same variables as in the stepwise regression of
selected signiﬁcant variables, was used to ascertain whether an alternate approach to
analysis would generate different outcomes. With backward regression, all variables
initially enter the equation and then are systematically removed depending on pre-
established removal criteria (Norusis, 1990b). In contrast to the stepwise model,
backward regression (Table 24) produced an equation that included all the variables
(caregiver age, family relationship, hours of physical care, amount of family help and
care recipient gender). This combination of variables and method produced a modest

increase in the amount of Type I variance explained (R2 = .229).

Table 24. Backward Regression of Selected Signiﬁcant Variables and Time 1, Type

   

139

 

 

 

I Adaptation
Variable b
Constant 2.655 4.682 .000
Hours of physical care .057 .190 3.154 .002
Spouse relationship 1.445 .416 3.055 .003
Amount of family help .082 .175 2.670 .008
Caregiver age -.039 -.297 -3.911 .000
Care recipient gender -.700 -.219 -2.392 .018
Daughter relationship .881 .272 2.737 .007
R2 .229
F 10.841
p-value .000
N 210

 

Variables out of equation = p > .10

Previous discriminant analyses results provided the signiﬁcant variables (caregiver
education, co-residence now, amount of family help, hours of physical care) for
additional two-group discriminant analysis of Time 1, Type H adaptation. This
combination of variables produced a model with overall signiﬁcance (p = .001). The
discriminant function for the selected variables and Type H adaptation correctly grouped
64% (64.16) of the cases.

For Time 1, Type III investigation, signiﬁcant variables from the previous Type
HI discriminant equations (family network, frequency of family help, total caregiver

involvement, hours of physical care) and the independent t-test (caregiver household

140

income) were used in the discriminant analysis. These variables generated a signiﬁcant
overall equation (p = .000) and correctly classiﬁed 75% (75.25%) of the cases.

Table 25 contains the canonical discriminant functions and Table 26 includes the
classiﬁcation outcomes for selected signiﬁcant variables and Time 1, Type H and Type
HI adaptations.

Table 25. Canonical Discriminant Functions for Selected Significant Variables and
Time 1, Type H and Type HI Adaptations

 

 

 

5‘53: 332$ 3.1%: x2 DP 5..
Type H
Time 1 .124 .332 .890 19.776 4 .001
(N =173)
Type HI
Time 1 .178 .389 .849 31.764 5 .000

(N= 198)

141

Table 26. Discriminant Function Classification for Selected Significant Variables
and Time 1, Type H and Type IH Adaptations

 

 

 

Predicted Predicted
Type II
(N = 173)
Actual Group Cases LOA N o LOA
LOA 44 (100%) 22 (50.0%) 22 (50.0%)
No LOA 129 (100%) 40 (31.0%) 89 (69.0%)
Percent of cases 64.16%
correctly
grouped
Type 1H
(N = 198)
Actual Group Cases Quit/Retire No Quit or

Retire

Quit/Retire 35 (100%) 24 (68.6%) 11 (31.4%)
No Quit/Retire 163 (100%) 38 (23.3%) 125 (76.7%)
Percent of cases 75.25%

correctly

grouped

142
In summary, selected signiﬁcant variables (derived from the separate Time 1

regression and discriminant analyses results or the independent t-test outcome) for
caregiver/care recipient characteristics, family support, and caregiver involvement offered
signiﬁcant explanations for each adaptation type. The combination of signiﬁcant variables
differed for each adaptation type. Regardless of the variable mix, the outcomes suggest
more explanation and differentiation power for Type HI adaptation (75% correct
grouping) than either Type I (stepwise R2 = .186; backward R2 = .229) or Type H (64%
correct grouping) adaptations at Time 1.

At Time 2, there were no signiﬁcant relationships identiﬁed with any of the
caregiver/care recipient, family support, or caregiver involvement variables and Type I
or Type H adaptations. Thus, additional analyses were not indicated for those adaptation
types.

However, two variables (caregiver education and hours of supervision) did have
signiﬁcant relationships with Time 2, Type HI adaptation and were used in two-group
discriminant analysis. The two variables produced a model with a signiﬁcant overall
equation (p = .001) and correctly classiﬁed almost 75% (74.75%) of the cases.

Table 27 contains the canonical discriminant functions and Table 28 the

classiﬁcation outcomes for selected signiﬁcant variables and Time 2, Type HI adaptation.

143

Table 27. Canonical Discriminant Functions for Selected Signiﬁcant Variables and
Time 2, Type HI Adaptation

   

Eigen- Canonical Wilks’s X2 DF

 

value Correlation Lambda 813’
Type 111
Time 1 .171 .382 .854 15.170 2 .001
(N =99)

 

Table 28. Discriminant Function Classification for Selected Significant Variables
and Time 2, Type HI Adaptation

 

 

Predicted Predicted
Type III
(N =99)
Actual Group Cases Quit/Retire No Quit or

Retire

Quit/Retire 16 (100%) 5 (31.3%) 11 (68.8%)
No Quit/Retire 83 (100%) 63 (75.9%) 20 (24.1%)
Percent of cases 74.75%
correctly
grouped

 

The speciﬁc approach and discussion for each hypothesis about key caregiver/care
recipient, family support, and caregiver involvement variables and employment

adaptation follow.

144

H1211 The hours of physical care are the best predictors of each type of employ-
ment adaptation at Time 1 and Time 2.

The results supported this hypothesis for Type H, but not Type I or Type III
adaptations at Time 1. The discriminant analysis showed that the standardized canonical
coefﬁcient (Table 29) for hours of physical care (.659) was higher than for the other
three variables in the equation (caregiver education, -.463; co-residence now, .304; and
amount of family help, -.359).

However, the stepwise regression analysis for Time 1, Type I adaptation (Table
23) indicated hours of physical care did not have the highest beta, but the second highest
of the three variables in the equation (caregiver age, -.283; hours of physical care, .212;
and amount of family help, . 160). Likewise, the backward regression (Table 24) showed
that hours of physical care did not explain relatively more variation than the other
variables in the equation (it was fourth highest of the ﬁve variables).

Similarly, discriminant analysis for Time 1, Type HI adaptation displayed hours
of physical care as having the second highest standardized coefﬁcient of the ﬁve variables
in the equation (Table 29). While hours of physical care was a good predictor of Time
1, Type I and Type III adaptations (coefﬁcient = .429), the best predictor was total
caregiver involvement (coefﬁcient = .536). Table 29 contains the unstandardized and

standardized canonical discriminant function characteristics for assorted variables and

Time 1, Type H and Type III adaptations.

 

 

145

Table 29. Canonical Discriminant Function Coefficients for Selected Significant
Variables and Time 1, Type H and Type III Adaptations

   

Type H Type HI

 

(N = 173) (N = 198)

Unstand— Standard- Unstand- Standard-

ardized ized ardized ized
Caregiver education -. 194 -.463 NA NA
Co-residence now .640 .304 NA NA
Amount of family help -.983 -.359 NA NA
Hours of physical care .137 .659 .826 .429
Caregiver income NA NA -. 176 -.300
Family members NA NA .150 .323
within 50 miles
Frequency of family NA NA -.444 -.212
help
Total caregiver in- NA NA .363 .536
volvement

 

Note: NA Variable not signiﬁcant for adaptation type (based on previous discriminant
equations).

At Time 2, there were no signiﬁcant relationships with any of the variables and
Type I or Type II adaptations. Thus, additional analyses were not indicated for the
caregiver/care recipient characteristics, family support, or involvement variables. While
there were signiﬁcant relationships with Time 2, Type HI adaptation, and subsequent
analysis, the variables did not include the hours of physical care (Table 30). Table 30
contains the unstandardized and standardized canonical discriminant function coefﬁcients

for selected signiﬁcant variables and Time 2, Type IH adaptation.

146

Table 30. Canonical Discriminant Function Coefficients for Selected Significant
Variables and Time 2, Type HI Adaptation

   
   

  
 
 

Type 111 (N=99) Type In (N=99)

 

 

 

Unstandardized Standardized
Caregiver education -.206 -.473
Hours of supervision .129 .838
H12b The amount of family help at both Time 1 and Time 2 is less effective in
predicting employment adaptation than are the other variables in the

equation.

The results supported this hypothesis for Time 1, Type I and Type IH adaptations,
but not for Time 1, Type II adaptation. In the stepwise regression equation for Type I
adaptation (Table 23), the amount of family support had a lower beta (.160) than the
other variables in the equation, suggesting it contributed less to explain variation in Type
I adaptation at Time 1 than did caregiver age (6 = -.283) and the hours of physical care
(,8 = .212). Similarly, the backward regression outcomes showed the amount of family
support as having the lowest beta of the six variables in the equation.

The analysis for Time 1, Type III adaptation did not include amount of family
help (no previous signiﬁcant relationship), thus implying that the variable contributes
little to predicting Type III employment adaptation.

However, the standardized canonical coefﬁcient (Table 29) for amount of family
help was the third highest (-.359) of the four variables in the equation to differentiate
Time 1, Type II adaptation. While the results indicated the variable was relatively less

effective than two of the variables (hours of physical care coefﬁcient = .659 and

147

caregiver education coefﬁcient = -.463), co—residence now (coefﬁcient = .304) offered
the smallest relative contribution to the discriminant function.

At Time 2, there were no signiﬁcant relationships with any of the variables and
Type I or Type H adaptations. While there were signiﬁcant relationships with two of the
variables and Type IH adaptation, they did not include the amount of family help. This
suggests that the amount of family help offered little explanatory power for Type 1, Type

H, or Type HI adaptations at Time 2.

H12c Caregiver and care recipient characteristics explain more variation in
employment adaptation than characteristics of caregiver involvement.

The results supported this hypothesis for Time 1, Type I adaptation, but not for
Time 1, Type H or Type III, or Time 2, Type III adaptations. The stepwise regression
results (Table 23) indicated caregiver age (B = -.283) contributed relatively more than
the other two variables (hours of physical care, B = .212; amount of family help,
B = .160) to explain the variation in Type I adaptation at Time 1. Similarly, when
backward regression analysis was used (Table 24), caregiver/care recipient characteristics
(family relationship, caregiver age, and care recipient gender) explained relatively more
of the variation in Type I adaptation than did the involvement variable (hours of physical
care).

However, for Time 1, Type II adaptation, hours of physical care had the highest
standardized canonical coefﬁcient (.659) of the four variables in the equation (Table 29).
The lower standardized canonical discriminant coefﬁcients of the two caregiver/care re-

cipient characteristic variables in the equation (caregiver education = -.463; co-residence

148

now = .304) suggest they offered relatively less to the discriminant function than did an
involvement variable.

For Time 1, Type 1H adaptation, total caregiver involvement had the highest
standardized canonical coefﬁcient (.536), and hours of physical care had the next highest
(.429) of the ﬁve variables in the discriminant equation (Table 28). This suggests that
involvement variables had a relatively greater contribution to the discriminant function
than did the only caregiver/care recipient variable to enter the equation (caregiver
household income = -.300).

At Time 2, there were no signiﬁcant relationships with any of the variables and
Type I or Type H adaptations. This suggests that both caregiver/care recipient
characteristics and caregiver involvement offered little contribution to explain
employment adaptation.

However, Time 2, Type III adaptation displayed a signiﬁcant relationship with
a caregiver characteristic variable (education) and an involvement variable (hours of
supervision). Discriminant analysis produced standardized canonical coefﬁcients for the
variables (education = -.473; hours of supervision = .838) that suggest that an
involvement variable contributed relatively more to the equation than did the caregiver
characteristic variable. Thus, the results do not support the hypothesis for Time 2, Type

IH adaptation.

CHAPTER V

Discussion

This section includes a summary of study ﬁndings, followed by discussion and
implications for practice, policy, and research. The results of the analyses are organized
by employment adaptation type and the relationship of each (Type 1, Type H, Type 1H)
with the groups of independent variables. Outcomes at the .05 level were considered
signiﬁcant.

The study, encompassing two time periods, broadened the scope of the usual
employment and family caregiving literature in several ways. First, it focused on
eldercare, not childcare, and employment. While substantial research about childcare
and employment exists (Friedman, 1988; Kola & Dunkle, 1988), the study of the
interaction of eldercare and employed caregivers is a relatively new phenomenon (Brody,
Kleban, Johnsen, Hoffman, & Schoonover, 1987).

Second, eldercare and employment adaptation were considered to be dynamic, not
static, processes that were likely to change over time. Existing longitudinal studies
(Given, Stommel, & Lin, 1991, Robinson & Thumher, 1979; Stoller & Pugliesi, 1989b;
Zarit, Todd, & Zarit, 1986) present a complex picture of change.

Finally, the researcher incorporated potential explanatory factors (caregiver/care

recipient characteristics, family support, and caregiver involvement) to assess their

149

150

inﬂuence on caregiver employment adaptations. The majority of existing family
caregiver and employment literature is not only cross-sectional in design, but also, with
some exceptions (Neal, Chapman, Ingersoll-Dayton, Emlen, & Boise, 1990), descriptive
and without an attempt to predict employment adaptation based on the presence of certain

variables.

Summary of Findings

This study investigated the inﬂuence of three blocks of variables (caregiver/care
recipient characteristics, family support, and caregiver involvement) on using short-term
employment adjustments, taking a leave of absence, and quitting or retiring from work
in order to provide eldercare. Two time periods, inception of the eldercare role and
three months later, were studied.

At inception of the eldercare role, the caregiver’s age and relationship to the care
recipient, whether she lived with the care recipient, the hours of physical care, and the
care recipient’s gender were likely to inﬂuence the short-term employment adjustments
made by the caregiver to meet family obligations. The family support characteristics did
not seem to have a strong relationship with the temporary work alterations used by the
caregiver. When the signiﬁcant variables from each block were studied together, three
variables, caregiver age, hours of physical care, and the amount of family help, were the
most useful in describing the caregiver more likely to use short-term work adjustments
in order to provide care.

Caregivers who chose to take a leave of absence from their employment at

inception of the eldercare role were inﬂuenced by slightly different factors. It appears

151

that a leave of absence may be more likely to occur with caregivers who have less
education, live with their care recipient, and provide more hours of physical care than
do those who choose to continue employment. The amount of family help provided to
either the caregiver or the care recipient exerted a limited effect on the caregiver’s
decision to take a leave of absence. However, when the four variables (caregiver
education, co-residence, amount of family help, and hours of physical care) were studied
together, they each were important in describing the caregiver likely to use a leave of
absence in order to meet demands of eldercare.

Finally, choosing to quit or retire from the workforce at inception of the eldercare
role seems more likely to be the decision of a caregiver with a lower household income
and who provides higher physical care hours. The family care network number, and the
amount and frequency of family help appeared to have little inﬂuence on the decision of
the caregiver to quit or retire. Nonetheless, when the family variables, household
income, and hours of physical care were studied as a group, each was important in
predicting caregivers more likely to quit or retire in order to provide care.

The results for the second time period of study were very different. Three
months after assuming the eldercare role, no caregiver or care recipient characteristic,
family support, or caregiver involvement variable inﬂuenced the caregiver’s decision to
make short-term work adjustments or to take a leave of absence from her employment.
However, the caregiver who chose to quit or retire three months after assuming the
eldercare role was more likely to be less educated and to provide more hours of

supervision than did those caregivers who continued employment. A complete summary

152

of results for both inception of the eldercare role (Time 1), and three months later

(Time 2) follows.

11019.1
Type I (maintain, but alter work) Employment Adaptation

Time 1 results indicated seven caregiver/care recipient variables (caregiver age,
household income, education, relationship to the care recipient, current employment
status, co-residence now, and care recipient gender) had signiﬁcant relationships with
Type I adaptation. Four variables (caregiver marital status, occupation, co—residence
before, and care recipient age) showed no signiﬁcant relationship with Time 1, Type I
adaptation.

In order to assess the relative inﬂuence of the signiﬁcant caregiver/care recipient
characteristics on Type I adaptation, a regression analysis was done. Employment status,
although signiﬁcant, was not included in the analysis so that the impact of the remaining
signiﬁcant caregiver/care recipient characteristics could be calculated. Three variables,
(caregiver age, relationship to care recipient, and care recipient gender), explained 19%
(R2 = .191) of the variance in Time 1, Type I adaptation.

From the group of three family variables (amount of family help, frequency of
aid, and family network), only the amount of family help related signiﬁcantly to Time
1, Type I adaptation. Although just one variable displayed a signiﬁcant relationship with
Time 1, Type I adaptation, all family variables were used in a regression analysis

because of family’s importance in the conceptual model for the study. As bivariate

153

correlations suggested, amount of family help was the sole variable to enter the equation
and explained just 5% (R2 = .047) of the variance in Type I adaptation.

Two involvement variables (total caregiver involvement and hours of physical
care) displayed a signiﬁcant correlation with Type I adaptation, but one (hours of
supervision) did not. Of the signiﬁcant variables, only hours of physical care met
criteria to enter the subsequent regression equation and accounted for 6% (R2 = .064)

of the variance in Time 1, Type I adaptation.

rype II (leave of absence) Employment Adaptation

For Time 1, Type H adaptation, only three of the eleven caregiver/care recipient
characteristic variables (caregiver education, current employment status, and co-residence
now) displayed signiﬁcant relationships. As with Type I adaptation, employment status
was not used in subsequent multivariate analysis. The other signiﬁcant variables
(caregiver education and co-residence now) contributed to a signiﬁcant discriminant
function and correctly grouped almost 59% of the cases. However, that percentage is
less than the identiﬁed practical signiﬁcance (60%) for the study.

Although just one family variable (amount of family help) showed signiﬁcance
with Time 1, Type II adaptation, as with Time 1, Type I analyses, all were used in the
discriminant equation because of family’s importance in the conceptual model for the
study. However, similar to Type I outcomes, only one variable, the amount of family
help, contributed to the discriminant function and correctly grouped 55% of the cases.
The low percentage of appropriate classiﬁcation is explained by the discriminant model’s

overall ineffectiveness with family variables.

154

Each involvement variable (total caregiver involvement, hours of physical care,
hours of supervision) demonstrated a signiﬁcant relationship with Time I, Type H
adaptation. Nonetheless, only one (hours of physical care) contributed to the discrimi-
nant function, produced a signiﬁcant model, and correctly grouped 66% of the eases.
Thus, hours of physical care appears to be a more effective variable to predict Type H

employment adaptation when compared with the amount of family help.

Type III (quit/retire) Employment Adaptation

Only one of the eleven caregiver/ care recipient characteristics (caregiver
household income) showed signiﬁcance with Time 1, Type HI employment adaptation.
An independent t-test conﬁrmed signiﬁcant differences related to the income of those
caregivers who chose to quit or retire (lower mean incomes) and those who did not.

Two of the three family variables (family network and frequency of family aid)
showed signiﬁcance with Time 1, Type HI adaptation. However, as with Type H
adaptation, the variables did not generate a discriminant model with overall signiﬁcance.
Even though both variables contributed to the discriminant function, they correctly
grouped only 57 % of the cases, not much better than the 50% expected by chance alone.

As with Type II adaptation, each involvement'variable displayed signiﬁcance with
Type HI adaptation. However, only two (total caregiver involvement, hours of physical
care) generated an effective model, contributed to the discriminant function, and induced

improved classiﬁcation results (72% correct grouping).

155
1111112
All Adaptation Types

Time 2 results for Type I and Type H adaptations were inconclusive. No
caregiver/care recipient characteristic, family support, or caregiver involvement variable
predicted Type I or Type H adaptations. Also, no family support variable predicted Type
IH adaptation.

However, a caregiver/ care recipient variable (caregiver education) and. an
involvement variable (hours of supervision) showed signiﬁcant relationships with Type
HI adaptation. Independent t-tests conﬁrmed signiﬁcant differences for each variable and
Type HI adaptation. Caregivers who chose to quit or retire at Time 2 had a lower mean
educational level and provided more hours of care recipient supervision than those

caregivers who continued their employment.

Compatibility with the Conceptual Model

The conceptual model guiding the study placed equivalent emphasis on each group
of independent variables and their inﬂuence on employment adaptation. Outcomes of
statistical analyses related to Time 1, Type I adaptation did not support the model,
however. At Time 1, caregiver/care recipient characteristics (R2 = . 191) evidenced
stronger predictive power than either family support (R2 = .047) or caregiver
involvement variables (R2 = .064). This outcome was incompatible with the preliminary
conceptual model (Figure 1, p. 38).

Type H results were more congruent with the model than Time 1, Type I effects.

Although differences existed between the variable groups, they were less dramatic than

156
ﬁndings for Time 1, Type I adaptation. For Time 1, Type H adaptation, the caregiver

involvement variables generated the highest correct grouping of cases (66%).
Caregiver/care recipient characteristics were next with 59 % correct, while the family
support variables had the weakest outcomes (55 % correct grouping).

Time 1, Type HI differences, as with Time 1, Type I contrasts, were incompatible
with the preliminary conceptual model. Classiﬁcation variations for Time 1, Type IH
adaptation, based on the three groups of independent variables, were less dissimilar than
ﬁndings for Time 1, Type I, but more diverse than Time 1, Type H adaptation.
Caregiver involvement generated the best (72% correct) grouping, family support
produced the next highest (57% correct), while caregiver/care recipient variables had the
weakest outcome (only one variable showed signiﬁcance).

The Time 2 data, particularly for Type I and Type H adaptations, did not support
either the expectations for the study or the appropriateness of the conceptual model. For
Type I and Type II adaptations, no variable from any independent group met criteria to
enter the regression equation. While Type III adaptation did have improved outcomes,
variables from only two of the independent groups (caregiver/care recipient characteris-
tics and caregiver involvement) showed signiﬁcance. The third variable group (family
support) showed no relationship with Type III adaptation.

Perhaps the general incompatibility of the model to employment adaptation at both
time periods represents the uniqueness of each care situation. Multiple factors inﬂuence
caregivers in varied ways and also are likely to change over time. The eldercare process
and its interaction with work and family may not lend itself to a distinct conceptual

model with equivalently inﬂuential variables. A model that includes all variables and

157

their signiﬁcance (or lack) with each adaptation type and over time may offer more
meaningful information about the unique inﬂuence each has on employment adaptation.

A transitional conceptual model depicts key outcomes of the present study (Figure 8).

23329.0( .eeE>o_nEm a 2.202m 2.5.“. 3. .033 3:32:30 Eco—22.2... .o 2.3.“.

 

158

€998: oaoumo 8393 2352.38 35:63 E .50...ch om_<u Ail a 08.: v.55} mgr—u O
EonEOnoo 1 1
95:96.9 c.9529 E8535» A I i l 2 9:5 eczaoosu i l.

 

8.3235 .0 manor

 

 

 

   

 

I“ \

9m 8...." 0330
\\\\_l o. .3“: :J

\ i
\ \ \ .coEm>_o>c_ 00 .30»
'''''' J - \ \
_ / \ t
\ \
\ \
\

_ 7 9o: 2.83 .6539“. J
2 as: _
_ L \ \ \ r so: 2.83 .595 J

llllll V \\ _i £0352 2.an J

 

c3282
5 on:

 

 

 

 

 

 

 

 

 

 

 

  

 

50.6525
__ on:

 
 

 

  

 

 

 

 

 

 

 

 

 

 

 

\ \
\ \ ‘ am
< O
k \ K ’I _ a J
1 llllll a I :05: oucoEmomoo
. c9662? .
u .. one; _ A :62. ooeoeaomoo
\ — oan:. 00 J

  

 

 

cessamo<
_ ~53

 
 

7 3.59600 00 J

\ .\
\ \
\ \ \ \ l i T 8:838 8 _

 

  

 

 

 

 

 

 

 

 

 

_ lllllll . »\ \\\\ \ \ i i l l i t. l l 7 3.2925200 i—

_ 8:283. _ .i t t t T seaside 8 i—
_ on;

_ e ..A . _ o? 8 J

r llllll

gang gangs

159
V ' l r

In the study, separate regression and discriminant analyses, plus independent t-
tests examined the inﬂuence of three groups of independent variables (caregiver/care
recipient characteristics, family support, and caregiver involvement) on each employment
adaptation type. Signiﬁcant variables from those procedures provided the independent
variable list for a ﬁnal query about their relative inﬂuence on employment adaptation.
This approach incorporated the interaction of the independent variables that is central to
Human Ecology theory and the employment adaptation outcomes that are important to
Choice and Exchange theory.

In the final Time 1, Type I regression equation, three variables (one from each
independent variable group) explained about 19% (R2 = .186) of the variance for Type
I adaptation. While caregiver age ()3 = -.283) made the highest relative contribution to
the equation, it was closely followed by hours of physical care (,8 = .212). The amount
of family help (B = .160) offered the least explaining power for the equation.

When a backward regression (out = p > .10) was used with the same variables
to see how results might differ, the R2 improved slightly (R2 = .229). The ﬁnal
equation included family relationship and care recipient gender as well as the variables
that entered the stepwise version. In the backward regression, the caregiver/care
recipient variables exhibited the highest betas (spouse = .416, caregiver age = -.297,
daughter = .272, care recipient age = -.219). The amount of family help and hours of
physical care had the lowest betas (.175 and .190, respectively). Thus, it appears that

some caregiver/care recipient characteristics, using either regression approach, were

160

relatively more inﬂuential in predicting Type I employment adaptation than the other
variable groups.

Time 1, Type II outcomes of separate discriminant analyses also indicated that
variables from each independent group were important in predicting Type II adaptation.
The combined signiﬁcant variables contributed to an effective discriminant model for
Type H adaptation and correctly classiﬁed 64% of the cases. The best predictor variable,
hours of physical care, had the highest standardized canonical discriminant function
coefﬁcient (.659) of the four variables in the equation (caregiver education = -.463,
amount of family help = -.359, co-residence now = .304).

Consistent with Time 1, Type I and Type II adaptations, Type III adaptation
outcomes included signiﬁcant variables from the three independent variable groups. Of
the ﬁve variables in the ﬁnal discriminant equation, total caregiver involvement had the
highest standardized canonical discriminant function coefﬁcient (.536). The other
variables, hours of physical care (.429), family members within 50 miles (.323),
caregiver income (-.300), and frequency of family help (-.212) offered relatively less to
the discriminant function.

The absence of signiﬁcant relationships with any variable and Time 2, Type I and
Type II adaptations undoubtedly depict disappointing Time 2 results. Only caregiver
education and hours of supervision showed signiﬁcant relationships with Time 2, Type

III adaptation. However, those variables, representing two of the three groups of
independent variables (caregiver/care recipient characteristics and caregiver involvement)

contributed to the discriminant function and generated 75 % correct classiﬁcation results.

161

The outcomes from the ﬁnal analyses of signiﬁcant variables from each
independent group and their inﬂuence on the adaptation types led to the study’s
concluding conceptual model. This model displays how, in relation to each other and
over time, the variables inﬂuenced employment adaptation. Figure 9 shows the

concluding conceptual model.

23:39.3 .eoanEEm a obtuse—um 2.5.“. .3 .302 33:00:00 255.230 .9 2:2“.

 

162

.8563 Smoumo 89.96 2962.36 .9... c. _So._.co.m 32.. All 8 9:5 2.5.2 827. O
E>5m50u00 1. I
2235 c050 3955 c. _eau.__ca,.wn A l l l 2 95: 3:89.? r I.

 

 

5.652% A _
5 25

8.2233 .0 2:0:

 

 

 

 

 

\\ _ Emu _mo_m>cn_ _o 2:01
_

_ co=mamc<_ ll _cmEozoE. 00 .99.
. 5 e9: _

 

 

 

 

9m: 22%”. _o >ocm=cm£ _

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

$3.32 3m: 22:8 38:2
__ 8a 6‘ _
é x5352 2E5“. _
1 llllll .— t
_ . ,
eczmamg _ / \ _
_ > . / x x 5950 mo .
. l l ..Imm .pl l l ’ \
\ 952v mocmoammoo
8:232 \ @885 Go
_ 8.: x x
x \ 8:828 00
\ \.
. lllllll _ \ x l l a. 22mm
_ CO——N—ch< - ' llllllllll .2 . _ m 00
m9 . A lllll
_rll.l lpll. A 09.00

 

 

 

 

a? gag

 

163

In spite of the original conceptual model’s imperfect ﬁt with study outcomes, the
ecological approach to the study was important. The alteration of the model to reﬂect
the perceived reality of the results illustrates adaptation, a basic tenet of ecological theory
(Bristor, 1990). Also, the study provided a ”holistic way of viewing human systems as
ecosystems with components bound together as functioning wholes in dynamic interaction
with the environment" (Andrews, Bubolz, & Paolucci; 1980, p. 42).

Illness or disability is not just an individual challenge, it involves the family
caregiver, other family members, the employer, and ultimately society. Family
caregivers respond, change, develop, act on information, and modify their environments,
both to meet eldercare and employment responsibilities and also to manage other family

and societal roles. Caregiver decisions, made in an environmental context, are used to

attain individual and family goals (Bubolz & Sontag, in press).

Conclusions

The study shows that employed family caregivers used various employment
adaptations to balance their work and family obligations. However, the particular
predictor variables differed in their inﬂuence on the adaptation types. Diverse factors
inﬂuenced caregiver decisions about both eldercare involvement and employment and,
in turn, affected both time periods.

The present study revealed some unanticipated results. The immediacy of the
inﬂuence of acquiring the eldercare role on the employment adaptation of women
represents a major ﬁnding. It was expected that increased employment adaptation (all

types) would occur over time, speciﬁcally from Time 1 to Time 2. However, caregivers

164

did not wait until Time 2 to make adjustments. The percentage of caregivers who
reported full-time or part-time employment prior to assuming the caregiver role related
substantial changes at Time 1 (an average of 27 days after screening for entry into the
study). Perhaps the Time 1 change (compared with prior employment) occurred in this
retrospective study because the caregiver had no speciﬁc time frame to consider when
responding to the question about employment prior to caregiving. Alternatively, the care
for a recently physically impaired elderly relative may create a sudden need for a family
member’s employment adaptation; thus, the caregiver decision to adapt work in order to
contribute the required services.

Regardless of Time 1 contrasts with reported prior (before assuming the caregiver
role) employment, contrary to expectations, Time 2 changes showed no signiﬁcant
differences in the expected direction (increased accommodation) for Type 1, Type II, or
Type III employment adaptations. Instead there were signiﬁcant changes with Type I and
Type II adaptations (less at Time 2) and there were no differences between the time
periods with Type III adaptation. Perhaps this outcome represents a return to the
caregivers’ balance of work and family obligations that existed before assuming the
eldercare role. It also may mean the caregiver is adapting to the eldercare role so
successfully, she ﬁnds it less necessary to adapt her employment. Clearly, the level of
Time 1, Type I adaptation (maintain, but alter work) did not inﬂuence the use of either
Time 2, Type 11 (leave of absence) or Type III (quit/retire) adaptations.

While the absence of signiﬁcant increased employment adaptation at Time 2 may
be due to the considerable adaptation that occurred immediately after assuming the

eldercare role, it also may reﬂect the study sample. A review of the reasons for the

165

decline in sample size at Time 2 reveals that 40 (17.9%) care recipients no longer
required care at Time 2. Some care recipients, because they were physically, not
mentally impaired, likely had conditions amenable to improvement through physieal
therapy or some other rehabilitative method.

In addition, 25 (11.2%) care recipients expired and four (1.8%) were either
hospitalized or institutionalized at Time 2. Thus, some care recipients who presumably
had high care demands (hence, more likely to inﬂuence employment adaptation) were no
longer in the study. Conceivably, the Time 2 decline in employment adaptation may
reverse at later periods of the core study (the complete study includes 6 time intervals,
3 months apart). Two snapshots in time (Time 1 and Time 2), although better than a
true cross-sectional design, do little to assess the daily, weekly, monthly, or yearly
adjustments likely to be part of the eldercare process.

As noted, existing longitudinal studies found caregiving to be a complex process
that changes over time. Robinson and Thurnher (1979), through their ﬁve-year
qualitative study of adult children and their aging parents, found that the process of
eldercare included a series of phases. Periods of increased caregiver tension, anxiety,
and feelings of restriction characterized these stages. While the Robinson and
Thumher study did not address caregiver employment issues, it is conceivable that
similar patterns related to work may emerge and trigger increased adaptation.
Caregiving is dynamic, not static; thus, intervals of varied intensity of adaptation may
be representative of the process.

A second unexpected study outcome is the negligible inﬂuence family variables

exerted on employment adaptation at either time period. One possible explanation is the

166

measurement of family involvement. The instrument used to assess the frequency of
family help had an acceptable alpha (.69). However, the tool utilized to learn the amount
of family help may be less reliable. For example, the caregiver was asked to indicate
the amount of help received from each network member. There may be no relationship
between the amount of reported help from family member #1, from member #2, from
member #3, and so on. Thus, the instrument may be less useful for obtaining important
information about the family.

Also, the self-report method of gathering information used with the core study,
provides subjective, not objective data. Caregivers may vary in how they interpret the
amount and frequency of help they receive from family members. Their assessment may
be clouded by the way they perceive the family member in question. For example, a
positive relationship may inﬂuence the caregiver to report substantial and frequent help
from a family member whether it is objectively true or not. Conversely, the reverse
situation also may occur. A negative relationship may sway the caregiver to assess the
frequency and amount of help as being poor, while in actuality, the opposite may be true.
The caregiver’s perception is her reality.

Former family relationships, values, structure, and patterns of care or mutual
support also may inﬂuence the involvement (or its lack) of network members with
eldercare. Family care is a life cycle event, beginning with childcare (for most families)
and ending with eldercare. A caregiver role for one individual may be sustained through
varied family stages. Consequently, substantial eldercare commitment from the extended

family may come only when the primary caregiver can no longer manage to provide

167

continued care. Studies show that one family member, usually a woman, provides the
most care (Finley, 1989; Horowitz, 1985).

Throughout the care situation, costs and exchanges occur. The caregiver’s
interpretation of the costs and beneﬁts of eldercare and/or employment adaptation
determines whether or not either activity is seen as a problem. Walker, Martin, and
Jones (1992) found that the costs of caregiving for daughters are reduced when there is
a good relationship with their relative, if the care recipient lives in separate housing, and
if the care recipient has fewer needs. While the present study included review of co-
residence and caregiver involvement, it did not investigate the quality of the care-
giver/care recipient relationship or caregiver burden.

Young and Kahana (1989) found employment to be a signiﬁcant predictor of
adverse reactions (costs) during the caregiving period. Other researchers reported
emotional strain, restrictions on time and freedom, and economic hardships (Buglass,
1989; Horowitz, 1985), as well as physical and mental exhaustion (Hooyman & Ryan,
1987).

Another probable eldercare cost encompasses the hours of care expended by the
caregiver. Care hours for the present study included separate categories for physical care
and supervision. While the literature reports care hours without differentiating care
rendered, studies document the consequences of those hours (Brody, Kleban, Johnsen,
Hoffman, & Schoonover, 1987; Scharlach, 1989). For example, the research ﬁndings
of Brody, Kleban, Johnsen, Hoffman, and Schoonover (1987) suggested that caregivers
providing about 38 care hours per week were more likely to quit employment than

caregivers with less time committed to care.

168

In the present study, there were signiﬁcant positive correlations with both hours
of physical care (about 26 hours/ week) and hours of supervision (about 49 hours/week)
and Time 1, Type II adaptation (leave of absence) and Time 1, Type III (quit/retire)
adaptation. A stronger relationship with hours of supervision (about 51 hours/week)
existed with Time 2, Type III adaptation. Discriminant analysis showed that hours of
supervision (along with caregiver education) correctly grouped almost 75 % of Time 2,
Type III cases.

Additionally, work days off without pay (a variable in the index of Type I
adaptation) can be a caregiver cost. However, the measure may be biased because of the
caregiver’s occupation and employment rank. For example, a professional or salaried
employee may take a day off for eldercare without losing compensation (more schedule
ﬂexibility), while an hourly employee probably will not have that option.

The same situation may occur with Type 11 (leave of absence) adaptation.
Perhaps the leave is paid, but it may not be. As with Type I adaptation, a professional
or salaried caregiver may have more options related to Type II adaptation than one who
is an hourly employee. The prevalent work environment, policies, and procedures
affecting work and family issues are not known for the study sample. In general, some
employers have responded to family needs in the workplace by adjusting policies,
beneﬁts, or services (Buglass, 1989; Fernandez, 1990; Kola & Dunkle, 1988; Scharlach,
Lowe, & Schneider, 1991), while others have not.

Consistent with other studies (Archbold, 1983, Scharlach & Boyd, 1989, Walker,
Martin, & Jones, 1992), caregivers at both time periods reported disruption with their

lives and daily activities. In time, most caregivers adjust to their eldercare role, but the

169

eventual equilibrium shifts with the care recipient’s next period of change. In this study,
counter to expectations, the adjustment included less employment adaptation at Time 2
than at Time 1. However, adaptation (regardless of the form) provides needed
stabilization for the family and the caregiver with multiple employment and family
obligations.

In addition to the costs that conﬂicting work and family obligations may create,
some researchers report beneﬁts for employed caregivers. Enright and Friss (1987)
related that employed caregivers of brain-impaired adults found work made their
caregiving obligations easier. Researchers also reported that, for some caregivers,
employment provided a psychological release and improved caregiver well-being (Barnes,
Given, & Given, 1991; Brody & Schoonover, 1986; Enright & Friss, 1987; Stoller &
Pugliesi, 1989a).

Thus, employment may provide not only respite from family obligations, but it
may also generate feelings of self-esteem because of job related knowledge and expertise.
In contrast, caregivers may lack the needed skills and competencies for some eldercare
tasks and, as a result, experience feelings of inadequacy. Caregivers frequently are
unprepared for the management of conditions and administration of treatments formerly
controlled in an acute care setting. It is possible that people continue working in order
to limit the time available for caring.

Further, varied reactions to caregiving and employment adaptation may relate to
socioeconomic status, family values, and personal and family goals. The ﬂow of

resources (monetary, education, culture) and information (perception, values,

170

competencies) leads the caregiver to the outcome, decisions about family care

involvement and employment adaptation.

Implications

Consistent with other studies (Creedon, 1987; Neal, Chapman, Ingersoll-Dayton,
Emlen, & Boise, 1990; Scharlach & Boyd, 1989; Stone & Short, 1990), data from the
present research showed that employment adaptation occurs for employed caregivers.
While the present study outcomes indicated less employment adaptive behavior at Time
2 than at Time 1, adaptations of all types still were evident. Also, this ﬁnding may
illustrate that the caregiver has truly "adapted" to the eldercare role, made lifestyle
modiﬁcations, and no longer ﬁnds it necessary to alter employment. However, the long-
term effect of eldercare may inﬂuence continued ﬂuctuation of employment adaptation.
Thus, important ramiﬁcations for practice, policy, and additional research become

evident.

Implications for employee assistance personnel, family educators, and family
professionals follow. In the workplace, persons who work directly with employee
assistance programs (EAP) may need to augment their own knowledge of gerontology
as well as increase their understanding of employed caregiver issues and possible
solutions. BAP staff should anticipate the immediacy of employee’s responses to a
family care crisis and be prepared for potentially rapid involvement. Long term

measures also may be essential. In addition, EAP personnel may co—ordinate caregiver

171

fairs to highlight formal resources available in the community or offer educational
sessions dealing with key issues.

Similarly, EAP staff may sponsor ongoing programs that increase management’s
realization of possible altered work productivity, utilization of caretaker beneﬁts (family
leave), and related costs because of conﬂicting work and eldercare demands. With
enhanced appreciation of the potential inﬂuence family eldercare has on employment,
managers may be encouraged to seek resolutions for work/family problems. A more
responsive and ﬂexible work environment for the family caregiver may evolve.

Some organizations have recognized and responded to eldercare’s signiﬁcance for
their company. These pioneers for work-friendly environments for employees with
family care responsibilities include: Travelers Insurance Companies, Bank of America,
IBM, and Phillip Morris (Scharlach, Lowe, & Schneider, 1991); Southwestern Bell and
Remington Products (Kola & Dunkle, 1988); Aetna Life and Casualty, Marriott (Buglass,
1989); and Chase Manhattan (Fernandez, 1990).

A second practice implication involves family educators in academic settings. The
educational preparation of family professionals requires content about family eldercare
and its possible ramiﬁcations (employment and others). While persons directly involved
in the care situation are cognizant of the costs (and the beneﬁts), others may not be. As
demographic and societal changes continue, work/family interactions will be increasingly
important considerations in curriculum decisions.

Knowledgeable instructors also should increase societal awareness of family
eldercare and its implications to the employed caregiver juggling multiple roles. Forums

on key issues (eldercare, employee needs, decision-making, values) offer one way to

172

reach a large audience. Persons attending such sessions may develop enhanced
understanding and support for co-workers coping with the dual work and family roles.

Family professionals represent a third group with practice implications derived
from the work/eldercare interaction. Persons working directly with either primary
caregivers or other family members may be resource brokers or providers of socio-
emotional services (education, counseling, stress management programs, facts about
community resources). Information about eldercare and resources available should be
accessible to both the caregiver and other family members. Families might need
encouragement to get involved with eldercare; conversely, the primary caregiver may
require permission and support to allow other family members care responsibility.

While the present study did not address whether or not the caregiver wished to
have more family assistance, clearly, little family help was provided. The support
available and the perception of the usefulness of the aid makes a difference in the
caregiver’s interpretation of the amount and frequency of family help. The family
professional may need to work with network members to improve communication skills
so that the family is able to recognize available help and use it as needed. Exploring the
caregiver’s willingness to provide care also may provide information about alternatives
and potential substitute caregivers or other resources.

In addition, the differential effects of any intervention must be considered by the
family professional. What is helpful to one caregiver may not be to another. It may be
necessary to make a concerted effort to change family task norms. For example, men
can be encouraged to have more involvement in family care activities throughout the

family life cycle. Intervention must be considered in the context of the caregiver and

173
family environment (past, present, future, psychological, and physical) to help identify

the interdependence of the family and how that relates to the family care situation.
Finally, family professionals may need to increase the ﬂexibility of available
service hours to meet the schedule needs of the employed caregiver. For instance,
offering early morning, evening, or Saturday hours to provide the services required by
either the caregiver or care recipient may allow the caregiver to meet family obligations

without the need to adapt employment.

Implicag’gns for Policy

Although the present study used a relatively small sample and has some
limitations, results provide additional information to help policy makers examine the
issues related to programs (existing or proposed) designed to help employees with family
care. The analysis of informal family care (eldercare or other dependent care) provides
information to help restructure social arrangements and economic resources.

Policy makers may need help altering the traditional image of "family" to one that
is more realistic in today’s world. Often the caregiver is single (25% of the caregivers
in the present study were unmarried) and many times the caregiver is from the
”sandwich” generation (61% of the present study were daughters, the average age was
52 years) and may have dependent children as well as responsibility for parent care
(Brody, 1985; Sidel, 1986). Policies should not jeopardize the caregiver who chooses
to take a leave of absence or leave the workforce temporarily in order to provide care.

Currently, either adaptation (Type II or Type III) may jeopardize the caregiver’s own

174

elder years because of the probable interruption in health and pension beneﬁts resulting
from that decision.

Flexible policies to meet the changing needs of family care throughout the life
cycle are needed. While 31 states and the District of Columbia have some version of a
family leave statute (most include maternity or adoptive leave; some also incorporate
other serious family problems), a national policy does not exist (”The Family and
Medical Leave Act,” 1991). Initiatives to establish national policies related to work and
family should be nurtured and supported. Hooyman (1990) maintains that the well-being
of family caregivers has not been the target for public policy.

In contrast, Hokenstad and Johansson (1990) report that Sweden recently
incorporated a care leave policy into the Swedish Social Insurance System enabling
Swedish family caregivers to take up to 30 days paid leave from their employment in
order to manage eldercare obligations. The Swedish government also emphasizes the
provision of caregiving salaries if caregiving is a part-time or full-time job (Hokenstad
& Johansson, 1990).

While policy changes require immense effort and enormous popular and
governmental support, they begin with individual involvement at the local level. This
commitment eventually extends to state and federal arenas. Ultimately, policy changes
supportive of work and family obligations will make a difference to the employed
providers of family eldercare, both today and in the future. The adaptation of current
policies to support family needs across the life cycle may come slowly, but it will come.
Adaptation means the modiﬁcation of behavior, feelings, and ideas to meet changing

demands and conditions. This applies to both political and family environments.

175

h sear h

The ﬁnal implications derived from the study are those for research. Additional
research, using the present study design, can provide data to determine whether there are
patterns in the ebb and ﬂow of employment adaptation. Future investigation should
include the quality of the caregiving dyad afﬁliation and the perceived burden (cost) of
the caregiver in order to evaluate differences in the costs of eldercare that may result
when the relationship quality varies.

Although the present study extends existing research by considering the interaction
of eldercare and employment over two time periods, a longer time frame is needed.
Longitudinal investigations (such as the core study) more ably address work/family issues
than do the cross-sectional studies that dominate the caregiving and employment literature
or the two time periods of the present inquiry.

The workplace has paid little attention to the problems employees may have with
eldercare responsibilities (Warshaw, Barr, Rayman, Schachter, & Lucas, 1986). While
managers are beginning to recognize the close connection of work and family, its effect
is still unknown. More studies are needed that compare the employment adaptation and
the work quality of employed family caregivers with workplace accommodations of non-
caregiving employees (Neal, Chapman, Ingersoll-Dayton, Emlen, & Boise, 1990).
Research also is needed to identify workplace characteristics that help employed
caregivers remain on the job as well as analysis to evaluate the effectiveness of existing
family beneﬁt programs (Scharlach, Lowe, & Schneider, 1991).

Investigation to assess the linkage with eldercare and other work-related events

(job turnover, productivity) provides yet another area of nwded research. Finally,

176

research that considers race and ethnicity and the diversity that those variables may bring
to employed caregivers’ (both men and women) employment adaptation is required
(White-Means & Thornton, 1990).

This study investigated the inﬂuence of acquiring the family caregiver role on
female employment adaptation at the inception of the eldercare function and three months
later. The primary caregivers shared their time, their talents, and sometimes their home
environment to meet eldercare obligations. They used internal resources (knowledge,
skills, and coping abilities) to manage the physical, instrumental, and health care
activities of daily living required by the care recipient. They altered work patterns and
career goals. Caregivers at both time periods elected to use a variety of Type 1, Type
II, or Type III adaptations in order to balance work and family care responsibilities.

How that equilibrium will shift remains for future research to describe.

LIST OF REFERENCES

LIST OF REFERENCES

American Association of Retired Persons. (1987). Caregivers in the work place.
Survey results. Overall summary. Washington DC: AARP.

Anastas, J. W., Gibeau, J. L., Larson, P. J. (1990). Working families and eldercare:
A national perspective in an aging America. Journal of the National Association
of Social Workers, 35(5), 405-411.

Andrews, M. P., Bubolz, M. M., & Paolucci, B. (1980). Study of the family.
Marriage and Family Review, 3(1/2), 29-49.

Archbold, P. G. (1983). Impact of parent caring on women. Family Relations, 32: 39-
45.

Babbie, E. (1989). The practice of social research (5th ed). Belmont, CA: Wads-
worth.

Barnes, C. L., Given, C. W., & Given, B. A. (1991). Parent caregivers: A compari-
son of employed and non-employed daughters. Manuscript submitted for
publication.

Barusch, A. S. & Spaid, W. M. (1989). Gender differences in caregiving: Why do
men report greater burden? The Gerontologist, 29(50), 667-675.

Bristor, M. W. (1990). Individuals, families and environments. Dubuque, IA: Kendall-

Hunt.

177

178
Brody, E. M. (1985). Parent care as a normative family stress. The Gerontologist,

25(1), 19-29.

Brody, E. M. (1990). Women in the middle: Their parent care years. New York:
Springer.

Brody, E. M. & Brody, S. J. (1989). The informal system of health care. In F.
Eisdorfer, D. A. Kessler, & A. N. Spector (Eds.) Caring for the elderly:
Reshaping health policy (pp. 259-277). Baltimore, MD: John Hopkins University
Press.

Brody, E. M. & Schoonover, C. B. (1986). Patterns of parent care when daughters
work and when they do not. The Gerontologist, 26, 372-381.

Brody, E. M., Kleban, M. H., Johnsen, P. T., Hoffman, C. & Schoonover, C. B.
(1987). Work status and parent care: A comparison of four groups of women.
The Gerontologist, 27(2), 201—208.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA:
Harvard University Press.

Bubolz, M. M., Eicher, J. B., & Sontag, M. S. (1979, spring). The human ecosystem:
A model. Journal of Home Economics 28-31.

Bubolz, M. M. & Sontag, M. S. (in press). Human ecology theory. In P. Boss, W.
Doherty, R. LaRossa, W. Schuman, S Steinmetz (Eds.) Sourcebook of family
theories and methods: A contextual approach, Chapter 17. New York: Plenum
Press.

Buglass, K. (1989). The business of elder care. American Demographics, 32-39.

179
Bureau of National Affairs. (1989). 101 key statistics on work and the family for the

1990s, Special Issue #21. Washington DC: Bureau of National Affairs.

Craft, J. L. (1990). Statistics and data analysis for social workers (2nd ed.). Itasca,
IL: F. E. Peacock.

Creedon, M. A. (1988). The corporate response to the working caregiver. In Aging.
Washington DC: US. Government Printing Ofﬁce.

Denton, K., Love, L. T., & Slate, R. (1990). Eldercare in the ’90s: Employee
responsibility, employer challenge. The Journal of Contemporary Human
Services, 349-359.

Eldercare beneﬁts meet new employee needs. (1988). Employee Beneﬁt Plan Review,
42(12), 21,24.

. Enright, R. B. (1991). Time spent caregiving and help received by spouses and adult
children of brain-impaired adults. The Gerontologist, 31(3), 375-383.

Enright, R. B. & Friss, L. (1987). Employed caregivers of brain-impaired adults: An
assessment of dual roles. San Francisco: Family Survival Project.

Feldstein, P. J. Health care economics (3rd ed.). New York: Wiley and Sons.

Fernandez, J. P. (1990). The politics and reality of family care in corporate America.
Lexington, Mass: Lexington Books.

Ferree, M. M. (1987). Family and job for working class women: Gender and class
systems seen from below. In N. Gerstel & H. E. Gross (Eds.), Families and

Work, (pp. 289-301). Philadelphia: Temple University Press.

180
Finley, N. J. (1989). Theories of family labor as applied to gender differences in

caregiving for elderly parents. Journal of Marriage and the Family, 51(1), 79-
86.

Fitting, M., Rabins, P., Lucas, M. J., & Eastham, J. (1986). Caregivers for dementia
patients: A comparison of husbands and wives. The Gerontologist, 26(3), 248-
252.

Franklin, S. T. (1991, November). The impact of acquiring the caregiver role on
employment responsibilities of female caregivers. Poster presented at the annual
meeting of the Gerontological Society of America, San Francisco, CA.

Friedman, D. E. (1986). Eldercare: The employee beneﬁt of the 19908? Across the
Board, 45-51.

Friedman, D. E. (1988). Family-supportive policies: The corporate decision-making
process. In F. E. Winﬁeld (Ed.), The work and family sourcebook (pp. 101-
116). Greenvale, NY: Panel.

Friedman, D. E. & Gray, W. B. (1989, October). A life cycle approach to family
beneﬁts and policies. The Conference Board, Perspectives #19, 1-6.

Frone, M. R. & Rice, R. W. (1987). Work-family conﬂict: The effect of job and
family involvement. Journal of Occupational Behavior, 8, 45-53.

Fullerton, Jr., J. N. (1991). Outlook: 1990-2005: Labor force projections: The baby
boom moves on. Monthly Labor Review, 11, 31-44.

George, L. K. & Gwyther, L. P. (1986). Caregiver wellbeing: A multidimensional
examination of family caregivers of demented adults. The Gerontologist, 25,

253-359.

181
Gibeau, J. (1988). Working caregivers: Family conﬂicts and adaptations of older

workers. In R. Morris & S. Bass (Eds.), Retirement reconsidered. New York:
Springer.

Gilhooley, M. L. M. (1984). The impact of care-giving on caregivers: Factors
associated with psychological well-being of people supporting a demented relative
in the community. British Journal of Medical Psychology, 57(1), 35-44.

Given, C. W. & Given, B. (1989). Caregiver responses to managing elderly patients
at home (Grant #2 R01 AG06584). Rockville, MD: HHS National Institute on
Aging. Principal Investigator, Charles W. Given.

Given, 8., King, 8., Collins, C., & Given, C. W. (1988). Family caregivers of the
elderly: Involvement and reactions to care. Archives of Psychiatric Nursing,
11(5), 281-288.

Given, 3., Stommel, M., Collins, C., King, 8., & Given, C. W. (1990). Responses of
elderly spouse caregivers. Research in Nursing and Health, 13(2), 77-85.

Given, C. W., Stommel, M., & Lin, C. S. (1991, September 25-28). Impact of
patients with cancer upon family caregiver reactions: A longitudinal study.
Presentation at the National Symposium "Respite Care in Chronic, Life-
Threatening, and Terminal Illness” sponsored by the American Institute of Life-
Threatening Illness and Loss, The Foundation of Thanatology, New York, NY.

Glass, G. V. & Hopkins, K. D. (1984). Statistical methods in education and
psychology, (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.

Graaff, J. De V. (1990). Social cost. In J. Eatwell, M. Milgate, and O. Newman

(Eds.) The new Pelgrave, (pp. 251-258). New York: Stockton Press.

182
Grosof, M. S. & Sardy, H. (1985). A research primer for the social and behavioral

sciences. Orlando: Academic Press.

Haber, D. (1989). Health care for an aging society: Cost-conscious community care
and self-care approaches. New York: Hemisphere Publishing.

Hokenstad, M. C. & Johansson, L. (1990). Caregiving for the elderly in Sweden:
Program challenges and policy initiatives. In D. E. Biegel & A. Blum (Eds.),
Aging and Caregiving Theory, Research, and Policy (pp. 254-269). Newbury
Park: Sage.

Hooyman, N. R. (1990). Women as caregivers of the elderly. In D. E. Biegel & A.
Blum (Eds.), Aging and Caregiving Theory, Research, and Policy (pp. 221-241).
Newbury Park: Sage.

Hooyman. N. R. & Ryan, R. (1987). Women as caregivers of the elderly: Catch-22
dilemmas. In J. Figueira-McDonough & R. Sarri (Eds.), The trapped wo-
man/Catch-ZZ in deviance and control (pp.l43-l7l). Newbury Park: Sage.

Horowitz, A. (1985). Family caregiving to the frail elderly. In C. Eisdorfer (Ed.),
Annual review of gerontology and geriatrics, (Vol. 5). New York: Springer.

Hosmer, D. W. & Lemeshow, S. (1989). Applied logistic regression. New York:
Wiley & Sons.

Hu, T., Huang, L., & Cartwright, W. S. (1986). Evaluation of the costs of caring for
the senile demented elderly: A pilot study. The Gerontologist, 26(2), 158-163.

1,600 days go to elder care, survey ﬁnds. (June 6, 1988). National Underwriter,
92(23), (Property/Casualty/Employee Beneﬁts), pp. 6, 88.

Jacobs, P. (1987). The economics of health and medical care. Rockville, MD: Aspen.

183
Kantor, D. & Lehr, W. (1975). Inside the family. New York: Harper Colophon.

Klecka, W. R. (1980). Discriminant analysis. Sage University Paper series on
Quantitative Applications in the Social Sciences, series no, 07-019. Beverly
Hills: Sage.

Kola, L. A. & Dunkle, R. E. (1988). Eldercare in the workplace. The Journal of
Contemporary Social Work, 569-574.

Lewis-Beck, M. S. (1989). Applied regression: An introduction. Sage University Paper
series on Quantitative Applications in the Social Sciences, series no, 07-022.
Beverly Hills: Sage.

Magnus, M. (1988). Eldercare: Corporate awareness but little action. Personnel
Journal, 67(6), 19,23.

Matthews, S. H. & Rosner, T. T. (1988, February). Shared ﬁlial responsibility: The
family as primary caregiver. The Journal of Marriage and the Family 50, 185-
195.

Montgomery, R. J. V. (1989). Investigating caregiver burden. In K. S. Markides & C.
L. Cooper (Eds.), Aging, stress and health (pp.201-218). New York: Wiley.

Neal, M. B., Chapman, N. J., Ingersoll-Dayton, B., Emlen, A. C., & Boise, L.
(1990). Absenteeism and stress among employed caregivers of the elderly,
disabled adults, and children. In D. E. Biegel & A. Blum (Eds.) Aging and
caregiving: Theory, research, and practice (pp. 160-183). Newbury Park: Sage.

Norusis, M. J. (1990a). SPSS/PC+ 4.0 Base Manual. Chicago: SPSS.

Norusis, M. J. (1990b). SPSS/PC+ Statistics 4. 0. Chicago: SPSS.

Norusis, M. J. (1990c). SPSS/PC+ Advanced Statistics 4. 0. Chicago: SPSS.

184
Nye, F. I. (1979). Choice, exchange, and the family. In W. R. Burr, R. Hill, F. I.

Nye, & Reiss, I. L. (Eds.) Contemporary theories about the family: General
theories/theoretical orientations (Vol. 2), (pp. 1-41). New York: The Free Press.

Older Women’s League. (1989). Failing America ’s Caregivers: A Status Report on
Women Who Care. Washington DC: OWL.

Orodenker, S. Z. (1990). Family caregiving in a changing society: The effects of
employment on caregiver stress. Family Community Health, 12(4), 58-70.

Pohl, J., Given, C. W., and Given, B. A. (1991). Caregiver responses to managing
elderly patients at home. Progress Report for Grant #2 R01 AGO6584, HHS
National Institute on Aging. East Lansing, MI: MSU.

Robinson, B. & Thumher, M. (1979). Taking care of aged parents: A family cycle
transition. The Gerontologist, 19(6), 586-593.
Rook, K. S. (1987). Reciprocity of social exchange and social satisfaction among older
women. Journal of Personality and Social Psychology, 52(1), 145-154.
Sakauye, K. M. (1989). Ethnic variations in family support of the ﬁail elderly. In M.
Z. Goldstein (Ed.) Family involvement in the treatment of the frail elderly (pp.
63-106). Washington DC: American Psychiatric Press.

Scharlach, A. E. (1989). A comparison of employed caregivers of cognitively impaired
and physically impaired elderly persons. Research on Aging, 11(2), 225-243.

Scharlach, A. E. & Boyd, 8. L. (1989). Caregiving and employment: Results of an
employment survey. The Gerontologist, 29(3), 383-387.

Scharlach, A. E., Lowe, B. F., & Schneider, E. L. (1991). Elder care and the work

force. Lexington, MA: Lexington.

 

185
Scharlach, A. E, Sobel, E. L., & Roberts, R. E. L. (1991). Employment and caregiver

strain: An integrative model. The Gerontologist, 31(6), 778-787.

Sidel, R. (1986). Women and work. Women and children last: The plight of poor
women in aﬁluent America (pp. 48-76). New York: Viking.

Stoller, E. P. & Pugliesi, K. L. (1989a). Other roles of caregivers: Competing
responsibilities or supportive resources. Journal of Gerontology, 44(6), 8231-
238.

Stoller, E. P. & Pugliesi, K. L. (1989b). The transition to the caregiver role: A panel
study. Research on Aging, 11(3), 312-330.

Stommel, M., Given, C. W., & Given, B. A. (1990). Depression as an overriding
variable explaining caregiver burden. Journal of Aging and Health, 2(1), 81-102.

Stone, R., Cafferata, G. L., & Sangl, J. (1987). Caregivers of the fragile elderly: A
national proﬁle. The Gerontologist, 27(5), 616-631.

Stone, R. I. & Short, P. F. (1990). The competing demands of employment and
informal caregiving to disabled elders. Medical Care, 28(6), 513-526.

The family and medical leave act. (1991, April). Congressional Digest, pp. 98-109.

The Travelers Companies. (1988). The Travelers employee caregiver survey. In F. E.
Winﬁeld (Ed.), The work and family sourcebook (pp. 239-249). Greenvale, NY:
Panel.

Touliaros, J. & Compton, N. H. (1988). Research Methods in Home Economics.

Ames, IA: Iowa State University Press.

186
U.S. Bureau of the Census. (1986). Women and the American economy (Current

Population Reports, Series P-23, No. 146). Washington DC: U.S. Government
Printing Ofﬁce.

Waldo, D. R., Sonnefeld, S., McKusick, D. R., & Amett HI, R. H. (1989). Health
expenditures by age group, 1977 and 1987. Health Care Financing Review,
10(4), 111-120.

Walker, A. J., Martin, S. S. K., Jones, L. L. (1992). The beneﬁts and costs of
caregiving and care receiving for daughters and mothers. Journals of Gerontolo-
gy, 47(3), 8130-139.

Warshaw, L. J., Barr, J. K., Rayman, 1., Schachter, M., & Lucas, T. G. (1986).
Employer support for employee caregivers. New York: The New York Business
Group on Health.

White-Means, S. I. & Thornton, M. C. (1990). Labor market choices and home health
care provision among employed ethnic caregivers. The Gerontologist, 30(6), 769-
775.

Willett, J. B. (1988). Questions and answers in the measurement of change. In E. Z.
Rothkopf (Ed.), Review of Research in Education (pp. 345—422). Washington
DC: American Educational Research Association.

Winﬁeld, F. E. (1988). Workplace solutions for women under eldercare pressure. In
F. E. Winﬁeld (Ed.), The work and family sourcebook (pp. 275-284). Greenvale,

NY: Panel.

187
Wisensale, S. K. & Allison, M. D. (1988). An analysis of 1987 state family leave

legislation: Implications for caregivers of the elderly. The Gerontologist, 28(6),
779-785.
Young, R. F. & Kahana, E. (1989). Specifying caregiver outcomes: Gender and
relationship aspects of caregiver strain. The Gerontologist, 29(5), 660-666.
Zarit, S. H., Todd, P. A., & Zarit, J. M. (1986). Subjective burden of husbands and

wives as caregivers: A longitudinal study. The Gerontologist, 26(3), 260-266.

APPENDICES

APPENDIX A

APPENDIX A

Instrument

Caregiver Employment Questionnaire (selected items)
1. Were you employed outside the home preyiqus to your caregiving responsibility?

_ YES (1)
_ N0 (2)

if yes, _ full-time (1) _ part-time (2)

Note: At Time 2, the caregiver is asked if there has been a change in employment status
in the last three months and if so, how it has changed.

2. What is your current employment status?

employed full-time (l)

employed part-time (2)

self-employed (3)

retired (4)

leave of absence (5)

unemployed; laid off/between jobs/disability (6)

not employed for pay outside the home; homemaker (7)
quit work to care for (8)

 

3. There are some ways that caregiving can affect or interfere with your work or

employment. Please answer "yes" or " no for each item W
currenﬂx malarial

Has caregiving affected or interfered with your work:
a. Caused you to be late for work/or leave work early?

_ YES (1) _ NO (2)

188

4.

189

b. Caused you to miss days of work without pay?
_ YES (1) __ N0 (2)

c. Caused you to take sick/personal days?

_ YES (1) _ NO (2)

d. Caused you to change the hours you work (shift)?
__ YES (1) _ NO (2)

e. Caused you to turn down a job or promotion?
_ YES (1) __ NO (2)

f. Kept you from looking for a job, or a better job?
_ YES (1) _ NO (2)

g. Affected your work in other ways not mentioned?
__ YES (1) _ NO (2)

Describe:

 

h. Caused you to take a leave of absence?
_ YES (1) _ NO (2)

i. Caused you to quit or take early retirement?

_ YES (1) _NO (2)

In the last three months, how many days of work have you missed in order to eare
for your relative?

(Write in Number of Days) __ Not Applicable

190

Sociodemographic Information

. Caregiver is: _ Male (l)_ Female (2)
Patient is: _ Male (l)_ Female (2)
Caregiver Birth Date _ month _ day _ year
Patient Birth Date _ month __ day °__ year

. What is your (caregiver) marital status? (CHECK ONE)

_ Single, never married (1)
_ Married/ remarried (2)
_ Divorced (3)

_ Widowed (4)

_ Separated (5)

. What is your relationship to ? You are his/her:
(Check one)

_ Spouse (1)

_ Parent (2)

Daughter/Son (3/4)
Daughter-in-law/Son-in-law (5/6)
Sister/Brother (7/ 8)
Sister/brother-in-law (9/10)
Granddaughter/ son (1 1/12)
Niece/Nephew (13/14)

_ Aunt/Uncle (15/16)

_ Other relative (l7) (specify)

_ Friend or companion with whom you live (18)
_ Other non relative (19)

What is your highest level of education? (Check one)

Completed grade school or less (1)

Completed some high school (2)

Completed high school (3)

Completed some college or technical training (4)
Completed college (5)

Completed some graduate or professional school (6)
Completed graduate or professional degree (7)

 

. Do you and

191

currently live in the same household? (If "yes", caregiver was

 

asked about prior)

Yes (1) _ No (2)

. Considering all sources of income for all members of your household, please indicate

your gross income, before taxes, as you would report on your Federal Tax Forms.
(Select one category)

__ 0 - 1,999 (l) _ 22,000 - 25,999 (9)

_ 2,000 - 3,999 (2) _ 26,000 - 29,999 (10)
_ 4,000 - 5,999 (3) _ 30,000 - 34,999 (11)
_ 6,000 - 7,999 (4) _ 35,000 - 39,999 (12)
_ 8,000 - 9,999 (5) _ 40,000 - 44,999 (13)
_ 10,000 - 13,999 (6) _ 45,000 - 49,999 (14)
_ 14,000 - 17,999 (7) _ 50,000 - 54,999 (15)
_ 18,000 - 21,999 (8) _ 55,000 - 59,999 (16)

_ 60,000 and over (17)

192
Family Network Questionnaire

The next set of questions asks about the families of you and the person for whom you
provide care.

First, we’d like you to identify all living adult relatives of the patient. We’d like you to
include parents, sisters, brothers, spouse, and children. For each relative, we’d like you
to give us the person’s initials, the relationship to the patient, and how far away they live
from the patient in approximate miles.

(THE INTERVIEWER FILLS OUT THE FIRST THREE COLUMNS OF THE TOOL
AT THIS POINT WITH THE INITIALS, RELATIONSHIP TO THE PATIENT, AND
DISTANCE IN MILES FOR EACH PERSON IDENTIFIED AS A LIVING ADULT
RELATIVE OF THE PATIENT)

Next, I’d like you to think about each one of these people in terms of how close they are
to the patient and how close they are to you. By closeness, we mean the strength of the
emotional tie between the patient and the relative, and you and the patient’s relative.
Please rate the closeness of the relationship on a scale of 1 to 5 where 1 stands for a
relationship that is not at all close and 5 stands for a relationship which is extremely
close. Then I’ll ask you about how much help each of the relatives helps with
caregiving. The answers you may choose from are NONE OR VERY LITTLE, A
LITTLE, SOME, QUITE A BIT, or A GREAT DEAL.

(THE INTERVIEWER CUES RESPONDENT WITH INITIALS AND RELATIONSHIP
AND FILLS OUT THE REMAINING COLUMNS)

CLOSENESS TO PATIENT CLOSENESS TO HELP PROVIDED
CAREGIVER
1 = not at all close 1 = not at all close 1 = none or very little
5 = extremely close 5 = extremely close 2 = a little
3 = some
4 = quite a bit

5=agreatdeal

193

Only the number in the caregiver’s family network and the amount of help
provided by each member was used in the study. At Time 2, the caregiver also is asked
to estimate both the number of hours per week that each person assisted in caregiving
and the total number of hours per week that others provided help. In addition, there are
questions about changes in the help received (stopped or decreased and by whom; started
or increased and by whom). Finally, the caregiver is asked whether help from the family

increased, stayed about the same or decreased.

194
Social Assistance (of the family) Scale

Now I’m going to ask you about the assistance that you receive from We to care
for your relative (relationship or name of the patient). By other people, I mean friends
or relatives whether you pay them or not.

For the next set of questions, I will ask "How often since discharge, over the past month
has anyone in your family given you assistance in the following ways," and I will read
a list of activities. I would like you to please pick from one of four categories of
responses: You can choose from "Rarely or None of the Time," ”Some of the Time,"
”Most of the Time,” or "Almost all of the Time.” I will repeat those categories again
if necessary.

1. How often since discharge or over tho past month, HAS ANYONE IN YOUR
FAMILY given you assistance in the following ways: (CHECK ONE FOR EACH
TASK)

SCALE
1 = rarely or none of the time
2 = some of the time
3 = most of the time
4 = almost all of the time

la. Help with physical care

1b. Spend time keeping your relative company

1c. Stayed with so that you could do something else for a few hours.
1d. Given YOU emotional support or encouragement

1e. Helped with transportation; for either you or

If. Helped YOU with money or other material goods

1g. Checked on YOU to be sure that you were alright

1h. Provided/encouraged diversional activities for _such as cards or Scrabble

 

Note: The core study also utilizes the questions to assess the assistance of friends. This
study used only family responses. The questions are identical at Time 1 and Time 2.

195
Caregiver Involvement Instrument (selected items)
The next set of questions addresses the PRESENT level of performance for the person
you care for on a number of activities and the way YOU help him/her. For each item,
please choose the response that most closely describes the patient’s PRESENT condition
and how you help him or her.
1. DRESSING
1a. With regard to dressing, would you say ...(check one)
_ IS INDEPENDENT
_ NEEDS SUPERVISION ONLY
_ NEEDS SOME PHYSICAL HELP
_ NEEDS TOTAL PHYSICAL HELP

_ IS NEVER DRESSED (always wears bed clothing)

1b. How frequently do YOU help your relative with dressing?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY
2. EATING
2a. With regard to eating, would you say ...(check one)

_ IS INDEPENDENT

_ NEEDS SUPERVISION ONLY

_ NEEDS SOME PHYSICAL HELP
_ NEEDS TOTAL PHYSICAL HELP

_ NOT APPLICABLE (needs tube feedings, IV’s)

196
2b. How frequently do YOU help your relative with eating?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY
3. BATHING
3a. With regard to bathing, would you say ...(check one)

_ IS INDEPENDENT
_ NEEDS SUPERVISION ONLY
_ NEEDS SOME PHYSICAL HELP

_ NEEDS TOTAL PHYSICAL HELP

3b. How frequently do YOU help your relative with bathing?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

4. WALKING INSIDE THE HOUSE

4a. With regard to walking inside the house, would you say ...(check one)

 

__ IS INDEPENDENT

_ NEEDS SUPERVISION ONLY

_ NEEDS SOME PHYSICAL HELP
_ NEEDS TOTAL PHYSICAL HELP

_ UNABLE TO WALK

197
4b. How frequently do YOU help your relative with walking?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

5 . TOILETING

 

5a. With regard to toileting, would you say ...(check one)
_ IS INDEPENDENT

_ NEEDS SUPERVISION ONLY

_ NEEDS SOME PHYSICAL HELP

_ NEEDS TOTAL PHYSICAL HELP

_ NOT APPLICABLE (has catheter and colostomy)

5b. How frequently do YOU help your relative with toileting?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY
6. TRANSFERRING IN/OUT OF BED
6a. With regard to transferring in/out of bed, would you say _ ...(check one)
_ IS INDEPENDENT
_ NEEDS SUPERVISION ONLY
_ NEEDS SOME PHYSICAL HELP
_ NEEDS TOTAL PHYSICAL HELP
_ REMAINS BEDFAST

6b. How frequently do YOU help your relative with transferring in/out of bed?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

198
7. COOKING/ PREPARING MEALS

7a. How much help does presently nwd with cooking? Does he/she need:

(check one)

 

_ NO HELP

_ SOME HELP

_ TOTAL HELP (doesn’t participate, but has done so in the past)
_ TOTAL HELP (doesn’t participate and never has)

_ NOT APPLICABLE (patient has tube feedings, IV’s)

7b. How frequently do YOU help your relative with cooking or cook for him or her?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY
8. HOUSEWORK

8a. How much help docs presently need with housework? Does he/she
need: (check one)

_ NO HELP

_ SOME HELP

_ TOTAL HELP (doesn’t participate, but has done so in the past)
_ TOTAL HELP (doesn’t participate and never has)

8b. How frequently do YOU help your relative with housework or do housework for
him or her?

NEVER ONCE A SEVERAL DAILY ' SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

199
9. SHOPPING

9a. How much help docs presently nwd with shopping? Does he/she nwd:
(check one)

_ NO HELP

_ SOME HELP

__ TOTAL HELP (doesn’t participate, but has done so in the past)
_ TOTAL HELP (doesn’t participate and never has)

9b. How frequently do YOU help your relative with shopping or shop for him or

her?
NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY
10. LAUNDRY
10a. How much help docs presently need with laundry? Does he/she need:

(check one)

_ NO HELP

_ SOME HELP

_ TOTAL HELP (doesn’t participate, but has done so in the past)
_ TOTAL HELP (doesn’t participate and never has)

10b. How frequently do YOU help your relative with laundry or laundry for him
or her?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

200
11. TRANSPORTATION

11a. How much help docs presently need with transportation? Does he/she
need: (check one)

_ NO HELP

__ SOME HELP

_ TOTAL HELP (doesn’t participate, but has done so in the past)
__ TOTAL HELP (doesn’t participate and never has)

11b. How frequently do YOU help your relative with transportation?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

12. MONEY MANAGEMENT

12a. How much help does
he/she need: (check one)

presently need with money management? Does

 

_ NO HELP

_ SOME HELP

_ TOTAL HELP (doesn’t participate, but has done so in the past)
_ TOTAL HELP (doesn’t participate and never has)

12b. How frequently do YOU help your relative with money management or do
money management for him or her?

NEVER ONCE A SEVERAL DAILY SEVERAL
WEEK/LESS TIMES A WEEK TIMES A DAY

Note: Responses are: 0 = never, 1 = once a week or less, 2 = several times a week,
3 = once a day, 4 = several times a day. In the core study, the involvement of
other people (family and friends) is also assessed with this section.

Tlnudu

Ed

(u

(

(NI ( (I

201

The next questions deal with activities of health care (HCA). The caregiver is asked if
the relative needs help with HCA. The caregiver indicates how frequently she helps with
the required care recipient activities.

Responses: 0 = never, 1 = once a week or less, 2 = several times a week (2-6), 3 =
daily, and 4 = several times a day.

URINARY CATHETER/CATHET ER CARE 0 1 2 3
OXYGEN ADMINISTRATION 0 1

IV, HICKMAN, BROVIAC

CATHETER CARE/DRESSING 0 1 2 3
IV MEDICATIONS/FLUIDS/FEEDINGS 0 1

TUBE FEEDINGS OR IV FEEDINGS 0 l 2 3
INJECTIONS (PAIN MEDICATIONS,

INSULIN) O 1 2 3
SPECIAL EXERCISES/ PHYSICAL THERAPY O I

CARE OF ULCERS/BEDSORES O 1 2 3
SKIN CARE (SPECIAL CLEANSING

LOTIONS) O l
COLOSTOMY/COLOSTOMY CARE 0 I

CARE OF POST OPERATIVE

INCISION/WOUND

ORAL MEDICATIONS

NASOGASTRIC TUBE AND CARE
INCONTINENCE OF URINE
INCONTINENCE OF STOOL
TRACHEOSTOMY/TRACHEOSTOMY CARE
RESPIRATOR/CARE OF RESPIRATOR
SUCTIONING

OOOOOOOO
r—tr—tr—sr—r—r—r—nr—A
NNNNNNNN
wmuuuwmw
##ﬁ-ﬁ-ﬁhh-k

Note: In the core study, this section also is used to assess the caregiver’s feelings of
competency to help with the required procedures and to measure how often others help.

202

The next set of questions is about how much time you spend providing care for your
relative.

1. PHYSICAL ASSISTANCE
1a. In a usual day how many hours do you spend providing physical assistance to
your relative? By physical assistance, we mean physically helping your relative
with activities such as eating, bathing, treatments, etc. (WRITE IN NUMBER)

_ NUMBER OF HOURS/DAY

2. SUPERVISION TIME
2a. In a usual day (excluding night hours), how many hours do you spend
supervising your relative? By supervision, we mean being with your relative to
make sure he/she is safe (WRITE IN NUMBER)
_ NUMBER HOURS/DAY

Note: At Time 2, the caregiver also is asked to indicate if she is providing more care,
less care, or the same amount of care compared to three months ago.

APPENDIX B

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a -8. 42.. I Jew Itm I New. - --....a. .....a. let-ext- Hum. lit-me.- -i.e,.-.t l ..m...: -l 2.. www.91-
ﬁ 8.. 2.- 8.- 8.- .2. ..8. ..8. .2. .2. .2. 2.- 2.- .2: .o. 2.5 2 L
‘ 8.. ..ﬁ. 8.- 8.- 8. ...8. ...8. ..ﬁ. ..2. ..8. 8.- ..8. .2. 3.3.. 25...:
_ 8.. .....n. ...a. 8. 8. 2. 2. 8. .2. ...on. ..2. 8.- :8. .... < .2
W 8.. 8. 8. 8. 2. 2. 8. 8. ...8. 2. 8.- 2. 32". a.
ﬁ 8.. 8. 8. ... ... :2. 8. 8.- 8. 8. n. .- E. ... ...
_ 8.. ...S. ...8. .32.. ...8. ...on. 8. .2.- ..8: ... me: .2 _
8.. ...S. ...8. :8. ...8. 8. :2.- ..2: 8.- 9. a: 0 “
V 8.. ...8. ...R. ...8. 8. 8. .2.- 8. .34. ..
u 8.. ...3. ...... 8.- ...- 8.- 8.- <2. .-
_ 8.. ...S. 2.- 8.- 2.- 8.- .9: .e
. 8.. 8. 8.- ..o..- 8.- .9. .n
8.. ...... 8. ... on... ..o .4
8.. ...8. :8.- e58... .n
8.. .n..- B .N
8.. a? no ..
v. n. a. .. 2 a a .. e m w ... u . ..E-p

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8...: 8.3.2.5

m “Gamma.

... 25 u ... ......- ... 2.....- u .. ...... .. 2...... n ...... ...... ......a. .o .525 n ... < 383.2.
3.2.... u 32 n. .20.. 2.28 .8 55:3... H :m .n. ”c2232.... .0 2.5.. H m a... .28 32%.... ..o 8.5.. H Um 2.. 352023...
.9328 .52 n 44.... ”823.2. 0.8 ...—.8.— " <0... $22. 2...... .0 822.2. 358.52.. n .55 35>: >3... .8 8.22.2. u
AQ< ”own 2.228.. 9.8 .I- ow< m0 628... 823.8 u 2.82.. 2.2.833 533.8 u ..m ”own 829.8 ﬂ ow< mo ”2.2.25.3...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.89. H 3.2.. 3c. ﬂ ...... ”no. H ... .302

8.- .... 2. we: .2 _
__ 8.. ...... ...... ...... ...... ...- 8.- .ﬁ: 8. 9...: ...
__ 8.. ...8. ....... ...»... 8. 2: 2.- 8. -.-.<...
8.. ...3. ...8. ...: .2: B.- 8. <2. .8
8.. 3.2.. 8.- 8.- 2.- 8. 3......
8.. z. 8.- .2.- 8. .9. .n
8.. 2. 8. 8. a... mu ..
8.. ...u. :8.- .58... ...
8.. :8.- ..m .N

8.. ...... no. _—

u. .. 2 a a .. o n v n a . «2...... _—

 

 

APPENDIX C

205

8.2.22.3 .o 2.2. u m 2: .28 .882... ..o 2.5.. H Um 2.. ”2.02023...
.9328 .52 u .5... .22. 2.8.... .o 8.532.. H mm .2 ...—2. $2.... .8 259.... H :m < ...-83.2. 3.22 H 32 m ”own 2.2202
28 u ow< mo 828... .6388 u 2:8... X28» .5 22.828 .3228 H ..m new“ .0288 n.- ow< wU .m:o_.~_>2nn<

N 25... .a 32.2... .2. 2.22.28... u <2 .202

 

 

on o 2 2 .h mmﬁ
vm o a... mm .N
hm o no.3 hm . 2 N
v . hm . co. 2
<2 <2 <2 <2
<2 <2 <2 <2
<2 <2 <2 <2
<2 <2 <2 <2
<2 <2 <2 <2
<2 <2 <2 <2
v o m. . _ ho. .
.82 ...:2 Gm x

 

82nz. m 22,.-

 

8 o .3 8.. m ....
z. o 88 e: o. 2..
an o 2.2 8.8 -.-.<
8.. . 2.. 8.. ... ..
2 c an... 8... .... <
2 o 2 .N SN 32 n.
8.8 ...8 2... 8.8 um... .6
8m... o8.m 2.8.x... 8.28.8 2.8... .6
8 c .3 8.2 ..m. 8
8.8 8.2 8... 8.8 uw< mo
8 8 8.. 8.. 8.5.8...
5.2 5.2 mm x .2?

 

Gmmuz. . as...

 

8322.3 .922..— 28 2.82

U Nun—22%;.

APPENDIX D

APPENDIX D
UCRIHS Approval Letter

mcmcam STATE umvensin'

 

mm m we! "mm“ uni ”MAIN! no lasts!“- 0 mount“ . clue-Mona
“I! m n 00 no hum! .m “”00!

April 20. 1992

Susan T. Franklin
2066 Hillside Dr.
Huskegon. Kl L9éél

RE: THE lNFLUlZXCE OF ACQUIRING THE FAMILY ELDER CARE ROLE 0!! FE‘MLE E‘iPlDYHENT
ADA?757103. 223 #92-157

Dear 3s. franklin:

I an pleased to advise that because of the nature of the proposed research. it
was eligible for expedited review. This process has been completed. the rights
and welfare of the human subjects appear to be adequately protected. and your
project is therefore approved.

You are reninded that UCRIHS approval is valid for one calendar year. If you
plan to continue this project beyond one year. please nake provisions for
obtaining appropriate UCRIHS approval prior to April 15. 1993.

Any changes in procedures involving human subjects trust be reviewed by the UCRIHS
prior to initiat.on of the change. UCRIRS nus: also be notified promptly of any
problens (unexpected side effeCts. complaints. etc.) involving human subjects
during the course of the work.

Thank you for bringing this project to our attention. If we can‘be of any future
help. please do not hesitrte to let us know.

1.142%

David E. wright. 3h. .. Chair
University Commit. - on Research
Involving Huzan Subjects (UCRIHS)

Sincerely,

   
  

cc: Dr. Sarbara Ames

V \l u a 2"vun r A. to». ! .‘aa' 'kfﬂs om inst-Ian .-

206

‘Mulloujuﬂmwoo“