MATERNAL MOTIVATIONS AND CRIMINAL DESISTANCE
By
Elizabeth A. Adams

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Criminal Justice-Master of Science
2014

ABSTRACT
MATERNAL MOTIVATIONS AND CRIMINAL DESISTANCE
By
Elizabeth A. Adams
Existing literature on reentry and community supervision experiences of mothers requires
expansion. An age-graded informal social control life course theory perspective provided the
framework for studying whether several indicators of bonds to children explained women’s
desistance from crime. The focus on bonds to children addressed a concern of feminist theorists,
who criticize life course theory for omitting bonds that are salient influences on women’s
behavior. Because prior research from inmate surveys and qualitative assessments suggest that
parental concerns are especially frequent, and they may be stressful for women offenders, stress
from parenting is also studied as an influence on desistance. In order to statistically examine
these distinct parenting-related challenges and protective factors, the connections of women’s
recidivism outcomes to maternal dispositions and motivations are investigated using a
longitudinal design on a large sample of women offenders. Based on transcripts from three
waves of interviews from a larger project’s data set, and guided by life course theory and
feminist criticisms of it, the current research examined parent-related factors and several
measures of recidivism for 193 mothers on community supervision. This study expanded
knowledge regarding maternal motivations, involvement with children, and stress from
parenting, in relation to criminal desistance processes conceptualized in the theory of age-graded
informal social control.

Dedicated to my mother and father for endlessly inspiring me by demonstrating strength and
positivity through life’s many obstacles.

iii

ACKNOWLEDGEMENTS

It is an honor to express my appreciation for all of those who encouraged and helped me
maintain balance during my first year in graduate school. I would not have been able to finish my
required courses and thesis without the support of numerous individuals and agencies.

First, I would like to acknowledge the funding for this project was provided by the National
Science Foundation (SES-1126162). Additionally, I’d like to thank those that facilitated this
important research on women under community supervision. The current examination would not
have been possible without the dedication and hard work of the investigators, graduate research
assistants, interviewers, and undergraduate workers. I also must express my appreciation for
Michigan State University’s School of Criminal Justice for aiding me in developing a career plan
to fit my interests by providing various opportunities for learning in the classroom as well as in
the community.

In addition, I’d like to thank my thesis committee members, Drs. Jennifer Cobbina and Mahesh
Nalla, for providing me with valuable input and support as I worked on my thesis. I also want to
extend my deepest gratitude for my thesis chair, Dr. Merry Morash, for challenging me to think
critically and teaching me how to utilize a mixed-methods approach for my thesis. Your daily
involvement with students and research is nothing short of inspirational. I hope to emulate your
work ethic and compassion for others throughout my life and career.

Additionally, I am pleased to acknowledge that several professors outside of my committee
offered me opportunities and encouraged me during my first year as a graduate student. I am
iv

thankful to Drs. Meghan Hollis, Jesenia Pizarro, and Ed McGarrell for offering me addtional
research experience and providing me with strategies to examine and address community issues
while collaborating with law enforcement personnel.

Finally, I am thankful for my family as they motivate and allow me to make my own decisions as
I navigate life. My aunt, Annette Horan, has assisted me through challenges and served as a
listening ear during many of my contemplations regarding my personal and professional life. In
addition, my parents have always supported and validated my life choices. Their endless
assurance has enabled me to confidently and creatively examine and attempt to address
controversial societal issues. The completion of my thesis was one vital step in my journey as a
researcher – this endeavor was made possible and exciting through the inspiration and
motivation I’ve received from others.

v

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................................... ix
Chapter 1: Rationale for the Study..............................................................................................1
Challenges of Incarceration and Reentry for Mothers ..............................................................1
Challenges for Mothers on Probation and Parole .....................................................................3
The Importance of Identifying Influences on Women’s Desistance ........................................4
Chapter 2: Review of Existing Literature ……………………………......................................7
Literature on Life Course Theories and Age Graded Social Control Theory............................7
Life Course Examinations of Female Offenders ....................................................................12
Qualitative Research using Age-Graded Social Control………………………………...12
Quantitative Research using Age-Graded Social Control……………………………….15
Conceptualizations of Recidivism for Mothers………………………………….15
Conceptualizations of Informal Social Control for Mothers…………………….16
Social Bonds to Children……………………………………………...…17
Stress from Parenting…………………………………………………….20
Relevant Variables in Research on Women’s Criminality and Recidivism……..22
Risks for Recidivism……………………………………………………..22
Age……………………………………………………………….22
Demographics, Neighborhood Contexts, Relationships, and
Peers……………………………………………………………...23
Educational and Occupational History…………………………..24
Criminal History, Substance Abuse, Mental Health Status, and
Exposure to Abuse……………………………………………….25
Conclusion ……………………………………......................................................................25
Hypotheses ……………………………………......................................................................26
Chapter 3: Methodology of the Current Study.........................................................................28
Sample .....................................................................................................................................29
Quantitative Data.....................................................................................................................30
Dependent Variables……………………………………………………………….…….30
Independent Variables.......................................................................................................31
Maternal Motivations…………………………………………………………….31
Social Bonds to Children………………………………………………………...31
Stress from Parenting…………………………………………………………….31
Control Variables ……………………………………………….......…………………. 32
Women’s Total Needs…………………………………………………………...32
Predictive Risks………………………………………………………….32
Employment and Financial Needs……………………………….32
Anti-Social Peers………………………………………………...33
Criminal History…………………………………………………33
Substance Abuse History………………………………………...34
Mental Health Disorders…………………………………………34
vi

1. Mental Illness…………………………………….......34
2. Current Depression and Anxiety……………………..34
3. Current Psychosis…………………………………….34
Interviewer’s indication of Women’s Anger and Hostility………35
Interviewer’s Indication of Women’s Anti-Social Attitude……...35
Women Residing in Un-Safe Homes…………………………….35
Predictive Strengths……………………………………………………...36
Educational Strengths……………………………………………36
Self Efficacy……………………………………………………..36
Family Support…………………………………………………..36
Age and Race…………………………………………………………………….37
Supervision Status………………………………………………………………..37
Neighborhood Context…………………………………………………………...37
Coding of the Qualitative Data ...............................................................................................37
Analysis of the Quantitative Data............................................................................................38
Chapter 4: Results .......................................................................................................................41
Qualitative Findings ................................................................................................................41
Maternal Motivations.…....................................................................................................41
Attachment to Children…………………………………………………………..41
Being a Good Parent……………………………………………………………..43
Being a Better Parent than Before……………………………………………….45
Proving to be a Good Parent……………………………………………………..46
Parental Role Assertion…………………………………………………………..49
Custody Motivations……………………………………………………………..50
Routine Activities with Children………………………………………………...52
Stress from Parenting .......................................................................................................53
Quantitative Findings ..............................................................................................................56
Bivariate Analysis ….........................................................................................................56
Correlations among Independent
Variables…………………………………………………………………………56
Correlations among Dependent Variables……………………………………….60
Correlations of Independent with Dependent Variables…………………………62
Regression Analysis ..........................................................................................................66
Alcohol Use Based on Supervising Officer Case Notes………………………....67
Number of Violations Based on Supervising Officer Case Notes……………….69
Women’s Self Reported Substance Abuse at the First Interview………………..72
Drug Use Based on Supervising Officer Case Notes…………………………….75
Official Arrest Records…………………………………………………………..77
Official Conviction Records……………………………………………………..80
Findings for the Control Variables………………………………………………84
Chapter 5: Discussion and Conclusion .....................................................................................87
Limitations...............................................................................................................................92
APPENDIX ..................................................................................................................................95
vii

REFERENCES ..........................................................................................................................100

viii

LIST OF TABLES
Table 1 - Attachment to Children .………………………………………………………….…...41
Table 2 - Being a Good Parent ……………………….………………………………….............43
Table 3 - Being a Better Parent than Before ………………......................................……..….…45
Table 4 - Proving to be a Good Parent ......…………..…………………………...….........……..46
Table 5 - Parental Role Assertion ...…………………………………………………...……..….49
Table 6 - Custody Motivations …………………...……………………………...……..……….50
Table 7 - Routine Activities with Children ………………….……………………….......……...52
Table 8 - Correlations among Dichotomous Independent Variables ...………………...……..…57
Table 9 - Correlations among Continuous Independent Variables ……………………...….…...59
Table 10 - Correlations among Dependent Variables ………………………………...…...…….61
Table 11 - Correlations of Dichotomous Independent Variables with Dependent Variables……63
Table 12 - Correlations of Continuous Independent Variables with Dependent Variables…..….65
Table 13 - Regression of Supervising Officer’s Case Notes of Women’s Alcohol Use while
under Supervision on Control Variables and the Dichotomous Independent Variables …….….68
Table 14 - Regression of Supervising Officer’s Case Notes of Women’s Alcohol Use while
under Supervision on Control Variables and the Continuous Independent Variables …….....….69
Table 15 - Regression of Supervising Officer’s Case Notes of Women’s Violations while under
Supervision on Control Variables and the Dichotomous Independent Variables …….……...….71
Table 16 - Regression of Supervising Officer’s Case Notes of Women’s Violations while under
Supervision on Control Variables and the Continuous Independent Variables ……………...….72
Table 17 - Regression of Self-Reported Substance Abuse at the First Interview on Control
Variables and the Dichotomous Independent Variables …………………………………...……74
Table 18 - Regression of Self-Reported Substance Abuse at the First Interview on Control
Variables and the Continuous Independent Variables ………………………….….....................75
Table 19 - Regression of Supervising Officer’s Case Notes of Drug Use while under Supervision
on Control Variables and the Dichotomous Independent Variables ……………………....…….76
ix

Table 20 - Regression of Supervising Officer’s Case Notes of Drug Use while under Supervision
on Control Variables and the Continuous Independent Variables ………………………...….…77
Table 21 - Regression of Official Arrest Records on Control Variables and the Dichotomous
Independent Variables ….........................................................................................................….79
Table 22 - Regression of Official Arrest Records on Control Variables and the Continuous
Independent Variables …………………………………………………………………….…….80
Table 23 - Regression of Official Conviction Records on the Control and the Dichotomous
Independent Variables ………………………………………………………………..……...….82
Table 24 - Regression of Official Conviction Records on the Control and the Continuous
Independent Variables …….……...……………………………………………………………..83
Table 25 - Findings of Significance for Controls in Dichotomous and Continuous Regressions
……………………………………………………………………………………………………84
Table 26 - Findings of Significance for IVs in Dichotomous and Continuous Regressions….....88

x

Chapter 1: Rationale for the Study
Offender populations in the United States have been marked by slight reductions over the
last few years (Bureau of Justice Statistics, 2013 a), but the number of females involved in the
criminal justice system remains higher than in previous decades (Salisbury & Van Voorhis,
2009). In 2012 females comprised 7% of those incarcerated, and accounted for 24% of
probationers and 11% of parolees (Glaze & Hererrman, 2013; Maruschak & Bonczar, 2013).
While females still comprise smaller percentages of offender populations than do men, from
1995 to 2012 their growth rate of imprisonment in jail and prison (80%), and on community
supervision (42%) was approximately double the growth rate for male imprisonment in jail and
prison (44%) and on community supervision (19%) (Bureau of Justice Statistics, 1997; 2013b;
2013 c; 2013d). Although females continue to experience disproportionate growth in offender
populations, criminological research and theoretical development remains inadequate in
assessing gender-specific risk and protective factors for women (Salisbury & Van Voorhis,
2009), particularly the types of bonds that contribute to their desistance from crime.
The next section briefly reviews literature suggesting why establishing bonds to children
might be problematic for women on parole and probation. Specifically, women’s experiences in
prison may make it difficult to maintain bonds, and all women supervised in the community may
face special challenges maintaining ties to their children.

Challenges of Incarceration and Reentry for Mothers
Research on offender populations has revealed distinct negative effects related to the
increasing numbers of mothers involved in the criminal justice system. From 1991 to 2007, there
was an 80% increase in children with incarcerated parents (Glaze & Maruschack, 2008), and this
was arguably attributable to the 75-80% of women that left behind an average of 2.11 minor
1

children upon incarceration (Eros, 2001; Greenfield & Snell, 1999). In 2007, 60% of
incarcerated women resided with children who relied on them for care prior to imprisonment
(Glaze & Maruschack, 2008). A related issue for mothers was finding placement for their
children upon incarceration, as they have been found to be far less likely to be able to rely on the
other biological parent (37%) for care during incarceration in comparison to incarcerated men
(88%) (Glaze & Maruschack, 2008). In effect, women more often have had to rely on
grandparents, other family members, and even foster care for child placement (Glaze &
Maruschack, 2008; Mumola, 2000). Qualitative research has shown that mothers reported fear
for the safety and protection of their children when placed with grandparents susceptible to
manipulation from children, anti-social adult role models, or complete strangers, all of whom
may reside in dangerous neighborhoods (Arditti & Few, 2008; Muhammad, 2012). In addition to
mothers’ stress over child safety, poor relationships or no relationships between the caregiver
and the incarcerated parent has been associated with decreased visitation, as caregivers may limit
children’s relations with incarcerated or offender parents (Arditti & Few, 2008; Muhammad,
2012).
Abrupt lapses in communication due to incarceration or commitments to in-patient
treatment facilities may be confusing and upsetting for both the child (Muhammad, 2012) and
the removed parent (Arditti & Few, 2008; White, 2012). Evidence exists indicating inmates’ that
reported fewer visits from their children experienced higher rates of parental stress, depression
(Poehlmann, 2005), and recidivism than those that were visited more often (Cochran, 2013).
Additionally, incarcerated mothers may rarely, if ever, receive visits from their children during
incarceration due to distances of facilities (Krisberg & Termin, 2001) averaging 160 miles
further from their families than inmate fathers (Hagan & Dinovitzer, 1999). For women on
2

probation and parole, recidivism has been found to be harmful for offenders and their children
due to exposure to illegal activity and increased risk for violations which may result in loss of
child custody and imprisonment (Hagan & Dinovitzer, 1999; La Vigne et al., 2005).

Challenges for Mothers on Probation and Parole
Although female parolees have been found to experience additional barriers in
maintaining family contact due to incarceration, for female offenders on probation as well as
those on parole, many of their over-arching issues and stipulations of supervision create
problems in forming and maintaining ties to children. For example, drug offenses that result in
community supervision can bring about child abuse investigations that lead to more intensive
supervision or imprisonment for an offender originally diverted to probation (Fagan, 1994). This
issue has been especially prevalent for women involved in the criminal justice system, as a
majority of them have drug charges, and more often than male offenders have histories of
substance abuse (Boyd, 1999; Morash, 2010; Salisbury & Van Voorhis, 2009). Compared to
men, women are more often the primary caregivers of their children, and this increased
involvement may contribute to their greater risk of losing custody and facing charges concerning
child abuse and endangerment (Glaze & Maruschack, 2008; Mumola, 2000). Disproportionate
criminalization for mothers compared to fathers has also been associated with punitive legal
policies for addressing drug use during pregnancy, which has been found to lead to child abuse
cases, court ordered abortions, incarceration, and loss of custody (Paltrow, 1990). In addition,
women more often than men report feelings of guilt and humiliation regarding their children’s
exposure to their criminal behavior; female offenders may experience more parental stress
(Giordano, 2010), which has been associated with increased substance abuse, depression and
recidivism (Arditti & Few, 2006; Arditti & Few, 2008; Salisbury & Van Voorhis, 2009).
3

The unique legal challenges that female offenders who are mothers face are often inflated
by social sanctions that further scrutinize them for violating gender roles stipulating that females
be lawful and nurturing to children (Clear, Waring & Scully, 2005; Heimer & DeCoster, 1999).
Assumptions about women bearing the brunt of care giving responsibilities for children have set
female offenders up for social ostracism by the community and the justice system, as these
women have been generally viewed as out of control, inept parents that threaten their children’s
well-being (Boyd, 1999). Some scholars have mentioned this gendered stressor as protective in
that it may deter maternal offending or motivate women offenders to seek substance abuse
treatment (Riehman, Hser, & Zeller, 2000). On the other hand, researchers have also argued that
in combination with legal sanctions, social sanctions may dissuade maternal or pregnant
offenders from seeking treatment for substance abuse, in fear that they may lose their parental
rights (Fagan, 1994).

The Importance of Identifying Influences on Women’s Desistance
abuse, and also that they In addition to the potential influence of bonds to children and
parenting stress on desistance, existing literature has identified several other possible influences
for women (Giordano, 2010; Giordano et al., 2002; Salisbury & Van Voorhis, 2009). Numerous
studies have found that female offenders more often struggle with substance abuse, mental
illness, sexual and physical are more often economically marginalized than male offenders, and
these factors influence desistance (Messina et al., 2006; Richie, 2001; Salisbury & Van Voorhis,
2009). Scholars have also found that programming varies in effectiveness across sexes and other
traits of offender groups depending on whether they address relevant risks and needs (Moloney
& Moller, 2009). Program effectiveness must be considered given the enormous cost of
corrections and the expenses of increased foster care caseloads (Mumola, 2000; White, 2012), in
4

addition to the future expenses related to children of offenders’ future criminal justice
involvement (Murray & Farrington, 2008; Giordano, 2010). Research on women offenders must
be increased due to the over-reliance of research on male offenders’ risks, needs and pathways in
and out of crime (Giordano, 2010; Salisbury & Van Voorhis, 2009).
In order to fill gaps in theory development, contemporary research has integrated gender
relevant factors with concepts from traditional criminology to better explain recidivism
(Salisbury & Van Voorhis, 2009; Giordano, 2010). Research on female offenders has found that
mothers’ commonly mentioned their concern and guilt regarding their offending and the
subsequent circumstances for their children (Giordano, 2010; Arditti & Few 2008). For example,
incarcerated women have expressed concern about their children’s well-being while apart and
their regret over the exposure their criminal actions may have had on their children (Arditti &
Few, 2008). Maternal offenders in the community have also expressed concerns and distress for
their children’s well-being, especially if they were no longer their children’s primary caregiver
and were suspicious of their child’s well-being and living circumstances (Giordano, 2010). To
examine maternal offenders, several researchers have used life-course perspectives, especially
the version of life course scholarship inspired by Sampson and Laub’s (1993, 2003) theory of
age-graded informal social control (Edin & Kafalas, 2005; Giordano, 2010; Giordano et al.,
2002; Kreager, Matuesda & Erosheva, 2010;Michalsen, 2011; Thompson & Petrovic,
2009).These studies used this theory to examine hypothesized gender specific variables related to
stability and change in criminal behaviors of mothers.
Age-graded informal social control theory is supported by research that has
methodological strength gained through using both quantitative data and the qualitative, in-depth
narratives that are arguably required to reveal the intricate nature of behavioral shifts throughout
5

the life course (Giordano, 2010; Laub & Sampson, 2003; Michalsen, 2010). Like many
longitudinal research endeavors, the framework suggests the use of a research design involving
the collection of quantitative measures of demographics, contextual factors and criminal
behaviors though self-report surveys and official records at different time periods (Laub &
Sampson, 2003). Especially distinctive to this approach is the supplement of quantitative data
with life narratives based on interviews with offenders. Narratives help reveal the subjective
perceptions of the offender regarding their experiences leading to desistance and persistence
(Giordano, 2010; Laub & Sampson, 2003). The current study contributed to this line of research
as the subject of motherhood and desistance for females on community supervision has never
been approached longitudinally using quantitative and qualitative measures on a large sample.

6

Chapter 2: Review of Existing Literature
In order to contribute to knowledge on desistance for maternal offenders under
community supervision, the current study examined key variables supported by existing life
course studies and feminist research. This chapter begins with an overview of Sampson and
Laub’s age-graded informal social control theory (1993), as the current study examined maternal
motivations and stressors using a gendered adaptation of this model. After the model is
described, the theories and politics that influenced the evolution of age-graded informal social
control theory are explored. This section is followed with a description of prominent studies and
general findings across existing literature regarding age-graded informal social control theory
applied to examining women and mothers. Overall, studies that considered heterogeneous
populations, situational contexts and historical conditions found differences in desistance
between individuals who varied in the quality of social bonds to their children. The majority of
studies provide support for the significance of motherhood and increased attachment to children
as turning points for female offenders (Edin & Kafalas, 2005; Kreager, Matuesda & Erosheva,
2010; Michalsen, 2011; Thompson & Petrovic, 2009).
After the review of literature connecting maternal offenders’ bonds to children with
desistance, a brief review of alternative factors commonly found to influence criminal
involvement of women is presented. Several variables that are alternative explanations were
incorporated in this analysis. Variables that were not in the available data set are identified, and
this limitation is noted in the Discussion section. Relatedly, those factors found to clarify the
nature of female offending and social bonds to children as a turning point but that this analysis is
unable to address were also mentioned as limitations. Overall, this review justifies the design of
the study, which the next chapter on methodology describes.

7

Literature on Life Course Theories and Age-Graded Social Control
For more than twenty years, criminal behavior over the life course has been examined by
studying the continuation of criminality for individuals and subgroups of individuals (Laub &
Sampson, 2003). Sampson and Laub (1993) made a significant shift in criminological focus to
include the desistance processes for individuals over their life course. They discovered it was
necessary to investigate desistance of formerly persistent offenders when they re-examined and
extended Glueck and Glueck’s data collected between 1939-1963 for 500 delinquent and 500
non-delinquent boys recruited for study between the ages of 10 and 17. Equipped with extensive
data on this large sample of delinquents, Laub and Sampson (2003) were compelled to explore a
prevailing controversy in the field of criminology regarding career criminals.
On one side of this controversy, findings from a study of a Philadelphia birth cohort
(1945-1963) inspired researchers’ to fixate on chronic offenders. The specific finding was that a
small percentage of their sample (6%) was responsible for the majority of crimes committed by
the entire cohort (52%) (Wolfgang, Figilio, & Sellin, 1972). In the 1980’s, developmental
criminologists also took interest in career criminals’ trajectories, and created typologies that
identified the etiological characteristics of chronic offenders and other groups. These studies
popularized the notion that a small percent of the population would become chronic offenders,
and they could be recognized at a young age based on risk factors such as low IQ, low SES, and
early and serious delinquency (Blumstein et al., 1986; Moffitt, 1993; Wolfgang, Figilio, &
Sellin, 1972).
Although Laub and Sampson (2003) mentioned that typologies may provide benefits in
terms of categorizing risk factors and suggesting treatment methods for groups of offenders, they
warned of several policy implications that could arise if assumptions were made regarding career
criminals. They mentioned potential policy issues such as the labeling of children as future
8

chronic offenders and discriminatory incarceration of those with childhood risk factors
(Greenwood, 1982). They also raised questions about the operationalization of variables
qualifying individuals as chronic offenders in research. In order to address these concerns, Laub
and Sampson (2003) carried out empirical study by examining the criminal careers of the
Gluecks’ sample over the participants’ life course.
To address criticisms of the career criminal approach, Laub and Sampson (2003) tried to
identify the chronic offenders in the Gluecks’ sample. They attempted to categorize chronic
offenders based on childhood risk factors supported by Mofitt (1993), one of the most influential
developmental criminologists of the time. After tracing the participants’ criminal records up to
age 70, they found that offending behaviors of the persistent group varied substantially by age
and crime type in adulthood (Sampson & Laub, 1993). Moreover, Laub and Sampson (2003)
found that childhood risk factors could not be relied on to prospectively identify persistent
offenders. This finding motivated them to explain the variations in desistance for formerly
persistent offenders in their study. In order to determine the causes of this variability, they set out
to interview at least five participants from each of the different criminal career trajectories they
found based on official records of the subjects’ criminal histories and supplemented by the
Gluecks’ original data.
The narratives collected through interviews with the men in the different trajectories
revealed common trends in life events of those who desisted earlier, especially in comparison to
those that persisted in crime into late adulthood. Sampson and Laub (1993) found that despite
cumulative risk factors indicative of persistence and the occurrence of antisocial behaviors in
early life and through early adulthood, those who desisted more frequently mentioned the
formation of quality pro-social bonds. The researchers concluded that variations in informal
9

social control were the primary causes for crime and desistance. They described bonds as
creating inter-reliant systems of responsibility and restraint which carried related costs and
rewards for the participants bringing them closer to or away from criminal behavior (Sampson &
Laub, 1993). More specifically, they found life events during adulthood – such as employment,
marriage, military commitment, and neighborhood changes – increased quality attachments and
reduced the criminal behaviors of previously persistent offenders in the sample (Sampson &
Laub, 1993; Laub & Sampson, 2003).
Noting the narrow scope of informal social control theory, they revised their theoretical
framework to include additional concepts that they argued were of equal significance to other
parts of the theory. They incorporated personal agency, situated choice, routine activities, age,
the macro level historical events of the time, local culture, and community context (Laub &
Sampson, 2003). They found these components helped explain the variation in bond formation
and desistance across their groupings of persistent offenders (Laub & Sampson, 2003). With the
expanded theory, the authors still concluded that the most influential types of informal social
control for the men were employment, marriage, military commitment, and neighborhood
changes (Laub & Sampson, 2003). Laub and Sampson’s (2003) development and revision of
age-graded informal social control theory provided the field of criminology with a
methodological approach that incorporated subjective narratives of offenders with quantitative
evaluations and official criminal histories.
Despite the in-depth nature of the theory, many feminist scholars have mentioned that
their findings neglected the potential variability of desistance processes for females (Edin &
Kafalas, 2005; Giordano et al., 2002; Kreager, Matuesda & Erosheva, 2010; Michalsen, 2011;
Thompson & Petrovic, 2009). Laub and Sampson (2003) warned of this, writing that their
10

findings lacked generalizability across gender groups and historical context. They suggested that
examinations should be conducted on females to identify the applicability of the framework and
the significance of the turning points they found for the Gluecks’ sample of men (Laub &
Sampson, 2003).
Prior research that provided insight into female desistance has also been criticized
(Giordano et al., 2002) for having deficient numbers of delinquent females in their samples
(Elliott, Huizinga, & Menard, 1989; Osgood et al., 1988). As women are more often processed
through the criminal justice system, gendered examinations grounded in research on large
samples of women offenders have more recently expanded. This more recent research on women
has found that their desistance was not explained by events associated with men’s desistance
(Edin & Kafalas, 2005; Giordano et al., 2002; Kreager, Matuesda & Erosheva, 2010; Michalsen,
2011; Thompson & Petrovic, 2009). For example, in Giordano and colleagues (2002) study,
education and employment were not found to be predictive life events in explaining desistance
for the women, as those that desisted from crime were unemployed or had low paying jobs,
which prompted the authors to question the influences on these women’s desistance (Giordano et
al., 2002).
Additionally, as women have been increasingly examined using life course propositions
and age-graded informal social control theory. These studies have consistently reported that life
events found to influence women’s transitions in and out of crime are significantly different from
influences on men’s transitions found in Laub and Sampson’s (2003) work, and transitions of
men from more current research (Thompson & Pertovic, 2009). Most notable is the influence of
parenting on women’s desistance from criminal behavior (Edin & Kafalas, 2005; Kreager,
Matuesda & Erosheva, 2010; Michalsen, 2011; Thompson & Petrovic, 2009); on the other hand,
11

parenting has not been found to have a significant or as strong of a role for male desistance
(Sampson & Laub, 1993; Giordano et al., 2002; Giordano, 2010; Thompson & Petrovic, 2009).

Life Course Examinations of Female Offenders
Qualitative Research Using Age-Graded Informal Social Control Theory
Life course scholars applying age-graded informal social control theory have regularly
made use of qualitative narratives to provide insight into the thought-provoking perceptions of
offenders regarding their personal struggles and turning points (Edin & Kafalas, 2005; Kreager,
Matuesda & Erosheva; Giordano, 2010; Michalsen, 2011). Although the present research did not
use data from complete life narratives, it used women’s qualitative accounts of their thoughts
about motherhood and their interactions with children. Such qualitative accounts are useful for
studying underrepresented topics and populations in existing research (Arditti & Few, 2008). In
addition, the qualitative data made it possible to identify the subjects’ perceptions of their own
parental motivations as triggers for cognitive and behavioral shifts away from crime (Giordano et
al., 2002).
Giordano and colleagues (2002) expanded on age-graded social control to include more
deliberate processes of change that they found in analyzing their interview data on a follow up
sample of serious male and female delinquents. They outlined four transformative stages
required for cognitive transformations, which are essential in the facilitation of change beyond
the source of social control. First of all, the person needs to be open to change, and then they
need to be exposed to a hook for change. Hooks for change vary for individuals, but usually an
opportunity is presented for the person to create ties, or strengthen ties with another person,
activity, or cause. This is followed by an ability to imagine their replacement self, which may be
displayed by their planning process to change. Finally, their views on deviant behavior, or the
lifestyle that accompanies deviant behavior, are converted and replaced with a new identity
12

(Giordano et al., 2002). These components were theorized to form mental filters based on
cognitive reinforcements which are held stable by the new identity. Meaning, the new identity
would be a stronger internal filter, providing more permanent change that would last upon
separation from their control source (Giordano et al., 2002).
In a follow up of the same cohorts of 127 girls and 127 boys, Giordano (2010) found that
female narratives were similar but more optimistic than those of males, especially concerning
children (Giordano, 2010). The prevalence of mothers that mentioned children in Giordano’s
study is consistent with other qualitative assessments (Arditti & Few, 2008; Edin & Kefelas,
2005). Qualitative assessments have found that female offenders maintain prominent expressions
of guilt and concern regarding their behavior’s impact on their children (Arditti & Few, 2008;
Giordano, 2010). Even female offenders that have lost custody of their children were found to
mention them as motivational in their self-reported desistance processes (Michalsen, 2011).
Relatedly, women have mentioned their love for their children as the reason they did not seek
custody. These women talked about their fears of removing their children from stable living
situations only to be unable to properly care for them. Mothers also spoke of these deficiencies
as incentives for change, as they saw their current abilities to support children as inadequate and
mentioned changes they needed to make to be a better role model and to provide more suitable
care for their children (Arditti & Few, 2008; Michalsen, 2011).
Examples of content mentioned by mother offenders during qualitative interviews have
been described as mending relationships, making up for lost time, and providing guardianship
for their existing or newborn children (Edin & Kafalas, 2005; Michalsen, 2011). Qualitative
accounts also reveal the daily routines that insulated women through new activity, pro-social

13

bonds and priorities (Edin & Kafalas, 2005; Hunt, Joe-Laider, & MacKenzie, 2005; Fleisher &
Krienert, 2004; Michalsen, 2011).
Although convincing evidence of cognitive transformations and motivations have
frequently been found in interview data of maternal offenders, in some cases, they are not
associated with desistance (Giordano, 2010). In compiling alternative explanations based on
survey responses and interview content, scholars described risk factors that may have impeded
these and similar motivational accounts for desistance. Several terms have been used to describe
these barriers. In research on juvenile offenders and their relations with parents, Schroeder and
colleagues (2011) mentioned cumulative risks for continuance or cumulative continuity, which
was described as events that snowball into more serious outcomes due to the impacts of prior
criminal behavior paving the way for additional disadvantages (p. 67). In addition, qualitative
accounts of mothers reentering the community have identified factors such as depression,
violence and addiction to comprise the “triple threat” which in combination, were hypothesized
to increase harmful stressors for maternal offenders with health problems, dysfunctional
relationships, trauma related to loss, and concern and guilt regarding children (Arditti & Few,
2008, p. 304). Relatedly, Giordano (2010) found qualitative evidence for a “respectable person
package” which comprised of marriage, and a job with stable income. She theorized that women,
who expressed cognitive shifts and identity transformations away from criminal behavior to
assume parenting responsibilities, but that still recidivated, lacked traits necessary in the
respectability package (Giordano, 2010). To address the lack of available longitudinal qualitative
research on mothers under community supervision, especially probation, the current study will
provide unique information on the moderating effect of variation in risk for recidivism on the
association of mothering-related variables to recidivism in a follow-up assessment (Edin &
14

Kafalas, 2005; Kreager, Matuesda & Erosheva, 2010; Michalsen, 2011; Thompson & Petrovic,
2009).
Quantitative Research Using Age-Graded Informal Social Control Theory
Conceptualizations of Recidivism for Mothers
As previously mentioned, Laub and Sampson (2003) noted several implications for the
operationalization of variables used by scholars for the identification of chronic offenders. Based
on their assessment, they promoted the use of data regarding multiple offending behaviors from
several sources such as self-report surveys, interview content with participants, information from
associates of the participant, and the use of official records (Laub & Sampson, 2003).
Researchers who have studied maternal offending also used a variety of measures of desistance.
For example, Michelson (2011) only considered the women’s yes or no responses regarding
reduction in their offending behaviors at the time of the interview compared to their behavior
prior to their last incarceration. Although the measure was simple and did not necessarily capture
the termination of offending behavior (Michelson, 2011), the study sheds light on the gradual
progression often found in quality social bond formation and the actual processes of desistance
(Elder, 1985; Giordano, 2010; Laub & Sampson, 2003).
Due to the high expense and the rigorous nature of life course longitudinal assessments,
many scholars have made use of existing data sets which limit their ability to measure variables
of interest and usually do not allow for follow-up with participants (Kreager, Matsueda, &
Erosheva, 2010; Laub & Sampson, 2003; Warr, 1998). For example, Thompson and Petrovic
(2009) measured the dichotomous response to subjects’ self-reported illicit substance abuse and
their responses for recurrent or single use over several interviews. The authors explained that
their use of an existing data set inhibited their ability to triangulate their measure (Thompson &
Petrovic, 2009). On the other hand, they argued that the existing dataset provided a timely
15

dependent variable through examining illicit use of drugs other than marijuana; this design
feature is important in light of recent policies around the country that have decriminalized
marijuana (Thompson & Petrovic, 2009). This example reveals some negative and positive
aspects of utilizing existing data on the operationalization of dependent variables.
Another issue is the term desistance broadly encompasses the termination of all criminal
behavior, which is problematic when broad and specific measures are compared without
consideration of their vast differences (Kreager, Matsueda, & Erosheva, 2010). As the scope of
criminal behavior has evolved, scholars also began to examine more specific behaviors for
particular groups of maternal offenders. For example, behavioral shifts such as self-reported
disaffiliations with gangs (Hunt, Joe-Laider, & MacKenzie, 2005) and reductions in self-reported
underage alcohol consumption (see example with small sample, Fleisher & Krienert, 2004) have
been examined and found to be common form of desistance for female gang members during
their transition to motherhood. Although these group specific measures, and other single
measures reviewed in this section provide specific insight into particular types of desistance, the
findings must be taken with skepticism, as they may not be as telling for other groups of people,
or for alternative behavioral changes. Studies that used age-graded social control theory and
provided several measures provide insight into the nature of cessation for maternal offenders
regarding multiple behaviors reported from varying agencies (Edin & Kafalas, 2005; Kreager,
Matuesda & Erosheva; Giordano, 2010; Warr, 1998). The dependent variables used to measure
desistance in these studies along with suggestions from Laub and Sampson (2003) were used to
formulate the current study’s dependent variables which will be elaborated in the next chapter.
Conceptualizations of Informal Social Control for Mothers
This section will review literature regarding social bonds to children and stress from
parenting. These factors have been found as primary independent variables from prior research
16

on maternal desistance. Many of the recent studies on women using a life course approach
obtained sample sizes that permit quantitative assessments to be generalizable across the
populations sampled (Edin & Kafalas, 2005; Kreager, Matuesda & Erosheva, 2010; Michalsen,
2011; Thompson & Petrovic, 2009). The large numbers of participants allowed researchers to
assess quantitative survey measures and to quantify qualitative data for empirical analysis.
Surveys mainly have been used to provide a sense of maternal circumstances thought to serve as
supports or barriers for desistance (Giordano, 2010; Kreager, Matuesda & Erosheva, 2010;
Michalsen, 2011; Thompson & Petrovic, 2009).
Social Bonds to Children
As social bond formation is central to age-graded informal social control theory,
attachment to children has been a dominant focus for studies on turning points for maternal
offenders (Giordano, 2010; Graham & Bowling, 1996; Kreager, Matsueda & Erosheva, 2010).
These measures of attachment have often been formulated based on situational dispositions and
characteristics of the maternal populations of concern or interest to researchers and policy
makers. For example, several criminological studies surveyed maternal offenders’ on their
maintenance of relationships with children during incarceration. The types of communication
examined have included self-report measures for inmates regarding: pre-imprisonment cohabitation with children; inmate mothers’ self-reports of caregiver relations; self-reported mail
contact, phone contact, and visitation during imprisonment (Barnes & Stringer, 2014);
comprehensive prison records of visitation (Cochran, 2013); and the frequency and timing of
visitation throughout subjects’ imprisonment (Bales & Mears, 2008). One of the studies found
support for social bonds theory for inmate mothers based on correlations between increased
communication with children, expectations of future child custody, and close ties to caregivers as
they related to more salient maternal identities upon reentry (Barnes & Stringer, 2014). However,
17

studies that measured recidivism rates in relation to child visitation have provided mixed
evidence of correlations for inmate mothers upon reentry. For example, one study found more
child visitation was related to decreased recidivism for women (Cochran, 2013), while another
examination found increased frequency and the timing of child visitation nearing the end
women’s incarceration to be associated with increased recidivism (Bales & Mears, 2008).
Mothers on community supervision that have not experienced imprisonment may have
stronger relations with their children (Leverentz, 2006), and therefore attachment measures can
address more specific parent-child involvement. Three indicators of attachments have been
suggested as important in desistance processes: supervision, involvement and maternal agency
(Giordano, et al., 2002; Giordano, 2010). Studies on offenders in the community have measured
amount of mothers’ supervision of their children. For example, through follow-up surveys on a
sample of serious delinquents, Giordano (2010) asked how often women called their children per
day for updates on their whereabouts. Another measure of social bonds to children has been the
extent of involvement parents’ maintain in children’s lives, in areas such as health behaviors,
school-related activities, and their children’s social lives. For example, items that assessed
involvement have included inquiries about how often parents assisted children with school work
(Miller et al., 2011), and to what extent they had met their children’s friends and the parents of
their friends (Giordano, 2010). Finally, measures of parental preparedness have also been
examined as influencing the effects of bonds with children (Giordano et al., 2002). These
measures provide detailed portrayals of parent-child attachment in the community while
touching on other theoretical components associated with Laub and Sampson’s (2003) revised
theory, such as routine activities, situated choice, and personal agency.

18

Although several studies of female desistance provided survey measures of attachment
that assessed specific types of involvement, the majority relied on more general measures such as
pregnancy and child-birth (Edin & Kefalas, 2005; Kreager, Matsueda, & Erosheva, 2010
Thompson & Petrovic, 2011; Wakefield & Uggen, 2004). It is possible that this is attributable to
the salience of the transition into motherhood in comparison to specific measures of attachment
such as those noted in the previous paragraph. However, broader factors may not provide insight
into definitive transitions like measures of contact with children, custody of children, and
residency with children. Next, the mixed findings regarding parent-related motivations,
attachment to children, and desistance will be discussed, as the variations in populations
examined and measurements of parent-child relationships certainly contribute to the inconsistent
findings (Giordano et al., 2002; Kreager, Matsueda, & Erosheva, 2010; Thompson & Petrovic,
2011).
In all, theoretical assumptions from age-graded informal social control theory suggest
that increased attachment to children results in decreased criminal activity (Giordano, 2010;
Thompson & Petrovic, 2011). However, prior research has provided mixed findings regarding
the relation of this theoretic concept with desistance; some scholars found attachment to children
to be protective against criminality (Edin & Kefalas, 2005; Kreager, Matsueda, & Erosheva,
2010; Thompson & Petrovic, 2011) while others have not (Blokland & Nieuwbeerta, 2005;
Giordano et al., 2002; Varriale, 2008). In order to determine the nature of this disjunction,
scholars have continued to re-examine more general measures of attachment for heterogeneous
populations (Kreager, Matsueda, & Erosheva, 2010; Thompson & Petrovic, 2011), and have also
made use of unique measures, like the “wantedness” of children (Giordano et al., 2011) to better
explain the variations in results. Although knowledge about attachment has been extended in
19

studies of diverse samples, no study has focused on parent-child attachments and desistance
processes for a general population of women offenders in the community, with varying parental
circumstances. By using secondary data for a sample of drug-involved women on community
supervision, it was possible to examine mothers with varying types of relations with children.
Stress from Parenting
Parenting has often been mentioned as a stressor that could lead to criminal involvement,
especially for at-risk populations (Arditti & Few, 2008). Stressors from parenting have been
linked to depressive symptoms, mental illness, substance abuse, and criminal behavior (Arditti &
Few, 2008). Some have found criminal behavior in disadvantaged populations, due to economic
stressors experienced while trying to provide for children (Giordano, 2010; Thompson &
Petrovic, 2009). In addition, parental stress has been associated with risks for those with present
and past criminal justice involvement. Stress from parenting may exasperate or be exasperated
by negative feelings surrounding formal criminalization and the challenges of meeting
requirements of probation and parole (Arditti & Few, 2008; Giordano, 2010). Giordano (2010)
used quantitative instruments to compare responses on the parental disposition of subjects that
were institutionalized as juveniles compared to a control sample non-institutionalized
delinquents. The survey measures she used to assess parental stress included questions like
“Raising my child can be a nerve wracking job” or “I’d like to be able to do a better job at
communicating with my child.” She found subjects with prior juvenile institutionalization
reported higher levels of depression and stress from parenting (Giordano, 2010).
Parental stress has been measured with many instruments expected to be related to mental
health issues and recidivism (Giordano, 2010; Thompson & Petrovic, 2009). For example, prior
research has measured stress experienced due to varying influences such as: parenting efficacy;
family, partner, or parental support; financial standing; perceptions of difficulty of parenting;
20

community conditions for raising children; and behavioral variations in reactivity of children
(Arditti & Few, 2008; Giordano, 2010). Guilt regarding criminal activity and time spent
incarcerated or away from children has been quantified for male and female offenders, and was
found to be more prevalent for females (Giordano, 2010). These stressors have been correlated
with increased depression, substance abuse, and even recidivism (Salisbury & Van Voorhis,
2009) Transitions into motherhood have also been associated with vulnerabilities for harmful
stressors that prompt negative coping and criminal activity. New mothers may experience more
stress causing them to turn to criminal outlets or negative coping more often than experienced
mothers (Thompson & Petrovic, 2009).
Although stress from parenting and transition into motherhood has often been associated
with risk, prior research has also found these negative or overwhelming emotions to be
associated with pro-active behavioral shifts (Edin & Kefelas, 2005; Thompson & Petrovic,
2009). For example, the majority of new mothers in a pool of gang affiliated young women with
prior heavy-alcohol use were found to abstain from alcohol consumption during pregnancy. Even
though the majority reported drinking after child-birth, the amounts were considerably lower,
and the subjects’ reported drinking alone and for the purpose of winding down after a busy and
stressful day rather than for social reasons (Hunt, MacKenzie, and Joe-Laidler, 2005). Other
research has found young women’s’ transition into single motherhood in disadvantaged
neighborhoods to be more strongly correlated with reductions in offending behaviors than for
their counterparts transitioning into motherhood with a marital partner or significant other
(Kreager, Matuesda & Erosheva, 2010; Thompson & Petrovic, 2009). Though this could be
explained by criminally-involved partners’ influence on the continuation of criminality
(Salisbury & Van Voorhis, 2009), the absence of partner support has been consistently
21

recognized as a stressor in research, yet single parenthood has been described as protective based
in some research (Edin & Kafalas, 2005; Kreager, Matuesda & Erosheva, 2010).
In the face of stress due to limited assistance, mothers may experience more pressure to
assume sole responsibility over children, which in turn creates incentives for identity
transformation and behavioral changes (Kreager, Matuesda & Erosheva, 2010). Stressors
experienced by these women may also influence their outreach to pro-social networks in the
community, such as other single mothers (Anderson, 1990; Edin & Kafalas, 2005; Kreager,
Matuesda & Erosheva, 2010).
Relevant Variables in Research on Women’s Criminality and Recidivism
Quantitative surveys have also identified variables that promote or inhibit the effects of
traditional pro-social events like child-bearing or assuming responsibilities for children
(Sampson & Laub, 1993; Giordano, 2010). Even in describing the operationalization of social
bonds to children and stress from parenting, effects of alternative variables were mentioned due
to the mixed findings across samples and the individuals in the samples. It is therefore important
to identify other major predictors of recidivism that could account for relationships of
independent and dependent variables. This section will detail the use of control variables from
prior research to help determine the directional effects these measures will posed on the current
sample and to justify their incorporation into the current study’s methodology.
Risks for Recidivism
Age
One of the characteristic findings from Sampson & Laub (1993) research is that with age
most people eventually desists from crime, whether desistance was brought on by cognitive
shifts, routine activity changes or barriers to criminal activity. They disentangled the age-crime
curve based on crime types and found that lesser crimes, such as property offenses, better fit with
22

Gottfredson and Hirschi’s (1990) model which identified property and violent crime peaks
during early teenage years between 16 and 18. Although Sampson and Laub (1993) found
support for this model while considering lesser offenses for men, they found more serious
offenses to have later peaks and more prolonged persistence through adulthood than suggested
by the age-crime curve. Scholars who included women in their samples have found mixed results
as well, with some finding that females’ criminal activity peaks in earlier years than do males
(Fergusson & Horwood 2002), and others finding that women have a late onset in offending starting after age 30 (Chung, Hill, Hawkins, Gilchrist, & Nagin, 2002), and that they commit
different types of crimes (Salisbury & Van Voorhis, 2009).
Demographics, Neighborhood Contexts, Relationships, and Peers
Prior research has indicated criminal justice involvement is positively associated with
neighborhood disadvantages such as depleted resources and service limitations (Visher & Travis,
2003). These areas have also been associated with high levels of racial minorities, mainly
African Americans, and single mother households. Despite these commonly documented
correlations, research on maternal desistance has provided mixed results of their effects
(Giordano et al., 2011). For example, several studies have indicated these traits decrease the
likelihood for mothers to form quality bonds to children, which is associated with their continued
persistence in crime (Blokland & Nieuwbeerta, 2005; Giordano et al., 2002; Varriale, 2008).
Extensions on research have found this effect to be reversed for those in disadvantaged areas for
mothers who expressed “wantedness” for their children, meaning women planned their
pregnancy, or accepted it and held positive attitudes about being pregnant (Giordano et al.,
2011).
Other life course research found the transition to motherhood increased desistance for
women in disadvantaged neighborhoods (Edin & Kafalas, 2005; Kreager, Matuesda & Erosheva,
23

2010; Thompson & Petrovic, 2009). Some of these studies cited support from Anderson’s (1999)
ethnographic study on an urban neighborhood in Philadelphia (Edin & Kafalas, 2005; Kreager,
Matuesda & Erosheva, 2010). Anderson (1999) found that women in these disadvantaged
neighborhoods were more likely to mature out of crime and create social bonds based on
children, through connections with other single mothers. He postulated that the effect of
motherhood was more influential on the maturation of these women due to their bleak likelihood
of forming traditional relationships or marriages with men in their communities (Anderson,
1999). These mixed findings indicate that single parenting is associated with increased
desistance for those in disadvantaged neighborhoods (Edin & Kefalas, 2005; Kreager, Matsueda,
& Erosheva, 2010; Thompson & Petrovic, 2011), but studies on other populations have found no
effect (Blokland & Nieuwbeerta, 2005; Giordano et al., 2002; Varriale, 2008) or the opposite
effect (Bales & Mears, 2008).
Although several studies have noted reductions in time spent with criminal peers upon
transition into motherhood (Edin & Kefelas, 2005; Fleisher & Krienert, 2004), others studies
found parenting to be insignificant regarding changes in peers without marriage commitments
(Warr, 1998). In expanding on age-graded informal social control theory, Giordano (2010)
describes the need to focus more attention on peers, as they were not found to be of importance
in Sampson and Laub’s (1993) original model, which relied on a sample recruited during a time
when marriage was more customary (1940) than it has been in more recent years (see also Warr,
1998).
Educational and Occupational History
Another variable that has been found to influence parental stressors and recidivism
outcomes is socio-economic status. Despite some researchers’ findings that disadvantage
provides incentive for maternal offenders, as mentioned in the prior section (Edin & Kefalas,
24

2005; Kreager, Matsueda, & Erosheva, 2010; Thompson & Petrovic, 2011), many studies have
found debilitating stressors to be created or inflated based on level of resources (Arditti & Few,
2008; Giordano, 2010). In order to control for these economic issues, researchers have used
simple measures such as income-level (Giordano et al., 2011) or status as a single parent in the
household (Thompson & Petrovic, 2009). In contrast, some studies used more complex measures
that assessed the financial and occupational prospects of offenders (Cobbina et al., 2014).
Complex measures were created to further refine the economic risk factors of marginalized
populations. Questions about previous employment challenges, current employment and
financial situation have been included in this type of measure. Measures of educational
attainment have also been factored in employment and financial related controls to provide for
future prospects of the offender.
Criminal History, Substance Abuse, Mental Health Status, and Exposure to Abuse
As mentioned in the above discussion of risk factors identified by developmental
criminologists, criminal history has been found to have strong effects on persistence of criminal
behavior. In addition to early onset of offending, scholars have shown that the seriousness of
criminal acts contribute to persistence in crime (Sampson & Laub, 1993; Giordano, 2010;
Gottfredson & Hirschi, 1990). Additional predictors of continued illegal behavior are history of
sexual and physical abuse, current abuse, mental illness substance abuse, anger, hostility, and
anti-social attitudes (Salisbury & Van Voorhis, 2009).

Conclusion
The review of the existing literature on the life course provides justification for the
current study. Prior findings suggest that individual differences and variations in heterogeneous
populations alter the effects of life events, like parenting, on desistance (Giordano et al., 2002).
In addition, unlike findings based on male offenders, female offenders’ more often fare better
25

over their life course upon establishing quality bonds to children. Contemporary findings
surrounding this proposition have been especially supported for African American females
residing in disadvantaged neighborhoods as they transition into parenthood (Edin & Kafalas,
2005; Kreager, Matuesda & Erosheva, 2010; Thompson & Petrovic, 2009).
Although prior research has found that stress from parenting can be debilitating on those
reestablishing connections with children, those seeking custody of children and those estranged
from their children through incarceration or legal termination of rights, females continue to
mention their children as primary reasons for change (Arditti & Few, 2008; Michalsen, 2011).
An important extension this study provided to prior research was to determine if motivations
associated with motherhood were related to desistance from crime for women on community
supervision. In addition, follow up research using the current study’s findings and data, will shed
light on those mothers with motivations, yet who remained in conditions that inhibited these
motivations from having an effect on cessation. It is possible that such identified barriers could
be targeted for interventions, providing better outcomes for mothers with desires to embrace their
parenting role, but who struggle with desistance.

Hypotheses
1. Women’s mention of maternal motivations during interviews will be negatively
associated with measures of recidivism.
2. The frequency with which women mention maternal motivations during interviews will
be negatively associated with measures of recidivism.
3. Higher levels of involvement with children based on a scale administered at the first
interview will be negatively associated with measures of recidivism.

26

4. Higher levels of stress from parenting based on a scale measure administered at the first
interview will be positively associated with measures of recidivism.

27

Chapter 3: Methodology of the Current Study
The data for the current subsample was collected in a study that examined 402 druginvolved women on probation and parole. From November 2011 to November 2012, three
interviews were conducted in urban and rural cities across 17 counties in Michigan, including
major cities such as Detroit, Lansing and Grand Rapids. Extending understanding of female
offenders using the current data set based on women under supervision in Michigan is of special
importance because the state established women-specific caseloads. The Department of
Corrections provides probation and parole agents with these specialized caseloads with training
on the use of female offender risk assessment tools and also provides them with supervision
methods to address unique challenges women offenders have been found to experience
(MacKenzie, 2013). Given that information, the mothers in this study may fare better than those
under traditional supervision, because their gender-unique challenges can be more specifically
identified and addressed under women-specific supervision (MacKenzie, 2013; Salisbury & Van
Voorhis, 2009).
Recruitment of the female offenders was based on their assignment to one of the 73
probation and parole agents that agreed to participate in the study. Three methods were used to
recruit the female offenders: the agent introduced the researcher to interested participants at the
office, the agents forwarded contact information to the researcher when permitted by their
clients, or the agents’ displayed and handed out flyers allowing their clients to contact the
researcher (Cobbina et al., 2014). Once recruited, participating female offenders were informed
that their participation was voluntary and that the information they provided would be
confidential. The female offenders were compensated in gift cards to their choice of stores in the
amount of $30 for the first, $50 for the second, and $75 for the third interview. The one-on-one
28

structured interviews were conducted by trained interviewers at various locations of convenience
for the participants such as supervision offices, the participants’ homes, or in public places.
Responses to the quantitative measures were entered into a computerized database, while openended responses to questions were audio-recorded and transcribed (Cobbina et al., 2014).

Sample
The selection of the subset of the sample for the current analysis was based on responses
to the question “Do you have kids?” First, those that reported, “I have children who I have
contact with” were included. Then those that responded, “I do not have children under 18” were
omitted because the current study seeks to uncover maternal challenges regarding
responsibilities, custody and contact with minor children. Lastly, respondents that responded
“children adopted out and thus no contact” were also omitted. Four mothers were excluded
because they permanently lost parental rights, which may alter the presence of maternal
motivations and its association with desistance if included in the current analysis. Based on this
selection process, 194 mothers that indicated they had contact with their minor children that they
did not lose parental rights for were included in the initial study sample. During the coding of the
interviews, the researchers found one of the women did not have minor children, or custody of
any minor children, reducing the sample to 193 women. During analysis, several women were
also excluded because they did not respond to at least one of the questions used to form the
variables, reducing the sample to 162 for the quantitative analysis.
In addition to selection variables, the women in this subset can be described using an
array of quantitative survey measures. This sample consists predominantly of white (44.1%),
black (36.6%), and multi-racial women (17%) in their 30’s ( x = 33, SD = 8.65), on probation
(82.5%), with no prior prison sentences (76.3%), that reported being unmarried (89.2%), single
mothers (69.6%) that had or sought custody of their minor children (91.8%). In addition, the
29

majority of the subsample reported the offense that led to their existing supervision term as druginvolved (61.9%), and that their income was below$10,000 per year (77.8%). These variable
percentages provide insight into the risks and protective factors the majority of the women in the
study experienced.

Quantitative Data
Dependent Variables
Six measures of recidivism were used as dependent variables in the current study. The
measures were from the women’s self-reported substance abuse at first interview, the supervising
agents’ case notes on the women’s probation and parole violations, the agents’ case notes on the
women’s substance abuse, the agents’ case notes on the women’s substance abuse, the agents’
case notes on the women’s substance abuse violations, official police reports on the women’s
arrests, and official court reports on the women’s convictions.
Self-reported substance abuse was examined during the first interview. The 7-item scale
combined the yes=1 or no=0 responses. The scale’s questions include: “Have you had any recent
(past 6 months) conduct violations, law violations, or technical violations related to drugs or
alcohol use?” “During the last 6 months have you received a drug screen that was rated positive
or diluted?” “Do you associate with individuals who drink heavily or use?” “In the past six
months, have you missed treatment appointments or stopped participating in support groups?”
“Does anyone in your home use drugs or alcohol?” “Are you currently using?” “Do you
currently have any feelings that you need to use drugs first thing in the morning?”
The total number of supervision officer’s reports of the women’s violations, alcohol use,
and drug use in 18 months was based on three follow-ups that were adjusted for the number of
months the women were under supervision. This was done to adjust for periods when detection
30

was not possible because supervision was revoked or ended. In addition, women’s recidivism
was measured by the number of arrests and convictions they had, which were taken from official
police records, beginning from the time women began supervision.
Independent Variables
Maternal Motivations
The primary independent variables reflect themes that captured women’s different types
of maternal motivations. These themes were coded from the qualitative data for the 181 women
who took part in all three interviews. The categories agreed upon and the coding process for the
transcripts will be explained later, in the Coding the Qualitative Data section.
Social Bonds to Children
Attachment to children was measured by a scale from the original data set that comprised
of the sum of yes=1 or no=0 responses to three questions. The questions included “Do any
children 18 or younger live with you? Or do you have at least monthly contact with a child 18 or
younger? If no, skip to next section.” “Are you involved in important decisions regarding your
children (e.g., school-related, health, outside activities)?” “Do you feel prepared to be a good
parent?” The higher numbers from this scale were created to reflect greater levels of parental
involvement.
Stress from Parenting
Stress from parenting was measured through an available scale that combined four
reversed responses for the following questions; “Raising children is a nerve-wracking job.”
“Raising children is harder than I expected.” “I have trouble keeping my kids from
misbehaving.” “My children are difficult to control.” The possible responses for these questions
included: 0=strongly agree, 1=agree, 2=disagree, and 3=strongly agree. In this existing scale, the
numerical values of the responses were reversed so that the larger values represented respondents
31

stress from parenting. The responses culminate in the following groupings zero representing the
lowest level of stress (0 thru 14=0), 1 representing mid-range parental stress (15 thru 24=1), and
2 representing the highest level of parental stress (25 thru Highest=2).
Control Variables
There are several factors that prior research suggested alternative explanations of
recidivism to the independent variables of interest. Those variables available in the existing data
set include: age, race, supervision status (parole or probation), relationship status, anti-social
peers, neighborhood context, educational needs, employment and financial needs, criminal
history, substance abuse history, mental health issues, exposure to abuse as a child, exposure to
abuse as an adult, and family support. Many of these variables are from an assessment measure
for women’s risk for recidivism that provides a composite score of items known to predict
women’s recidivism (Van Voorhis, Bauhman & Brushett, 2013). The composite index was
created by adding together risk scores and subtracting the women’s strengths scores.
The description that follows includes the title of the subscale and one or two examples of the
questions. In addition, some of the variables in the available scales were reversed so the
cumulative sum of the responses would provide a valid and consistent measure of the control
variable, similar to those mentioned in the Independent Variable section. The control variables
that include reversed responses include anti-social peers, neighborhood context, educational
needs, employment and financial needs, and family support.
Women’s Total Needs
Predictive Risks
Employment and Financial Needs
To control for employment and financial issues, the current analysis made use of the
available 10-item scale that comprises of questions regarding employment and financial needs.
32

The first question, “Are you currently employed?” provides three possible responses, fulltime=0,
part-time or unable to work because of child/family care, poor health, student, etc. =1, and
unemployed but able to work=2. An example of a variable that’s numerical response was
reversed in this scale is question 10, “Do your children have medical insurance? If not
responsible for children, score N/A” The applicable responses include yes=1 or no=0 and these
values were swapped so the higher value indicates a greater disadvantage when compiled with
the other responses.
Anti-Social Peers
Association with anti-social peers was measured with an 8-item scale comprising of
questions questions regarding their friends’ criminality. For example, the first two questions
asked “Have any of your close friends been in trouble with the law?” “Have any of you close
friends done prison time?” The composite variable responses for the questions in this scale were
operationalized as 1=yes and 0=no.
Criminal History
In order to measure the participants’ criminal history, a 6-item scale was used from the
available data set. The first three questions were related to arrests. For example, the second
question asked “What is the number of prior felonies you were convicted of? Was this the only
one? Were there one or two others? Were there three or more?” The responses included only
one=0, one or two others=1, three of more=2, don’t know=7, or refused=9. The next three
questions asked about the participant’s criminal history related to supervision. For example,
question 7 asked “Were you ever revoked on a current or prior term of probation or parole?” The
possible answers include yes=1 and no=0.

33

Substance Abuse History
The substance abuse history of the subjects was measured using an 8-item scale from the
available data set using yes=1 or no=0 responses. Two examples of these questions are, “Have
drugs or alcohol ever made it difficult for you to perform at work or in school?” “Have you ever
experiences health or emotional problems resulting from alcohol or drug use?”
Mental Health Disorders
The existing data set provided several scales to measure participants’ mental health issues
including Mental Illness, History of Depression and Anxiety, Current Depression and Anxiety,
History of Psychosis, and Current Psychosis.
1. Mental Illness
A 5-item scale was made available through the existing data set to measure mental
illness. The scale included the five questions with yes=1 or no=0 responses. For example one of
the questions asked, “Have you ever taken any prescribed medication to help you feel better
emotionally?”
2. Current Depression and Anxiety
The participants’ history of depression and anxiety were measured based on yes=1 or
no=0 responses to 6 questions from the available dataset. For example one the questions asked,
“Worrying so much about things that you have trouble getting going and getting things done?”
Current depression and anxiety was measured using the same question, but also included a
supplemental yes=1 or no=0 question for each which asks, “If yes, has this happened in the last
several days?”
3. Current Psychosis
Two items from the available dataset were used to measure the participants’ history of
psychosis. This scale combined subjects’ yes=1 or no=0 responses to, “Have many thoughts that
34

others are out to harm you?” “Seeing things or hearing voices that are not really present?” The
current measure of psychosis comprised of the same two items but each item was supplemented
by an additional yes=1 or no=0 question that asked, “If yes, has this happened in the last several
days?”
Interviewer’s indication of Women’s Anger and Hostility
The indicator for women’s anger and hostility was measured through a scale that
combined the subjects’ yes=1 or no-0 responses to 7 items asked at the first interview. Examples
of questions include, “Would you describe yourself as having a strong temper?” and “Do you
have trouble controlling your temper when you get upset?” The responses were summed, so that
7 would be the highest score for anger and hostility, while 0 would be the lowest.
Interviewer’s Indication of Women’s Anti-Social Attitude
Following the first interview, the interview completed a 6 item scale of yes=1 and no=0
questions regarding the offender’s attitude about her crime during the interview. Two questions
asked whether the offender attributed the offense to others, one focused on the women’s mention
of co-defendants, and the other question focused on whether the offender blamed victims or
others. The final question appraised whether the woman displayed feelings of remorse for the
offense, not including remorse for apprehension. The women’s antisocial attitude was measured
by the sum of these responses, with the 6 being the highest and most indicative of women’s antisocial attitude during the first interview.
Women Residing in Un-Safe Homes
A four item scale was administered to determine the women’s thoughts about the safety
of her residence, the certainty of her residence, and if there was violence and substance abuse at
the residence. The score was summed and served to measure whether the home was un-safe.

35

Predictive Strengths
Educational Strengths
An available scale from the existing dataset was used to measure women’s educational
attainment. The items in this scale capture the yes=1 or no =0 responses regarding education. For
example, the one of the questions asked, “Have you graduated from high school or received a
G.E.D.?”
Self-Efficacy
A 17 item scale was used to measure the women’s general self-efficacy based. The
responses to these items included often=2, sometimes=1, and seldom/never=0. Some examples
of questions asked include, “When you make plans, are you fairly certain that you can make
them work?” and “Does failure just make you try harder?” When the items scale items were
summed, larger numbers indicated higher levels of general self-efficacy.
Family Support
The measure for family support comprised of 3 items that measured women’s responses
regarding general family life. In addition, yes=1 or no=0 responses to 3 items that measured
family support, and 2 items that measured family conflict. The questions regarding conflict were
reverse coded when combined with family support measures, so that a higher number would
reflect increased family support. The first question incorporated into this scale asked, “How is
your relationship with your parents (parent figures) and/or siblings (check the option that best
applies)?” and provided the responses: Good, just minor conflicts=3, Conflictual some of the
time(mixed)=2, Conflictual most of the time=1, or Family, no contact=0. An example of one of
the items that were reverse coded to measure family support asked, “Do your parents or any
siblings tend to be critical of you when they communicate with you?”

36

Age and Race
Women in the study were asked to provide their age at the first interview, and the average
the women was 33 years old. Although the current study was unable to rigorously examine
trajectory changes of participants’ during childhood and their recidivism beyond 18 months after
the first interview, the wide age range of the subset (19-59) provides insight into age-crime
relationships for mothers of varying ages, risk factors, and recidivism scores. In addition, Race
was operationalized based on a scale that compiled responses as “White only” “Black only”
“Multiracial woman of color include Hispanic” “Asian only” “Unclear” “Native American.”
Supervision Status
To determine supervision status, the participants were asked, “Are you on probation or on
parole?” The responses included “probation” or “parole”. The measure of supervision status for
the women found that 82.5% of the mothers were sanctioned to probation, 16.5% were
sanctioned to parole, and 1% selected both responses, indicating they were under both forms of
supervision. The current study considered 17.5% as parolees in consideration of the unique
disadvantages associated with imprisonment and reentry that may provide different outcomes for
them compared females’ on probation alone (Leverentz, 2006).
Neighborhood Context
The available data set included an 8-item scale to measure issues of neighborhood safety.
A few examples of questions from this scale include “Are there drugs in your neighborhood?”
and “Do the police come into your neighborhood a lot?”

Coding the Qualitative Data
Audio recordings were transcribed, formatted into Microsoft Word documents and then
read into NVivo software for qualitative data analysis. After being placed into NVivo, two
trained coders unitized and coded all transcript information from the interviews the women in the
37

sample participated in. The units of analysis were comments indicating involvement or
motivation related to being a mother. During this process of unitization, the coders agreed that
one of the women in the original sample did not have custody or contact with minor children
based on transcript content, and this participant was excluded from the analysis, leaving 193
women in this study’s sample. Based on 20 cases, Guetzkow’s U (Guetzkow, 1950) was
calculated to determine if agreement on the units of analysis was acceptable, and U = .056,
which met an acceptable level of agreement.
After unitizing the data, the coders agreed upon several categories based on informal
social control that reoccurred for women in the sample on topics of motherhood during
unitization. The coding scheme was created inductively based on the content of the data, but was
influenced by variables and trends found in prior research on the theory of informal social
control (Giordano, 2010; Michalsen, 2011; Sampson & Laub, 2003). The categories included
attachment to children, being a good parent, being a better parent than before, proving to be a
good parent, parental role assertion, custody motivation, and routine activities with children.
Once the categories were determined, the two coders returned to the data and coded 20
transcripts in order to establish initial coding reliability as well as to assess discrepancies that
might exist between coders. Once adequate reliability was established and all discrepancies were
resolved, the remaining transcripts were coded independently by the first coder. Acceptable
levels of reliability (Cohen, 1960) were obtained for the themes relevant to motherhood (k =
.913).

Analysis of the Quantitative Data
After all of the interviews were coded for the 193 women in the sample, the categories
were exported to SPSS software as both dichotomous and continuous variables. Retaining those
38

two variables for each category allowed the presence (i.e., mentioned, not mentioned) and
frequency (number of times mentioned) of these maternal factors to both be compared and tested
as correlates and predictors of the dependent variables. In addition to these variables, parental
involvement and parental stress, both based on data collection at the first interview, were
included as independent variables from the existing data set. The dependent variable measures
also came from the existing data set and included the women’s self-reported substance abuse at
the first interview, alcohol use based on the supervision officer’s case notes, drug use based on
the supervision officer’s case notes, violations based on the supervision officer’s case notes,
arrests based on official police reports, and convictions based on official court reports. Several
control variables were also found in the existing data set and included in the regressions (e.g.,
age, a measure for bad neighborhood conditions, being a single parent, and supervision status).
The two remaining controls used in the regression analyses, race and needs, were created based
on variables from the available dataset. Dummy variables for race were generated based on an
existing measure and indicated whether women were black, multi-racial or other. The other
category was the omitted category in regression analyses. As mentioned earlier, in order to
reduce the complexity of the regression analysis, a needs measure was created through the
combination of theoretically relevant variables indicating risk for recidivism, minus those
strengths expected to aid in desistance. The risk variables included employment and financial
needs; family conflict; anti-social friends; criminal history; substance abuse history; current
depression/anxiety; current psychosis, living in an un-safe home; the interviewer’s indication that
the woman exhibited anger or hostility at the third interview; and the interviewer’s indication
that the woman exhibited an anti-social attitude at the third interview. The support variables
included educational strengths, family support, and self-efficacy.
39

In order to address the research questions, several statistical tests were conducted. First,
bivariate analyses were generated between the independent variables, between the dependent
variables, and between the dichotomous independent variables and dependent variables to
determine the extent of association between each variable. Correlations were also generated for
the continuous independent variables that indicated the number of times the women expressed
each parental behavior or motivation over the three interviews. The continuous independent
variables were also correlated with the dependent variables to determine the relationship between
the frequency of parental behaviors and motivations and the dependent variables.
Next, regression diagnostics were carried out to determine whether there were problems
with multicollinearity or outliers. Although most of VIF scores were under 4, and the
historgrams and scatterplots were distributed normally, the large number of control variables
prompted the creation of a needs measure, which was previously explained. Multiple regression
analyses were then conducted for all of the dependent variables to determine if the independent
variables explained significant variance beyond what control variables explained. In this process,
linear regressions were conducted between each dependent variable, with the control variables
entered in step 1, and the independent variables entered in step 2. First the full regression model
R2 was tested for significance, then the R2 change with the addition of the independent variables
to the model with just control variables was examined. Then tests were conducted for the
significance of Betas for each variable. Regressions for each dependent variable were generated
this way for the grouping of both the dichotomous and continuous independent variables with the
controls.

40

Chapter 4: Results
Qualitative Findings
Maternal Motivations
Based on concepts from prior research on age-graded informal social control, and
interview content from the sample, seven categories were created to classify the ways women in
the current study talked about their parent-related motivations and behaviors. The coding
process is detailed in the Codebook, which is located in the Appendix. As stated in the
Codebook, statements that fit with the definition of several categories were coded for all relevant
motivations. Exceptions for multiple coding procedures were limited to the three categories
where women talk about properly parenting their children, as Proving to be a Good Parent, and
Being a Better Parent than Before, would not also be coded in Being a Good Parent. The logic
of this was that the definitions for being a good parent overlapped for the three areas, but the
reasoning for the motivation and behavior differed. Next, an overview of the qualitative findings
highlight the number of women and the frequency of women’s statements coded for each
category. In addition, explanations and examples of each category are detailed.
Attachment to Children
Table 1 – Attachment to Children
# of Times Talked
About in Interviews

1

2

3

4

5

6

7

12

Number of
Women

67

22

16

8

3

2

1

2

Total Number of Women that
Talked About
Attachment = 121

Above, Table 1 shows 121 women in the sample referred to bonds between them and
their children as making their lives better or keeping them away from criminal activity. This
table also displays the number of women and the frequency with which they talked about
attachment to their children. For example, 67 women talked about attachment once, 22 women
talked about it twice, and so forth.
41

Inspection of message content revealed that women in the sample spoke about attachment
in a variety of ways, and in some cases talked differently about how bonds served as forces in
their lives. For example, several women attributed their driving force being their transition into
motherhood during their current criminal justice involvement and mentioned their attachment to
their first child as facilitating their life changes and motivations. Relatedly, some women with
existing children mentioned that their bonds to children became stronger after they experienced
an event that facilitated a realization of their bonds, or increased bonds to children. Some
examples of events that brought about changes in or awareness to these attachments included
separation from children, sobriety, and even changes in interactions that caused their children to
display increased connection to the mother. In the excerpt, below, a woman was asked how her
personal message to be positive and abstain from drug use, helped her do something she was
proud of. This excerpt detailed the women’s response, as she attributed her continued positivity
and abstinence from drugs to her subsequent strengthened relationship with her daughter and the
realization that, as a single mother, she and her daughter only had each other.
Q:
Has the message helped you to do something you were proud of?
A:
Staying home more often with my daughter, growing up. I mean being able to
stay home and not … today when I went to parole, just saying bye to my daughter was
like, “I don’t want to leave you.” I mean we’ve grown a bond closer. I mean I thought we
had before, but we didn’t and now me being home and me being clear minded just … it
helps a lot.
Q:
Mhmm (yes), like you said you’ve developed a closer bond with your daughter
too.
A:
Definitely.
Q:
Like you thought you had that before, but now you really see that you do have
it.
A:
Right.
Q:
How is that really helping you?
A:
It’s making me feel better as a person, and making me, you know … not look
back. I’m a mom, single mom, not married who doesn’t have a father in her life for her
daughter. I’m doing well, I’m doing fine, and she loves me. And I can feel the love, I can
feel the way she looks at me, she knows that, “I love my mom, she’s doing good.” You
know what I mean. It’s just … I don’t know, it’s hard to explain.
42

Similar to the excerpt above, many women in the sample also talked about their new or
existing children as being all they had, or as central in their life. For some women, and in the
example below, the thought of being removed from children caused anxiety.
Q:
And has this message helped you stop from doing something you would later
regret?
A:
Oh, absolutely, you know … he does all the time, sometimes when I have a
stressed out day, it’s like, “God, you know, I would love to smoke a joint,” and I just look
at him, you know, the just the thought of doing something … to risk him not being with
me every day, that’s pretty much all it takes.
Being a Good Parent
Table 2 – Being a Good Parent
# of Times Talked
About in Interviews

1

2

3

4

5

6

7

12

14

18

Number of
Women

45

44

16

14

6

8

1

1

1

1

Total Number of Women that
Talked About
Being a Good Parent = 137

Based on Table 2, above, 137 mothers in the sample spoke about being a good parent
during the interviews. This table also displays 45 of the women in the sample mentioned actions
they’d taken or motivations related to this topic once, 44 spoke of this topic two times during the
interviews, and so on.
A review of the statements coded into this theme revealed that there were numerous ways
women talked about being, or being motivated to be good parents. Many women talked about
their need to be and the actions they’ve taken to serve as a financial support for their children.
Women also talked about many ways they protected or were motivated to safeguard against
threats to their children. Common issues that the sample mentioned were physical harm, mental
illness, neighborhood dangers, and even their efforts or motivations to protect their children from
following a criminal lifestyle. Some of the ways women talked about protecting their children
included teaching them safety measures, knowing their whereabouts, watching them enter school
43

or while they were outside, keeping them away from bad influences, and even confronting those
that posed a threat. An interesting example of a woman that spoke of protecting her daughter due
to dangerous housing conditions is referenced below.
Q:
So you said that your neighborhood isn’t, like, the safest
A:
Like 12 houses on my block is abandoned … 10. They knocked two down.
Q:
Did they? Are they going to knock the rest of them down?
A:
I have no idea. I think I’m going to have to call in and start complaining because
the house next door to mine was boarded up all downstairs … front door and everything.
Now I come out … last night … and noticed that the window across from my daughter’s
room … the board they knocked off from the inside of the house and the doors is wide
open on the house. So I’m like, “Okay no.” I’m going to have to go to sleep, because I
ain’t been to sleep since I noticed that the board been knocked off and the window is
right directly across from her room. So I ain’t been to sleep.
Some women talked about raising their children differently than the way they were
raised to prevent their children from experiencing similar negative experiences and trauma. An
example of this can be seen in the statement below.
Q:
Has the message helped you to do something you were proud of?
A:
Yeah, because I, I look at it like me being a teenager and going through all the
stuff that I went through – being molested, being on drugs, being on the streets, and stuff
like that – that, listening to my mom, that helped me in raising my kids. Because I don’t
have those problems, I don’t have … like my mom wasn’t there for me to talk to her, she
wasn’t that ear that I needed as a teenage girl with certain stuff that you need to know,
she wasn’t that mom, you know. Me learning the hard way which I had to be removed
from my mom’s home and put with somebody else. Me learning that way – that … that’s
not how I want to raise my kids. So everything that my mom didn’t do, I’m doing, you
know, everything she didn’t teach me, I taught my daughters so there will never be a
repeat of what happened to me, you know what I mean. And I, I think the stuff that I went
through in my life made me the person that I am because I’m way stronger than anybody
that I know, you know, some people can’t deal with the fact of losing somebody and then
trying to start over or raising five kids and not having no help for half of the time that you
been raising these five kids. My daughter is gunna be 18, her baby dad got killed when I
was five months pregnant so I never really had that, in the beginning, nobody to help me
out, nobody to get up and make bottles with me, nobody to change a diaper or none of
that. I did that on my own and I’m proud of the fact that my daughter is a great student
and about to go to college, you know, so I feel like I learned from my mom.

44

Several other ways that women talked about being a good parent include providing their
children with motivation or support in school and extra-curricular activities. In addition, women
who said they were motivated to take parenting classes, or said parenting classes helped them
were also included in this category.
Being a Better Parent than Before
Table 3 – Being a Better Parent than Before
# of Times Talked
About in Interviews

1

2

3

4

5

6

7

8

Number of
Women

42

28

9

5

2

1

2

1

Total Number of Women that Talked
About
Being a Better Parent = 90

Table 3, above, indicates that 90 women mentioned a statement related to being a better
parent than before. Additionally, this table shows this topic was mentioned by 42 women just
once, 28 women twice, and so forth.
As mentioned earlier, women that talked about being better parents than they were before
were categorized separately from those that did not mention prior guilt or reference to poor
parenting in the past. A common theme that appeared in these statements was the women’s
desire, or compensatory actions they’ve made, to make up for lost time, and the damage
experienced by their children. A few examples of the past behaviors women talked about making
up for included being incarcerated, emotionally distant, physically distant, and even negligent to
their children. Below is an example of a mother that spoke of being more emotionally available
for her son after becoming sober.
Q:
And how do you feel about this whole situation now, looking back on it?
A:
I never really thought of like the danger that I put my son in making the choices
that I made. I’ve learned a lot, like them sending me to CPI. I’ve learned a lot in ways to
deal with things better – tools to stay clean, NA and AA meetings. I don’t know; it’s
changed me, like I know I hate the idea of being in recovery court because I get scared
sometimes. What if I fail, like what would the consequences be? But, it’s made me a
45

better person and clean. And a mother … I mean I was a mother before, but it’s made me
a better mom and actually being there for my son – because before I was there, but not
mentally or emotionally there.
Another example exhibits the effect their children’s behaviors were capable of having on
a mother’s new ideals on parenting and staying out of trouble with the law. In the example
below, a woman said she stays out of trouble because she thinks about the pain her son
experienced while she was incarcerated, and how he continued to wake up in the middle of the
night to check if she was still there.
Q:
What about … how do you not get caught up in, like, the drugs, or just the
drama, like how do you not get mixed into it?
A:
Kids. I see my kids’ faces.
Q:
Yeah?
A:
I think about all the tears they cried, and all that stuff while I was gone. So it
makes it a lot easier to stay away and say no. My baby coming home and saying,
“Mommy, don’t leave no more.” He woke up … we went to bed at … he went to bed at,
like, two something. I stayed up until, like, 7:00am … 7:30am. But at like 5:30am I got in
the bathtub and get all this stuff off me, shaved, all that stuff. And he wakes up; I thought
he was asleep. He still does it to come make sure I’m still home. So that’s … he shouldn’t
have to be … they should haven’t to go through that. But, like, they say that when I’m –
when your family member is locked up; you do the time with them.
While many women spoke of the effects incarceration on their children, women also
mentioned being better parents in that they changed the roles they held with their children. For
example, instead of acting like a friend, women would talk about how they started serving a
parental role, which was different from their prior inconsistent or lenient parenting style.
Proving to be a Good Parent
Table 4 – Proving to be a Good Parent
# of Times Talked
About in Interviews

1

3

Number of
Women

6

1

Total Number of Women that Talked
About
Proving to be a Good Parent = 7

46

Above, Table 4 shows that 7 women spoke about proving themselves as a good parent.
The table also displays the information that 6 women talked about this once, and one woman
spoke of this three times throughout the interviews.
This variable was created based on women’s mention of proving themselves as good
parents as something that was important to them. Women in the sample spoke of a few types of
people they wanted to prove themselves to, including their parents, the criminal justice system,
governmental programs, their children’s father or his parents, their children, and even
themselves. The women also spoke about a variety of ways they would prove themselves, and
these areas mainly comprised of supporting children and abstaining from drug use. Below, is an
excerpt of a woman that talked about her need to prove to everyone that she was a good parent,
and especially better than her boyfriend’s sister, as the participant found her to be a bad mother.
She also mentioned that she received praise from her mother for proving everyone wrong by
being a good mother.
Q:
So it’s just … it’s a different thing for anyone. It’s just a little message that like
you think about someone saying it and it changes your behavior kind of.
A:
I don’t think it’s … I don’t think it’s a message that someone has given me but
everybody kind of, like people that I … besides my boyfriend and my grandparents …
everybody seemed like I was gunna fail and not get clean and not be a good mother. And
I’ve had my mom tell me … completely showed me, “Like, you’ve proved me wrong. I
really thought that you weren’t gunna be a good mom.” So I guess you could say that and
it just, when I had the baby I just wanted to prove to everybody that, you know, I can do
this, I can be the greatest mom ever. And the way that my boyfriend’s sister is, she’s the
worst mom I’ve ever seen and we live with his parents and she still lives there. And I just
wanted to show everybody … especially that I could be better than her. Cause it like, it
breaks my heart, she’s got three kids and we were pregnant at the same time, pretty
much. Her baby will be one July like 10th I think.
Q:
Okay so only, like, what three months?
A:
Yep.
Q:
Three months age difference.
A:
Yeah, which is crazy and I swear … when you live with someone, I think
getting pregnant is like contagious. Like, when someone’s already pregnant in the house
… cause I just … I … that was weird cause I tried having a baby for the longest time with
47

my ex-fiancé and it just never worked. And that first couple times with my new
boyfriend, it was just like, “Whoa, I’m pregnant.”
Q:
So has thinking about this, kind of, you know your determination to show
people that you can be a good mom and thinking about your mom saying that she didn’t
think that you were gunna do it. Has that helped you to do something that you were really
proud of?
A:
Oh yeah definitely.
Q:
Like what?
A:
My daughter for example.
Q:
Yeah?
A:
Yeah, I’m proud of myself because I didn’t think I could do it. I was scared.
And when I was doing drugs, I didn’t think I was ever gunna be clean. I always just said,
“I’m just gunna do … I like doing drugs, I’m gunna keep doing it. I don’t have to stop.
And I definitely proved myself.
The example below is of a woman that talked about proving to her supervising officer
that she could be a good parent to her children and refrain from drinking and drunk driving.
Q:
Do you have any message that has become memorable to you that you received
since you went on probation or parole?
A:
Just basically she don’t think I can do it and I’ll have to prove to her I can,
basically.
Q:
What is that message word for word?
Q:
Okay, what was that message word for word?
A:
It was just about my children. About, you know, they’re better off because I
violated my probation six times now and I can’t have my children in a home where I
drink and just to let… give up custody, basically.
Q:
And that message was from the PO?
A:
Yes.
Q:
Has the message, like, helped you in any way or has it…?
A:
Yeah, just because I’m very prideful and I have to show her now that, you
know, I can get her back.
Q:
Has the message helped you stop from doing something you would later regret?
A:
Yes.
Q:
Okay, what did it stop you from doing?
A:
Going out and drinking and driving.
Q:
Has the message come to mind when you did something you were not proud of?
A:
Yes.
Q:
And how did it come to mind?
A:
I drank a couple of beers and I’m like “now I’m going to show her, you know,
she’s right.” And so I just stopped drinking.

48

Parental Role Assertion
Table 5 – Parental Role Assertion
# of Times Talked
About in Interviews

1

Number of
Women

8

Total Number of Women that Talked
About
Parental Role Assertion = 8

Table 5 displays the finding that 8 women talked about asserting their parenting role one
time during the interviews. This category included women that spoke of their need to or their
actions taken to display maternal role or authority over their children. Two women talked about
issues they had with the step-mother, or the partner of the child’s father telling the participants’
children call them, “Mom”. In addition, women talked about difficulties parenting due to
inconsistencies in how they wanted to raise their children compared to their children’s prior or
current caregiver. Women also mentioned children being non-compliant, and of their need to
convey themselves as an authority figure to their children. In the example below, the woman
talked about her difficulty and need to serve as the authority figure in her daughter’s life, because
her daughter would call her grandmother when the participant would not allow her to stay at a
friends, as her grandmother served as her caregiver for four years while the participant was in
prison.
Q:
[Multiple Choice] And so you said that situation is getting better or worse?
A:
[Multiple Choice] That … with my kids?
Q:
[Multiple Choice] Yeah, so you guys are starting to kind of figure out what
you’re going to do?
A:
[Multiple Choice] Well the thing is that is pretty much I just want my 12 year
old with me but she’s been with my mom for so long that she basically needs reprogrammed is how I’ve been putting it. But with my mom in the background, what
happens is like my mom went to see my grandma when she was dying. And I would say,
“No Cameron, I’m not letting you do this.” “Well my grandma let’s me do it.” “Well I’m
not your grandma, you’re not letting your friend spend the night on a school night, when
you didn’t get your homework done and you left it at school. And then you want a friend
to spend the night.” “Well then I’ll call my grandma.” And then my grandma says,
“Alright, let her have her friend stay.” Okay so that’s where my issues come in, like we
49

need severe counseling because if this child is gunna ever have any respect and have any
rules, she’s gunna have to stay out of it. And that’s hard for her to on her defense because
she’s raised her on her own for four years until I came back in the picture, so. We’re
having a lot of issues with that and we don’t have the money for counseling and right
now I don’t even have my own home to get her if I wanted her, so I’m working on it
though.
The criminal justice system and the public assistance locations were also mentioned by
women as targets for their attempts to assert their parental roles, because they felt their role as
parents were discounted. In the example below, the woman talks about how it bothered her that
government agencies assessing applicants for cash assistance would say she wasn’t a parent
because she didn’t have full custody of her children.
A:
It doesn’t, it really doesn’t. And to the fact that they tell me I’m not a parent. I
carried two children inside of me for nine months; do not tell me that I’m not a parent
because I do not live under the safe roof as my kids. I am a parent. I parent my children,
you know, I may not live with them but I’m still their mom. I’ll always be their mom.
That’s not a fair thing for them to do, you know.
Custody Motivation
Table 6 – Custody Motivation
# of Times Talked
About in Interviews

1

2

3

4

5

6

7

Number of
Women

36

16

8

6

4

1

3

Total Number of Women that
Talked About
Custody Motivation = 74

Table 6, above indicates that 74 women talked about motivations or actions they took
regarding the custody of their children. This table also shows that during the interviews, 36
women mentioned this topic once, 16 women mentioned it twice, and so forth. Several themes
emerged for actions women made or were motivated to make. One of the most common themes
was women complying with the terms of their supervision to gain support in custody hearings.
The example below shows a woman that talked about staying out of trouble with the law because
she didn’t want to pass up her chance on gaining custody of her children.

50

Q:
And what sort of strategies have you used in the recent months to keep yourself
from getting in trouble with the law and staying away from drug activity?
A:
I just work for my children …
Q:
Did they work? Why?
Q:
How’s that helping … why is that working?
A:
Because I believe deep in my heart I will take about another six months to a
year, probably more towards the year and I will be on my way very close to either having
them back or getting them back. And if I get into any more trouble than it’s not gunna
look good for the courts for me trying to get these children back. And I still have a
chance and I’m not gunna pass that up.
In addition, women talked about how they took actions with Child Protective Services, or
took parenting classes in order to improve their chances of gaining custody. Women also spoke
of hiring lawyers and even just speaking to their supervising officer to safeguard against having
accusations from their families, their children’s fathers, and others putting them at risk of losing
custody or reducing their chances of gaining custody. This category also included women that
spoke of their happiness and thankfulness for regaining custody of children despite great odds. In
the example below a woman described her pride and thankfulness in gaining custody of her
children, and explained how that kept her out of trouble and provided her incentive to take care
of her kids.
Q:
Is there a message you can remember in the past two or three months you
received from anyone that stuck with you at all?
A:
It might have been a little over three months, but the probate judge gave me
custody back, my kids, and told me I should be really proud of myself because a lot of
people don’t get their kids back … so that stuck with me.
Q:
Has the message helped you to do something you were proud of?
A:
Just taking care of my kids.
Q:
Has the message helped you stop from doing something you would later regret?
A:
Definitely, every day [laughs].
Q:
Can you tell me a little about that?
A:
Well, I’m an addict so … but I remember what I lost and have back now and I
can still hear the judge saying that he would never give them back twice [laughs]. You
know that message also sticks with me every day.

51

Routine Activities with Children
Table 7 – Routine Activities with Children
# of Times Talked
About in Interviews

1

2

3

Number of
Women

42

12

5

Total Number of Women that Talked About
Routine Activities with Children = 59

Table 7, above, indicates 59 of the mothers in the sample talked about their children
occupying their time. In addition, this table shows that 42 women talked about this once, 12
talked about it twice, and three spoke of it five times over the interviews. Some women
mentioned time spent on children as a distraction from criminal behaviors, while others said time
spent with their children was all they had time for, or all they had time for among other prosocial or probation stipulated activities. Some women would reference the number of children
and their ages to be especially time consuming. Below, is an example of a woman that said her
toddler age children expended her time, keeping her out of trouble with the law.
Q:
So what about from keeping yourself from getting in trouble. Like, how do you
avoid, just, like, the drama and the drugs and things like that?
A:
I got kids. I keep busy with my kids.
Q:
Did they work? Why?
Q:
Just staying busy with your kids. And that works for you?
A:
Yeah.
Q:
Why do you think that works?
A:
Because they three and four. They need my all the time [laughs].
Although some women simply described their children as being time consuming, others
recognized the pattern change in their daily routines, and some attributed living a normal life
with their children as contributing to obedience with the law. In the example below a woman
talked about how her patterns changed since she became a mother, because the party lifestyle did
not align with raising children.

52

Q:
Do these strategies generally work for you though?
A:
Yeah.
Q:
How specifically like..?
A:
Because if you keep doing it, it gets easier. You know, I used to always drink
and party like every Thursday, Saturday, maybe Sunday, you know what I mean? In the
beginning of my life, when I was in high school, I mean we drank every weekend. I mean
how can you drink every weekend with a kid? It just doesn’t happen. So my patterns have
changed.
Women in the current study also mentioned having children as influencing shifts in their
social networks, however, only one woman in the sample mentioned becoming a mother as
influential in her formation of a new pro-social network with other mothers in her community.
This issue will be mentioned in the Discussion section, as research on mothers has found that
women commonly replaced their negative social networks with pro-social networks of mothers
after becoming parents (Anderson, 1999; Edin & Kelalas, 2005; Kreager, Matuesda & Erosheva,
2010).
Stress from Parenting
While coding interviews for maternal motivations, many illustrations of parental stress
were observed and noted. However, since the focus of the qualitative analysis was on motivation,
and a quantitative measure of parenting stress was used, the stressors were not coded.
Nonetheless, several trends were still recognized for women’s experiences with parent-related
stress in the sample. For instance, some women talked about how difficult it was to make
community supervision a priority, when their children were their main concern. Many spoke of
struggles experienced in trying to support their children financially, and some mentioned being a
single mother as increasing this challenge. The example below provides insight into the common
occurrence for women in the sample that mentioned challenges in legally supporting children
due to poverty. In this example the woman was seeking advice from others in jail on how to

53

avoid criminal activity when her kids were hungry and she didn’t have resources to support
them.
A:
I was in jail, wondering, like, “What am I gunna do when I got home? I ain’t got
no money, I ain’t got this.” And talking to people in my pod, like, “How do you keep
from not doing something illegal with your kid’s sitting there looking at you …
“Mommy, I’m hungry,” and you can’t feed them?”
Relatedly, some women spoke of stress from parenting as causing them to become
involved in the law in the first place due to their inability to provide support in a lawful manner,
or as reducing their abilities to comply with probation orders. Challenges for complying with
supervision due to parenthood and challenges of parenting children while on supervision were
both common for the sample. Mothers also noted challenges in finding child care, and some
mentioned that their stipulations on probation became overwhelming due to their inability to
trust others for child care in their area or social network. Several women even mentioned the
inconvenience resulting from rules that prohibited taking their children with them when they
reported to their supervising officer. Other examples of these trials include mother’s speaking of
being prohibited to have contact with their children’s father due to the father’s criminal
involvement, women being unable to secure benefits because they did not want to go to court
against their children’s father for child support, and women lacking the time to commit to work,
school and parenting children.
Children were also described by the women as victims in their offenses, which many
mentioned as a continuing source of guilt. In some cases, the women explained this victimization
and guilt was related to the separation from their children due to sanctions, their own life
choices, or their prior mental detachment from children. In one case a mother mentioned that she
involved her children in her criminal activity and that this provided her with a great deal of
stress, but that she wanted to do better for her grandchildren because of this. Although the
54

current study did not systematically code the data for stress, this example was coded as Be Better
than Before. In the example below, another mother with parent-related stressors spoke of being a
better mother than before, as she stayed clean after the death of her children’s father, because she
became her children’s only source of support.
A:
It’s like MDOC rehab. So I violated and went there and that’s when he passed
away but, like, I’m glad I was there when it happened.
Q:
So was it a brutal passing away, how’d it happen?
A:
Yeah, he got shot.
Q:
Did he?
A:
Yeah, but I think it would’ve been way worse if I would’ve been out because
that would’ve just escalated a whole other issue and drinking would’ve been a really big
option and—but that I was there and that, you know, I could think about it and I dealt
with it without any substances, at all, like I did cry for three days though but um…
Q:
Typically you turn to the substance, naturally.
A:
Yeah, that’s what I do or used to do.
Q:
Now how long has it been since he’s…?
A:
July.
Q:
This year?
A:
Yes.
Q:
So how long did you have to stay in there after he was killed?
A:
I didn’t get out until September.
Q:
So that really gave you time to grieve. Were you able to go to the funeral?
A:
Oh yeah.
Q:
So that really gave you some good time.
A:
Yeah, and like it just, like, with that happening, like, that’s just more of an
inspiration for me not to want to do anything, because now my son doesn’t have anyone
else but me so like it’s a good and a bad. I kept saying that there and I was like, “You
know, I probably—this doesn’t sound right. It sounds like it’s mean but I’m not trying to
be mean.” That’s… I don’t know.
Q:
It keeps you on track.
A:
Yeah, like a reality check for real, seriously. Like, “Wow, what are you really
doing, you know, doing with yourself?” You know, and if I continue to drink and live
that kind of lifestyle then I won’t be around much longer to be with my son either, so.
I know that my kid needs me more now than ever, that, that just makes it to where I don’t
do anything stupid. Because then he wouldn’t have anyone. Who would take care of my
son, if I’m sitting in jail? So I think about that, like before I didn’t think, and I think that
the in-patient helped me a lot with that because I didn’t think at all. I went to a Cognitive
Behavioral Therapy.
Stressors were frequently mixed in with mothers’ statements in Be a Better Parent than
Before category, and others, including Proving to be a Good Parent, Parental Role Assertion,
55

Attachment to Children, and Custody Motivations. Although many statements within these
categories reflected anxiety and stress, the mothers were still motivated in some way to reduce
their parenting-related stress or guilt. These examples shed light on how parental stress can
become an incentive for change.

Quantitative Findings
Bivariate Analysis
Correlations among Independent Variables
There were several statistically significant correlations among the independent variables
based on Table 8, below. Looking first at the women's expression of maternal motivations that
were coded from the qualitative data as present or not (i.e., yes/no), there are significant
associations of women’s expressions of attachment to their children with motivations to be a
better parent, gain or maintain custody of children, and desire to be a good parent in general.
Mention of being a better parent was also significantly associated with speaking of being a good
parent in general. In addition, parental role assertion was significantly associated with women’s
mention of maintaining or gaining custody and proving themselves as a good parent. Women
who spoke of being a good parent in general also mentioned spending time with their children on
routine activities. There were significant correlations for the measures of parenting involvement
as well. Both motivation to be a good parent and parent-related routine activities had a positive
association with parental involvement scores at the first interview. The final significant
association among the independent variables is for the women’s mention of being a better parent
which displays a significant negative association with the parental stress scores from the first
interview.

56

Table 8 – Correlations among Dichotmous Independent Variables

57

In Table 9, below, correlations were also generated for the continuous independent
variables that indicated the number of times the women expressed each parental behavior or
motivation over the three interviews. Results from the continuous independent variable
correlations indicated that the number of times women talked about parental role assertion was
significantly associated with the numbers of women’s statements surrounding child custody
attainment or maintenance, being a good parent, and proving to be a good parent. The number of
women’s expressions of attachment to their children was also significantly associated with
number of comments about custody motivations and being a better parent. Mentions of being a
better parent had significant negative associations with parental stress scores at the first
interview. Lastly, continuous independent variable correlations showed that statements of routine
activities with children were significantly associated with parenting involvement scores for the
first interview.

58

Table 9 – Correlations among Continuous Independent Variables

59

Correlations among Dependent Variables
Several significant correlations among the dependent variables are displayed below in
Table 10. The bivariate correlations for the dependent variables indicate that women’s selfreported substance abuse at the first interview was significantly associated with their supervising
officer’s case notes regarding their drug use per month while on supervision. Supervising
officer’s notes on violations were related to their case notes on alcohol and drug use, in addition
to official reports from the state police of the women’s total number of arrests during the 18
month period following the first interview. These official arrest reports were also associated with
official court records of the women’s total number of convictions throughout the 18 months
following the first interview.

60

Table 10 – Correlations among the Dependent Variables

61

Correlations of Independent with Dependent Variables
Significant findings were also found in correlations of the both the dichotomous and
continuous independent variables with the dependent variables. Table 11, below, displays
bivariate correlations of the dichotomous independent with the dependent variables, which
indicate that self-reported substance abuse at the first interview, had a significant negative
association with both parental involvement and stress at the first interview. In addition, parenting
involvement scores from the first interview were negatively associated with the supervising
officer’s case notes on the women’s alcohol use while under supervision. Lastly, the
dichotomous independent correlations with the dependent variables showed that the PO’s case
notes on women’s violations per month while under supervision were negatively associated with
women’s mention of routine activities with children, and their parental involvement scores at the
first interview.

62

Table 11 – Correlations of Dichotomous Independent Variables with Dependent Variables

63

Bivariate correlations of the continuous independent measures with the dependent
variable are displayed below, in Table 12. Similar to the bivariate correlation between the
dichotomous independent variables and the dependent variables, the correlation including the
continuous independent variables indicated a significant relationship between the supervising
officer’s notes of the women’s violations while under supervision and the number of times they
talked about routine activities with children. Conversely, the inclusion of the continuous
independent variables illustrates several other trends that did not appear in the correlation of
dichotomous independent variables. One of the significant correlations showed the number of
times women mentioned being a better parent was positively associated with the supervising
officer’s case notes of women’s alcohol use and violations while under supervision. The last
trend found upon continuous independent variable inclusion showed the number of times women
expressed attachment to children had a significant positive association with the PO’s case notes
on drug use during supervision.

64

Table 12 - Correlations of Continuous Independent Variables with Dependent Variables

65

Regression Analysis
Multiple regression analyses were conducted for all of the dependent variables to
determine if the independent variables explained significant variance beyond what control
variables explained. For each dependent variable, an explanation for the regression is provided.
As mentioned in the Method section, the analysis incorporated analyses for the dichotomous
variables for maternal motivations and behaviors, which indicate that the woman had any code
for the theme, and the continuous variables, which indicate the frequency with each topic was
mentioned during all three interviews. Below, regressions of the dependent variables on control
variables and independent variables that include the dichotomous, yes/no codes for presence of a
theme are referred to as dichotomous regressions. Regressions of the dependent variables on
control variables and independent variables that include the frequency with which themes were
coded in the interview transcript are referred to as continuous regressions.
For each regression, the test or significance of the model that includes control and
independent variables, referred to as the whole or full model, is presented first. This test
determined the predictive validity of the regression when both the control and independent
variables were included. Then most relevant to this study, the significance of the R2 change was
reported to indicate whether the inclusion of independent variables explained more variance in
the dependent variable than just the control variables. When the R2 change was significant, the
results of tests of significance of each independent variable were also examined and reported
based on the beta values found in the column entitled β on the two tables presented below each
description of findings. In the tables, one asterisk indicates that the test or variable was
significant at .10, two asterisks indicate these items were significant at .05, and those with no
asterisk were significant at a level above .10. In the description, the level of significance for
betas between .05 and .10 were reported, and those of greater significance, below .05, were just
66

described as significant. The direction of each measure’s relation was based on the positive or
negative values in the column titled B on the regression tables.
In cases where the R2 changes were not significant at .05, meaning the independent
variables did not explain more variance in the dependent variables than the control variables
alone, the independent variables with significant betas were mentioned. However it was noted
that those betas should be considered with caution, as findings for those independent variables
were not as reliable as when there were significant R2 changes. A final paragraph will outline the
significant findings for the control measures and the betas of significance. Theoretical and policy
implications relevant to the findings from the correlations and the regressions are presented in
the Discussion section.
Alcohol Use Based on Supervising Officer Case Notes
Based on Tables 13 and 14, below, the F values for the full model are significant,
meaning the entire model has statistically significant predictive validity in both regressions. The
change in R2 was also significant, meaning the independent variables explained significantly
more variation than the controls alone for supervising officer case note reports of women’s
alcohol use while under supervision in both the dichotomous (R2 change = .061 df =16,162, p =
.013) and continuous (R2 change = .077 df = 16, 162 p = .005) regressions. Considering both
regressions, parenting involvement measured with a scale administered at the first interview was
the only independent variable to have a significant relationship with the dependent variable, and
this relationship was negative. In other words, women’s high parenting involvement was related
to their low levels of alcohol use as indicated in the case notes. In the continuous regression,
there was also a significant positive relationship between women speaking of being a better
parent and their supervising officer’s notes on the women’s alcohol use while under supervision.
In the dichotomous regression, the variable indicating that women talked about proving
67

themselves as a good parent was also positively related to the dependent variable, but this
association was only significant at the level of .058. Overall, women speaking of being better
parent in the continuous regression, and speaking of proving to be a good parent in the
dichotomous regression were both related to higher levels of alcohol use being noted in the case
notes.
Table 13 - Regression of Supervising Officer’s Case Notes of Women’s Alcohol Use while under
Supervision on Control Variables and the Dichotomous Independent Variables
Β
B
Std. Error
Control Variable
Black
0.003
0.005
0.045
Multi/Other Race
-9.62E-05
0.006
-0.001
Age
2.86E-05
0
0.008
Supervision Status
-0.004
0.005
-0.057
Total Needs
7.47E-05
0
0.023
Bad Neighborhood
-0.002
0.001
-0.183**
Single Parent
0.001
0.005
0.02
Independent Variables
Assert Parental Role
-0.007
0.011
-0.053
Attachment
-0.005
0.005
-0.087
Better than Before
0.004
0.004
0.066
Custody
-0.003
0.005
-0.05
Good Parent
-0.005
0.005
-0.085
Prove Good Parent
0.021
0.011
0.15*
Routine Activities
0.004
0.005
0.073
Involvement in Parenting
-0.018
0.005
-0.264**
based on PV Scale at T1
Stress from Parenting based
0
0
0.045
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.145
.061
1.718**
16, 162
.116
2.432**
9, 162

68

Table 14 - Regression of Supervising Officer’s Case Notes of Women’s Alcohol Use while under
Supervision on Control Variables and the Continuous Independent Variables
β
B
Std. Error
Control Variable
Black
.003
.005
.059
Multi/Other Race
-.001
.006
-.010
Age
7.799E-05
.000
.023
Supervision Status
-.004
.005
-.054
Total Needs
1.616E-05
.000
.005
Bad Neighborhood
-.002
.001
-.195**
Single Parent
.002
.005
.028
Independent Variables
Assert Parental Role
-.004
.011
-.028
Attachment
.000
.002
-.011
Better than Before
.003
.002
.186**
Custody
.000
.002
.020
Good Parent
-.001
.001
-.097
Prove Good Parent
.010
.007
.098
Routine Activities
.004
.003
.114
Involvement in Parenting
-.019
.005
-.275**
based on PV Scale at T1
Stress from Parenting based
.000
.000
.051
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.160
.077
1.922**
16, 162
.130
2.785**
9, 162

Number of Violations Based on Supervising Officer Case Notes
In both the dichotomous and continuous regressions on Tables 15 and 16, the F values
were significant, indicating the whole model had statistically significant predictive validity.
However, based on the significance of the R2 change, only for the continuous regression did the
independent variables explain significant variation in the dependent variable beyond what was
explained by the control variables (adjusted R2 = .136 df = 16, 162 p = .021). In this continuous
regression, there was a significant positive relationship between the number of violations and the
69

measure indicating women spoke of being a better parent. In both regressions, the independent
variables for involvement based on a scale value assessed in the first interview and the measure
indicating that a woman spoke of routine activities with children had significant negative
associations with number of violations. However, this finding should be taken with caution for
the dichotomous regression, because the R2 change was not significant.

70

Table - 15 Regression of Supervising Officer’s Case Notes of Women’s Violations while under
Supervision on Control Variables and the Dichotomous Independent Variables
β
B
Std. Error
Control Variable
Black
-.003
.018
-.012
Multi/Other Race
.024
.024
.080
Age
-.001
.001
-.101
Supervision Status
.034
.019
.136*
Total Needs
.002
.001
.158*
Bad Neighborhood
-.002
.003
-.050
Single Parent
-.026
.018
-.108
Independent Variables
Assert Parental Role
-.032
.040
-.062
Attachment
-.005
.017
-.022
Better than Before
.003
.016
.013
Custody
.005
.017
.022
Good Parent
.002
.019
.010
Prove Good Parent
.027
.042
.051
Routine Activities
-.036
.017
-.161**
Involvement in Parenting
-.045
.020
-.173**
based on PV Scale at T1
Stress from Parenting based
-.001
.002
-.065
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.179
.098
2.205**
16, 162
.064
1.405
9, 162

71

Table 16 – Regression of Supervising Officer’s Case Notes of Violations while under Supervision
on Control Variables and the Continuous Independent Variables
β
B
Std. Error
Control Variable
Black
-.006
.017
-.029
Multi/Other Race
.015
.023
.052
Age
-.002
.001
-.136*
Supervision Status
.038
.018
.150**
Total Needs
.002
.001
.121
Bad Neighborhood
-.002
.003
-.034
Single Parent
-.020
.018
-.084
Independent Variables
Assert Parental Role
-.022
.040
-.043
Attachment
-.003
.006
-.041
Better than Before
.016
.006
.229**
Custody
.000
.006
.002
Good Parent
-.001
.003
-.020
Prove Good Parent
.001
.027
.003
Routine Activities
-.023
.011
-.160**
Involvement in Parenting
-.049
.020
-.187**
based on PV Scale at T1
Stress from Parenting based
-.001
.002
-.030
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.214
.136
2.749**
16, 162
.099
2.260**
9, 162

Women’s Self-reported Substance Abuse at the First Interview
Below, Tables 17 and 18, show that the F values for the full models for both the
dichotomous and continuous regressions are significant. However the R2 change when the
independent variables were added was only significant at .099 for the continuous regression, and
was not significant for the dichotomous regression of women’s self-reported substance abuse at
the first interview. In the continuous regression, custody motivations, parenting stress scores, and
parenting involvement scores all displayed significant negative relationships to self-reported
72

substance abuse at the first interview. Based on the dichotomous regression results, the only
independent variable that significantly explained variance was parenting stress measured with a
scale during the first interview, which exhibited a significant negative relation to self-reported
drug use. Parenting involvement scores measured at the first interview also had a negative
relation with self-reported drug use in the dichotomous regression, but the association only met
significance at .075. As previously mentioned, findings about individual variables when there
was not a significant R2 change should be considered with caution, because they are less reliable
than those findings when there was a significant R2 change.

73

Table 17 – Regression of Self-Reported Substance Abuse at the First Interview on Control
Variables and the Dichotomous Independent Variables
β
B
Std. Error
Control Variable
Black
.464
.196
.181**
Multi/Other Race
.296
.263
.085
Age
-.007
.011
-.045
Supervision Status
.024
.209
.008
Total Needs
.051
.012
.342**
Bad Neighborhood
.049
.038
.092
Single Parent
-.209
.203
-.073
Independent Variables
Assert Parental Role
.247
.440
.041
Attachment
-.230
.186
-.089
Better than Before
-.023
.177
-.009
Custody
-.131
.188
-.051
Good Parent
-.067
.206
-.024
Prove Good Parent
-.142
.462
-.022
Routine Activities
-.261
.191
-.097
Involvement in Parenting
-.402
.225
-.129*
based on PV Scale at T1
Stress from Parenting based
-.039
.019
-.151**
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.282
.211
3.973**
16, 162
.062
1.562
9, 162

74

Table 18 – Regression of Self-Reported Substance Abuse at the First Interview on Control
Variables and the Continuous Independent Variables
β
B
Std. Error
Control Variable
Black
.435
.195
.170**
Multi/Other Race
.395
.261
.113
Age
-.005
.011
-.031
Supervision Status
-.002
.209
-.001
Total Needs
.052
.012
.347**
Bad Neighborhood
.044
.038
.083
Single Parent
-.197
.201
-.068
Independent Variables
Assert Parental Role
.454
.457
.075
Attachment
.066
.066
.086
Better than Before
-.071
.071
-.084
Custody
-.138
.066
-.158**
Good Parent
-.036
.039
-.067
Prove Good Parent
-.178
.307
-.041
Routine Activities
-.077
.120
-.046
Involvement in Parenting
-.459
.224
-.148**
based on PV Scale at T1
Stress from Parenting based
-.040
.019
-.157**
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.286
.215
4.055**
16, 162
.066
1.676*
9, 162

Drug Use Based on the Supervising Officer’s Case Notes
Below, Tables 19 and 20 indicate that F values were significant for both full-model
regressions. However, the R2 changes were not significant for either regression in explaining
variation in supervising officer’s case notes on the participant’s drug use during supervision.
Also, none of the independent variables had betas that significantly explained the variance of this
dependent variable.

75

Table 19 – Regression of Supervising Officer’s Case Notes of Drug Use while under Supervision
on Control Variables and the Dichotomous Independent Variables
β
B
Std. Error
Control Variable
Black
.005
.009
.048
Multi/Other Race
-.018
.012
-.119
Age
-.001
.000
-.137*
Supervision Status
.035
.010
.259**
Total Needs
.002
.001
.280**
Bad Neighborhood
.000
.002
.009
Single Parent
-.022
.009
-.167**
Independent Variables
Assert Parental Role
.010
.021
.038
Attachment
.005
.009
.042
Better than Before
-.011
.008
-.100
Custody
-.004
.009
-.039
Good Parent
.000
.010
-.002
Prove Good Parent
-.019
.022
-.067
Routine Activities
-.002
.009
-.017
Involvement in Parenting
-.005
.011
-.036
based on PV Scale at T1
Stress from Parenting based
.000
.001
-.040
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.192
.133
2.709**
16, 162
.019
.426
9, 162

76

Table 20 – Regression of Supervising Officer’s Case Notes of Drug Use while under Supervision
on Control Variables and the Continuous Independent Variables
β
B
Std. Error
Control Variable
Black
.005
.009
.044
Multi/Other Race
-.020
.012
-.126
Age
-.001
.000
-.146**
Supervision Status
.032
.010
.238**
Total Needs
.002
.001
.269**
Bad Neighborhood
.000
.002
.016
Single Parent
-.019
.009
-.147**
Independent Variables
Assert Parental Role
.025
.021
.094
Attachment
.004
.003
.123
Better than Before
.001
.003
.029
Custody
-.003
.003
-.088
Good Parent
-.002
.002
-.092
Prove Good Parent
-.018
.014
-.092
Routine Activities
.000
.006
-.005
Involvement in Parenting
-.004
.010
-.032
based on PV Scale at T1
Stress from Parenting based
.000
.001
-.016
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.192
.159
2.967**
16, 162
.034
.796
9, 162

Official Arrest Records
Tables 21 and 22 indicate that the F value did not meet significance, meaning the whole
model was not significant for this regression. In addition, the R2 change was not significant at .05
or .10 in explaining the variance in the women’s official arrests. However, based on the beta
values, several of the variables significantly explained variation independently. The continuous
regression showed a negative relationship between official arrests and the independent variable
77

measures indicating how often women spoke of attachment to children at a significance of .058
and how often they spoke of asserting their parental role at a significance of .076. In the
dichotomous regression, women that spoke about asserting their parental role also had a negative
relation to arrests based on police records with a significance of .058. Despite these findings for
single variables, there is very limited evidence that any of the variables explain a significant
amount of the variation in official arrests.

78

Table 21 – Regression of Official Arrest Records on Control Variables and the Dichotomous
Independent Variables
β
B
Std. Error
Control Variable
Black
-.029
.130
-.019
Multi/Other Race
.062
.174
.030
Age
-.010
.007
-.113
Supervision Status
.274
.139
.155**
Total Needs
.005
.008
.059
Bad Neighborhood
-.015
.025
-.049
Single Parent
.035
.135
.021
Independent Variables
Assert Parental Role
-.557
.292
-.156*
Attachment
-.032
.124
-.021
Better than Before
.089
.118
.061
Custody
-.058
.125
-.038
Good Parent
.044
.137
.026
Prove Good Parent
.416
.307
.110
Routine Activities
-.079
.127
-.050
Involvement in Parenting
.140
.149
.077
based on PV Scale at T1
Stress from Parenting based
-.011
.013
-.072
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.089
-.001
.987
16, 162
.044
.875
9, 162

79

Table 22 - Regression of Official Arrest Records on Control Variables and the Continuous
Independent Variables
β
B
Std. Error
Control Variable
Black
-.035
.129
-.023
Multi/Other Race
.047
.172
.023
Age
-.013
.007
-.145*
Supervision Status
.325
.138
.183**
Total Needs
.003
.008
.034
Bad Neighborhood
-.012
.025
-.039
Single Parent
.023
.133
.014
Independent Variables
Assert Parental Role
-.540
.302
-.152*
Attachment
-.083
.043
-.181*
Better than Before
.064
.047
.129
Custody
-.008
.043
-.016
Good Parent
-.001
.026
-.004
Prove Good Parent
.146
.203
.056
Routine Activities
-.081
.080
-.082
Involvement in Parenting
.087
.148
.048
based on PV Scale at T1
Stress from Parenting based
-.011
.013
-.074
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.102
.013
1.152
16, 162
.058
1.154
9, 162

Official Conviction Records
Below, in Tables 23 and 24, the F values for the full models were not significant for
either regression. Similar to the dependent variable measuring official arrest, the variance in
official convictions of the women based on police reports could not be significantly explained by
the grouping of the controls before or after independent variable inclusion based on R2 changes.
Although, none of the independent variables explained conviction at a level of significance, in
80

the continuous regression, parental stress based on a survey at the first interview was significant
at .084 and the number of times women talked about asserting their parental role was significant
at .096. Both of these variables were negatively associated with convictions, and explained more
variation in conviction than the other independent variables. These significant betas, and those
for official arrest records, are both less reliable because the R2 changes were not significant.

81

Table 23 - Regression of Official Conviction Records on the Control and the Dichotomous
Independent Variables
β
B
Std. Error
Control Variable
Black
-.028
.121
-.019
Multi/Other Race
-.102
.162
-.052
Age
-.006
.007
-.069
Supervision Status
.380
.129
.227**
Total Needs
.000
.007
-.004
Bad Neighborhood
-.015
.023
-.051
Single Parent
-.092
.125
-.057
Independent Variables
Assert Parental Role
-.402
.272
-.119
Attachment
-.020
.115
-.014
Better than Before
.088
.110
.063
Custody
.009
.116
.007
Good Parent
.190
.127
.120
Prove Good Parent
.278
.285
.077
Routine Activities
-.113
.118
-.075
Involvement in Parenting
.157
.139
.090
based on PV Scale at T1
Stress from Parenting based
-.021
.012
-.143*
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.118
.031
1.359
16, 162
.059
1.208
9, 162

82

Table 24 - Regression of Official Conviction Records on the Control and the Continuous
Independent Variables
β
B
Std. Error
Control Variable
Black
-.030
.122
-.021
Multi/Other Race
-.133
.163
-.069
Age
-.009
.007
-.106
Supervision Status
.401
.130
.240**
Total Needs
-.002
.008
-.024
Bad Neighborhood
-.014
.024
-.048
Single Parent
-.084
.125
-.052
Independent Variables
Assert Parental Role
-.476
.285
-.141*
Attachment
-.048
.041
-.111
Better than Before
.040
.044
.086
Custody
.026
.041
.053
Good Parent
.018
.024
.061
Prove Good Parent
.116
.191
.047
Routine Activities
-.078
.075
-.082
Involvement in Parenting
.142
.139
.082
based on PV Scale at T1
Stress from Parenting based
-.021
.012
-.147*
on PV Scale at T1
R2
Adjusted R2
F
df
R2 Change
F Change
df

.107
.019
1.216
16, 162
.048
.969
9, 162

83

Findings for the Control Variables
Table 25 - Findings of Significance for Controls in Dichotomous and Continuous
Regressions
Dependent Variables

Dichotomous
Control / - or +
relationship with DV

Continuous
Control / - or +
relationship with DV

Alcohol Use from Case Notes

Bad Neighborhood / -

Bad Neighborhood / -

Violations from Case Notes

Supervision Status / +
Age/ -

Supervision Status / +
Age/ -

Self-Report Substance Abuse

Needs / +
Black / +

Needs / +
Black / +

Drug Use from Case Notes

Needs / +
Supervision Status / +
Age / -

Needs / +
Supervision Status / +
Age / -

Arrests based on Police Records

Supervision Status/ +
Age/ -

Supervision Status/ +
Age/ -

Convictions Based on Court Records

Supervision Status / +

Supervision Status / +

Note: Control Variables that meet significance below .05 are colored black, while those that meet significance below
.10 are colored gray.

Although not the focus of the present study, it is informative to examine all the
relationships of control variables to various dependent variables, net of the effects of other
variables (condensed above in Table 25). Based on results in Tables 13 and 14, in the
dichotomous and continuous regressions, the scale measure for bad neighborhood conditions had
a significant negative relationship with women’s alcohol use indicated by supervising officer’s
case notes on the . Specifically, living in a bad neighborhood was significantly associated with
fewer case notes indicating alcohol use.
In addition, Tables 15 and 16 display violations based on case notes of supervising
officers as having a significant positive relationship to women’s supervision status, which
indicated that parole status was significantly associated with more violations while on
supervision. The control measure for age was negatively related to the dependent variable, but

84

this association was only significant at the .064 level. Older age was associated with fewer
violations noted in the case notes.
A few of the control variables in Tables 17 and 18 reached significant levels in
explaining the variation in the women’s self-reported substance abuse at the first interview. In
both the dichotomous and continuous regressions, the measure for the women’s total needs and
the indicator of women being black had significant positive relationships with self-reported
substance use at the first interview. In other words, women with higher needs, indicated by risk
for recidivism minus strengths, and being black were both associated with higher scores on selfreported substance abuse based on a scale administered at the first interview.
In Tables 19 and 20, supervision status and the measure of the women’s needs both had
significant positive relations to the dependent variable, so those women on parole and those with
higher needs scores were both associated with more indications of drug use based on their
supervising officer’s case notes. On the other hand, the case note indicator for drug use was
significantly negatively associated with age and being single parents. However, age was only
significant at the .097 level in the dichotomous regression, and the .065 level in the continuous
regression.
In the dichotomous and continuous regressions on Tables 21 and 22, supervision status
had a significant positive relationship to the arrest variable, meaning that parole status was
associated with more official arrests. In the continuous regression, age had a significant
negative relationship with arrest, meaning older age was associated with fewer arrests based on
official police records. The dichotomous regression also revealed this negative relation between
age and arrests, but this was only significant at .065.

85

In Tables 23 and 24, the only control variable to significantly explain variation in official
records of conviction was supervision status, and it explained a significantly positive association
with conviction in both the regressions. This means that parole status was significantly
associated with more records of official conviction.

86

Chapter 5: Discussion
Several statistical tests were carried out to address the research hypotheses. First, the
explanatory power of the whole model and the significance of change for the independent
measures was considered for dichotomous and continuous regressions, as these both differed for
each of the six measures of recidivism that were included. Women’s alcohol use based on their
supervising officer’s case notes was the only measure that was explained significantly more by
the independent variables than the controls alone for both the dichotomous and continuous
regressions. For the continuous regressions, for which the number of comments women made
about material motivations were the independent variables, the variance in violations based on
case notes and women’s self-reported substance abuse scores from the first interview were also
significantly explained by the inclusion of the independent measures. However, the relationship
for women’s self-reported substance abuse was only significant at a level of .099. Therefore, the
independent variables were not able to explain significant variance beyond the controls for the
dichotomous regressions of violations indicated by case notes and of women’s self-reported
substance abuse, or for the dichotomous and continuous regressions for drug use based on case
notes, official arrest records, and official conviction records.

87

Table 26 - Findings of Significance for IVs in Dichotomous and Continuous Regressions
Dependent Variables

Dichotomous
IV/ - or + relationship with
DV

Continuous
IV/ - or + relationship with
DV

Alcohol Use from Case Notes

Prove Good Parent/+

Be Better than Before/+

Violations from Case Notes

Routine Activities/ -

Routine Activities/ Be Better than Before/ +
Custody Motivations/ -

Self-Report Substance Abuse
Drug Use from Case Notes

Attachment/ -

Arrests based on Police Records
Convictions Based on Court Records

Parental Role Assertion/ -

Parental Role Assertion/ -

Note: Independent Variables that are more reliable due to R2 Significance changes are in black, while those that are
less reliable because the whole model did not indicate a significant change beyond the controls are colored gray.

Second, each independent measure’s significance and Beta value were also considered to
determine each variable’s directional strength in explaining women’s recidivism. The
independent variables reflected seven types of maternal motivations coded from the qualitative
data and two quantitative measures from the first interview, one an indication of parenting
involvement, and the other of parenting stress. Based on the associations between the
independent variables and the measures of recidivism from the regressions in Tables 13-24
(condensed above in Table 26), several conclusions were made regarding the research questions.
First, a few of the parent-related motivations and behaviors found in the interview content were
associated with reduced recidivism, because they had a negative relationship to the dependent
variables based on Beta values. Most reliably, both regressions for women’s talk of routine
activities with children were associated with fewer violations based on case notes. In the
continuous regression, women’s custody motivations were significantly associated with
decreased self-reported substance abuse at the first interview. The other independent variables
based on interview content, and that had significant negative Betas, were less supported as
88

significant, because inclusion of the independent variables did not explain significant change in
the dependent measure beyond the controls. These less reliable findings showed that fewer
arrests were related to attachment (for the continuous regression) and parental role assertion (for
the dichotomous regression). Parental role assertion was also less reliably related to fewer
convictions for both regressions.
A review of the Betas with significant positive relationships with the dependent variable
shows that, contrary to the proposition of the first research question and second research
questions, two parent-related motivations and behaviors were associated with increased
recidivism. In continuous regressions, the number of times women spoke of being a better parent
than before was related to both increased alcohol use and increased violations based on
supervising officer’s case notes. In addition, women’s mention of trying to prove they were
good parents was also related to alcohol use based on supervising officers’ case notes, but only
in the dichotomous regression. These findings, paired with those linking women’s parent-related
motivations and behaviors to desistance, provide mixed results for the first two hypotheses. In
some cases, the dichotomous and continuous regressions identified significant associations of
different independent variables relations with the recidivism measures. The different findings for
the dichotomous and continuous maternal motivation variables suggest that whether a woman
spoke of a particular motivation and the number of times she spoke of a motivation vary in what
they tell us about the motivation’s relation desistance. This trend appeared for the motivation to
be a better mother than before, which was significant in the continuous regression and was
spoken of more often by more women than the measure for proving to be a good parent, which
was only significant in the dichotomous regression.

89

Based on the quantitative scales included as independent variables, parenting
involvement significantly explained fewer indications of alcohol use and violations based on the
case notes, and lower scores for women’s self-reported substance abuse at the first interview.
Higher levels of parenting stress also explained fewer violations based on case notes, and fewer
convictions based on official records. However, stress as an explanation for convictions is less
reliable due to the same issues mentioned before, since there was no significant change in
explained variance for convictions for the whole model beyond the model with just controls.
In addition to providing insight to address the research questions, the findings contribute
to the debate on the components necessary for bond formation to be effective in contributing to
desistance. As mentioned in the review of the literature, Giordano and colleagues (2002)
proposed that cognitive shifts were necessary to form strong bonds for desistance. In a later
study, Giordano (2010) investigated women’s parent-related motivations and labeled these as
indicators of a cognitive shift, but they were not able to statistically support this concept, and
concluded that barriers in other aspects of life impeded the effect of motivations on reducing
recidivism. Sampson and Laub (2003) maintained that involvement with the source of bonds?
(e.g., children) would be sufficient in supporting desistance. The interesting findings were that
routine activities with children were significantly associated with fewer violations based on
supervising officer case notes. In addition, parenting involvement was significantly associated
with fewer violations and fewer indications of alcohol use based on supervision case notes, in
addition to reduced self-reported substance abuse. With this finding, the second research
question was addressed, affirming that, parental involvement was related to less recidivism
indicated by two different measures.

90

Taken as a whole, these findings suggest routine activities and parental involvement were
more indicative of compliance with the law than the other motivations. This conclusion is
strengthened because it holds for indicators of recidivism based on both the women’s and their
supervising officers’ accounts. On the other hand, motivation to be a good parent had no
significant relationship to any dependent variables, parental role assertion had less reliable
relationships to desistance, and attachment also had less reliable relationships with desistance, or
in some cases was even positively related to recidivism. Most importantly, motivation to be a
better parent than before and to prove oneself as a good parent were both significantly and
positively related to recidivism. These findings suggest that informal social control through
bonds to children may be based more on spending time with children and on activities with
children than on comments about motivations to be a good or better mother. This finding is in
line with Sampson and Laub’s (2003) proposition that involvement alone was sufficient in
forming bonds related to desistance. This provides further support that cognitive transformations
mentioned in the literature review may not be necessary, or effective in determining those that
will desist. Alternatively, motivation based on interview content may not be sufficient in
capturing the process of cognitive transformation.
The hypothesis for the fourth research question was contradicted by opposite results,
more specifically, stress from parenting was significantly related to lower scores in women’s
self-reported substance abuse at the first interview, and less reliably related to fewer convictions.
In addition, custody motivations, which, based on the content from the qualitative data caused
the women’s anxiety and stress, were also significantly related to lower scores in women’s selfreported substance abuse at the first interview. These findings provide support for prior research
suggesting that women may be motivated by their parent-related stressors (Arditti & Few, 2008;
91

Edin & Kefalas, 2005; Giordano, 2010; Kreager, Matsueda, & Erosheva, 2010; Thompson &
Petrovic, 2011). In addition, these findings also are in line with the motivations found for many
of the women that experienced stress in the current study’s qualitative data.

Limitations
In all, there were mixed findings across dependent and independent measures for all four
research questions. There were two primary issues based on the hypotheses that contributed to
inconsistent findings. First, the hypotheses did not consider the range and complexity of accounts
women would have regarding their parent-related motivations and behaviors. A more thorough
or more specific investigation should be conducted in the future to provide more conclusive
results for the hypotheses. The second issue was the hypotheses did not predict the extent of
differences that would exist in the influence independent measures would have on the six
recidivism variables. Although the limitations based on the hypotheses provide inconclusive
answers for the research questions, the analyses provided a vivid portrayal of statements that
were related to a range of desistance and persistence for an array of parent-related motivations
and behaviors.
Unfortunately, the current study was also not able to assess which control variables
affected women that persisted, but talked about parent-related motivations and behaviors. These
issues will be addressed in a follow-up study using this data that will examine the interaction
effects between control variables of interest and independent variables suggested by prior
research and the findings in the current study. Although a follow-up study will provide more
insight into the control variable’s effects on motivations and desistance, abuse throughout the
women’s lives was not captured in the current study, even though it is considered an important
factor in previous research on women’s criminality (Salisbury & Van Voorhis, 2009). In
addition, the needs measures in the current study combined several risk factors and protective
92

factors against recidivism, so the interaction effect each of these measures present on their own
will not be uncovered if identical data is used in the follow-up.
In addition, as a secondary study of an existing data set, this examination was unable to
capture several parent-related variables that were identified as significant mediating or
moderating factors in previous studies. First, this study did not account for the transition into
motherhood, which was found as an important indicator of women’s desistance in prior research
(Edin & Kefalas, 2005; Kreager, Matsueda, & Erosheva, 2010; Thompson & Petrovic, 2011).
Although the qualitative assessment provided some insight into the maternal circumstances for
some of the women’s transitions into motherhood, there was no quantitative measure of this and
it was not uniformly detectable based on the interviews because there were no questions to that
end. Similarly, there was no measure for women’s custody transitions or whether they were
separated and reunited with children. Future research on parent-related motivations and
desistance should ask questions about these issues to refine the conceptualization of bond
changes between women and their children.
Several other issues that were not addressed in the current study were the women’s
relationships with caregivers of their children; their contact with children while separated, either
during imprisonment or through other forms of separation like residential treatment; the number
of and the age(s) of children (Barnes & Stringer, 2014). In addition, the qualitative analyses only
discovered one woman that mentioned replacing a negative social network with a pro-social
network of other mothers. However this idea, inspired by Anderson’s (1999) concept of “baby
clubs”, was found to be related to women’s transitions into motherhood, and their desistance in
studies that used age-graded informal social control to examine mothers in disadvantaged
neighborhoods (Edin & Kelalas, 2005; Kreager, Matuesda & Erosheva, 2010). Mother’s social
93

networks changes, and the other parent-related variables that were not captured in the current
study should also be investigated in future studies on maternal desistance to reveal the influence
more specific parental circumstances have on parent-related motivations and desistance over the
life course. Investigations of this nature will help uncover unmet needs of maternal offenders,
and services that may help women live up to their parent-related motivations and abandon
criminal lifestyles.

94

APPENDIX

95

APPENDIX

Codebook for Maternal Involvement and Motivation
Rules
1. Read over each passage originally coded as motivations and stressors and code for the
motivations based on the themes on the next page.
2. Do not code in categories when questions are leading and the responses are closeended, yes or no responses. For example, you would not code the following
statement:
-“Does having your kids around make your life better?”
“Yes.”
3. Statements that fit with the definition of several categories were coded for all relevant
motivations. Exceptions for multiple coding procedures were limited to the three
categories where women talk about properly parenting their children, as Proving to be
a Good Parent, and Being a Better Parent than Before, would not also be coded in
Being a Good Parent. The logic of this was that the definitions for being a good
parent overlapped for the three areas, but the reasoning for the motivation and
behavior differed.
4. If children are in the content or are the senders of memorable messages and the
women says these messages serve as motivations, help them, or if the women says the
message helped her do something she was proud of or stopped from doing something
she would later regret, then code in the suitable theme. See example below with a
slight variation.
-“What is that message?”
“I just think about my children’s faces”
“Has that helped you do something you’re proud of?”
“Yes, I didn’t use because I thought about them.” Code as Attachment to
Children
-“What is that message?”
“I just think about my children’s faces”
“Has that helped you do something you’re proud of?”
“Yes, I didn’t use because I thought about them and how that would affect their
future.” Code as Good Parent
5. When unborn children, grandchildren or other children that are not the biological
child of the offender should still be coded in the themes. These will hold the same
weight as codes for biological children, but will be referenced as examples of how
96

women are motivated by children in these categories similarly and differently than
motivations for biological children.
6. There are statements related to parental stressors within the unitizing document that
will be as examples, but not coded for because the quantitative measures for stressors
are more reliable because they were uniformly posed.
Themes
1. Attachment to children
Expresses children as being central in their thought processes; expresses enjoyment in
spending time with their children; expresses that being away from kids is upsetting.
Expresses that she wants to keep children around her or give birth and/or care for child
despite challenges.
For example, in response to questions about what keeps them away from criminal activity
or what makes their life better:
-“It’s about my kids now”
-“I spend more time with my kids because they are the most important thing in my life”
-“I stay away from those activities because I take my kids everywhere and I hate
spending time without them”
-“Everyone tried to persuade me to get an abortion because it would be very difficult to
get through school, but I didn’t because I wanted to keep the baby.”
For statements like:
-“What are you doing to keep yourself out of trouble?”
“I just hang out with my kids.” Code as Routine Activities because this is not explicitly
attachment
-“I could never touch drugs again, because I remember how difficult it was being away
from my kids while I was in prison and I would never want to put them through that
again” Code as Be Better Parent then Before
2. Routine Activities
Expresses involvement in children’s life and activities as occupying their time or being of
priority over other activities, especially criminal activities; kids serve as a distraction and
consume time.
For example, in response to questions about what keeps them away from criminal activity
or what makes them abide the law or change ways:
-“I just stay focused on my kids …”
-“My kids and I go to the park”
-“My kids keep me busy”
-“I don’t have time to even think about criminal activity because I am too busy doing
things with my kids”
-“Spending more time with my kids”
97

If the above statements are in response to a question about making life better, or the
woman expands and says more about enjoying this time spent with children, or how
spending time with children makes their life better, code as Attachment to Children
instead.
3. Good Parent
a. Good Parent
Serve as a positive role model (practice what you preach, be strong for children
either through guarding against temptations or addressing parental difficulties);
serve as financial support; serve as a guardian; protect child from threats like
mental illness, physical harm, bad influences, neighborhood characteristics etc.
Pursues parenting classes due to challenges in disciplining children and says that
she needs or has benefited from parenting classes. Expresses aspiration and
appreciation for parenting tips from PO or others but must also speak of this as
necessary to combat parental stress. Also code for women that talk about things
that their partners or fathers could do to make to would benefit children.
-“I watch my kids as they walk to the bus.”
-“I won’t let my kids be raised the way I was, because I don’t want them to end
up breaking the law like I did.”
-“My PO referred me to parenting classes and it made things better because I was
able to learn new ways to discipline my child consistently.”
-“It would be better if my child’s father would take a parenting class, because he
has issues raising our child.”
Only code for parenting classes that the women mention for being better parent,
those that are mentioned to fulfill custody requirements should be coded in
Custody.
b. Prove Good Parent
For new moms, or those that mentioned needing to prove self as good parent
without indication of prior parenting failures or guilt, prove to others that said you
would not be a good parent like her PO, her parents, a partner, her children, and
even to herself.
-“When I was pregnant, everyone told me to get an abortion, and it just made me
want to show them how good a mom I am going to be.”
-“My grandma keeps telling me that I don’t know how to take care of my kids,
this makes me mad every time and I just want to show her how good of a mom I
am by taking even better care of my children.”
“I have to prove that I am a good parent to myself, and then I think the guilt will
go away.”
c. Be Better Parent than Before
98

Motivated to spend time with children, make up for lost time with children, or be
a better parent due to guilt brought on by prior actions or criminal sanctions either
with the child that was effected, new children, grandchildren.
-“I was out of their lives for two years, and I feel so bad about that. Now I am
trying to be the best mother that I can for them.”
-“My daughter keeps trying to guilt me into being lenient on my rules. It’s very
difficult to get past the guilt I have from being in jail and emotionally distant all
these years, but I am just trying to be a good parent now to make up for that.”
-“My kids are mostly grown and I didn’t do that great of a job with them, so now
I am doing all I can to make sure I play a big role in raising [Granddaughter].”
4. Custody
Keep kids in custody, gain more custody or visitation, or get kids back
This includes women who are motivated to be proactive in behavior or complying with
supervision stipulations (letting PO know of any issues like people that may contact PO
with lies about female’s treatment of children) to maintain or gain legal parental rights.
-“I am seeking full custody of my child.”
-“I am fighting for my kids.”
-“The father of my child threatened to call my PO and tell her lies about how I treat my
kids bad, because he doesn’t want me to get to see them. I showed [PO] the voicemail
just in case he was to contact her about this.”
5. Asserting Parental Role
The women talks about how she wants to or had to talk to her child, family members,
caregivers, and even the criminal justice system about how she is the parent to this child,
because she feels her parent status or her parenting is being diminished, threatened, or
undermined.
For example:
-“No, no, the only problem that I ever have is I tend to – family tends to be mother-henish. [Laughs] So, there are times I have to tell my mom or Grandma that “I’m the parent,
you’re the grandparent. So, you know, I can’t ground them for this and then you guys go
behind my back and it’s okay to, you know, do that sort of stuff.” [Laughs] Family drama
…”
-“My ex-husbands wife called me and told me that she wants [Son] to call her Mommy,
and I told her that wasn’t going to happen because I am the mother to this child.”

99

REFERENCES

100

REFERENCES
Altheide, D.L., & Coyle, M.J. (2006).Smart on crime: The new language for prisoner release.
Crime, Media, Culture: An International Journal, 2(3), 286-303 .doi:
10.1177/1741659006069561
Arditti, J. A. and Few, A. L. (2006). Mothers’ reentry into family life following incarceration.
Criminal Justice Policy Review, 17(1), 103-123.
Arditti, J. and Few, A. (2008).Maternal distress and women’s reentry into family and
community life. Family Process 47(3): 303-321.
Austin, J. (2001). Prisoner reentry: Current trends, practices and issues. Crime & Delinquency,
Vol. 47(3), 314-334.
Austin, J., Bruce, M. A., Carroll, L., McCall, P. L., & Richards, S. C. (2001). The use of
incarceration in the United States: American Society of Criminology National Policy
Committee. Critical Criminology: An International Journal, 10, 17-41.
Bales, W. D., and Mears, D.P. (2008). Inmate social ties and the transition to society: Does
visitation reduce recidivism? Journal of Research in Crime and Delinquency, 45(3), 287321.
Barnes, S. L. and Stringer, E. C. (2014). Is motherhood important? Imprisoned women’s
maternal experiences before and during confinement and their postrelease expectations.
Feminist Criminology, 9(1), 3-23.
Beck, A. J. (2000).Prisoners in 1999. Washington, DC: Bureau of Justice Statistics.
Benda, B. B. (2005). Gender differences in life-course theory of recidivism: a survival analysis.
International Journal of Offender Therapy and Comparative Criminology, 49, 325-342.
Blaustein, M. and Kinniburgh, K. (2007).British Psychological Society, Briefing Paper 26, 4853.
Blumstein, A., Cohen, J., Roth, J. A., and Visher, C. A, eds. (1985). Criminal careers and
career criminals'. Washington, DC: National Academy Press.
Blokland, A. A. J., & Nieuwbeerta, P. (2005). The effects of life circumstances on
longitudinal trajectories of offending. Criminology, 43, 1203–1240
Braman, D. (2004). Doing time on the outside: Incarceration and family life in urban
America. Ann Arbor: University of Michigan Press.
Boyd, S. C. (1999). Mothers and illicit drugs: transcending the myths. Toronto; Buffalo:
University of Toronto Press.
101

Bureau of Justice Statistics.(1997). Correctional populations in the United States, 1995.
Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics. (2014). Recidivism of prisoners released in 30 states in 2005:
Patterns from 2005 to 2010. Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics.(2013a). Correctional populations in the United States, 2012.
Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics.(2013b). Probation and parole in the United States, 2012.
Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics.(2013c). Jail inmates at mid-year 2012 – Statistical tables.
Washington, DC: U.S. Department of Justice.
Bureau of Justice Statistics.(2013d). Prisoners in 2012 – Advance counts. Washington, DC: U.S.
Department of Justice.
Chesney-Lind, M. & Shelden, R.G., (1998). Girls, delinquency, and juvenile justice (2nd ed.).
Belmont, CA: West Wadsworth.
Clear, T. R. (1997). Evaluating intensive probation: The American experience’, In Mair, G. (ed.)
Evaluating the effectiveness of community penalties.Aldershot: Avebury.
Clear, T.R. (2003). The problem with “addition by subtraction”: The prison-crime
relationship in low-Income communities. In Mauer, M., & Chesney-Lind, M. (Eds.)
Invisible Punishments. (56-74). New York: The New Press.
Clear, T. (2007). Imprisoning communities: How mass incarceration makes disadvantaged
communities worse. New York: Oxford University Press.
Clear, T., Waring, E., and Scully, K. (2005). Communities and reentry: Concentrated reentry
cycling. In J. Travis and C. Vishers (Eds.), Prison reentry and crime in America (pp. 179208) Cambrige, UK: Cambrige University Press.
Cobbina, J. E., Morash, M., Kashy, D. A., and Smith, S. W. (2014). Race, neighborhood danger,
and coping strategies among female probationers and parolees.Race and Justice, 1-26.
Cochran, J. C. (2013). Prison experiences, social ties, and inmate behavior: Examining visitation
and its effects on incarceration and reentry outcomes. Dissertation submitted to the
College of Criminology and Criminal Justice, The Florida State University.
Cohen, J. A. (1960). A coefficient of agreement for nominal scales. Education and Psychological
Measurement, 20, 37-46.
Edin, K., and Kefalas, M. (2005). Promises I can keep:Why poor women put motherhood before
marriage. Berkeley, CA: University of California Press.
Elder, G. H., Jr. (1985). Perspectives on the life course. In G. H. Elder,, Jr. (Eds.), Life course
102

dynamics: Trajectories and transitions, (pp. 23-49). Ithaca, NY: Cornell University
Press.
Elliott, D. S., Huizinga, D., and Menard, S. (1989). Multiple problem youth: delinquency,
substance abuse, and mental health problems. New York: Spinger-Verlag.
Fagan, J. 1994. "Women and drugs revisited: Female participation in the cocaine economy.
The Journal of Drug Issues, 2(2),179-225.
Fleisher, M. S., and Krienert, J. L. (2004). Life-course events, social networks, and the
emergence of violence among female gang member. Journal of Community Psychology,
32, 607-622.
Garland, D. (2001). The culture of control and social order in contemporary society. Oxford:
Oxford University Press.
Glaze, L. E., &Maruschak, L. M. (2008).Parents in prison and their minor children.
Bureau of Justice Statistics Special Report (NCJ 222984). Washington, DC: U.S.
Department of Justice, Office of Justice Programs.
Glueck, S., and Glueck, E. (1950). Unraveling Juvenile Delinquency. New York, NY:
Commonwealth Fund.
Giordano, P. C. (2010). Legacies of crime: A follow-up of the children of highly delinquent girls
and boys. New York, NY: Cambridge University Press.
Giordano, P. C., Cernkovich, S. A., and Rudolph, J. L. (2002). Gender, crime, and desistance:
Toward a theory of cognitive transformation. American Journal of Sociology, 107(4),
990-1064.
Giordano, P. C., Seffrin, P. M., Manning, W. D., and Longmore, M. A. (2011). Parenthood and
crime: The role of wantedness, relationships with partners, and ses. Journal of Criminal
Justice, 39, 405-416.
Gottfredson, M. R., and Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford
University Press.
Graham, J., and Bowling, B. (1996). Young people and crime. London: Home Office Research
Study.
Greene, S., Haney, C., & Hurtado, A. (2000). Cycles of pain: Risk factors in the lives of
imprisoned mothers and their children. Prison Journal, 80, 3-23.
Greenfield, L. A., and Snell, T. L. (1999). Women offenders. Bureau of Justice Statistics
Washington, DC. U.S. Department of Justice, Office of Justice Programs.
Greenwood, P. W. (1982). Selective incapacitation. Report to the National Institute of Justice.
Santa Monica, CA: Rand Corporation.
Guetzkow, H. (1950). Unitizing and categorizing problems in coding qualitative data. Journal of
Clinical Psychology, 6, 47-58.
103

Hagan, J. and Dinovitzer, R. (1999).Collateral consequences of imprisonment for children,
families and prisoners.Crime and Justice 26: 121-162.
Hagger, M.S., Anderson, M., Kyriakaki, M. and Darkings, S. (2007). Aspects of identity and
their influence on intentional behavior: Comparing effects for three health behaviors.
Personality and Individual Differences 42, 355-367.
Hardyman, P. L., & Van Voorhis, P. (2004).Developing gender-specific classification systems
for women offenders. Washington, DC: USDOJ, National Institute of Corrections.
Hatterly, A., and Smith, E. (2010). Prisoner reentry and social capital: The long road to
reintegration. Lanham, MD: Lexington Books.
Heimer, K., and De Coster, S. (1999). The Gendering of Violent Delinquency, Criminology
37(2), 277 - 318.
Hunt, G. P., MacKenzie K., and Joe-Laidler, K. (2005). Alcohol and masculinity: The case of
ethnic youth gangs. In T.M., Wilson, (Eds.) Drinking cultures: Alcohol and identity (pp.
225–254). Oxford, UK: Berg.
Joplin, L., Bogue, B., Campbell, N., Carey, M., Clawson, E., Faust, D., Florio, K., Wasson, B.,
and Woodward, W. (2005). In What works and why: Effective approaches to reentry. 109
145.American Correctional Association. Lanham, Maryland: Library of Congress
Cataloging-in-Publication Data.
Kreager, D. A., Matstuesda, R. L., and Erosheva, E. A. (2010).Motherhood and criminal
desistance in disadvantaged neighborhoods.Criminology, 48(1), 221-258.
Krisberg, B.A., & Temin, C.E. (2001). The plight of children whose parents are in prison. NCCD
Focus. Retrieved from http://www.nccd-crc.org/nccd/n_pubs_main.html
Kubrin, C.E., Stucky, T.D., Krohn, M.D. (2009). Researching theories of crime and deviance.
New York, NY: Oxford University Press. 198-217.
La Vigne, N. G. Naser, L. E. and Castro, J. L. (2005). Examining the effect of incarceration and
in-prison family contact on prisoners’ family relationships. Journal of Contemporary
Criminal Justice, 21(4), 314-335.
Laub, J. H. and Sampson, R. J. (2003).Shared beginnings, divergent lives: Delinquent boys to
age 70. Cambridge, MA: Harvard University Press.
Leverentz, A. (2006). The love of a good man? Romantic relationships as a source of support or
hindrance for female ex-offenders. Journal of Research in Crime and Delinquency, 43,
459-488.
MacKenzie, D. L. (2013). First do no harm: A look at correctional policies and programming
today. Journal of Experimental Criminology, 9, 1-17.
Maurua, S. (2001). Making good: How ex-convicts reform and rebuild their lives. Washington,
DC: American Psychological Association Books.
104

Mazza, C. (2002). And then the world fell apart: The children of incarcerated fathers.
Families in Society: The Journal of Contemporary Human Services, 83 (5/6), 521- 529.
Messina, N., Burdon, W., Hagopian, G., and Prendergast, M. (2006). Predictors of prison
TC treatment outcomes: A comparison of men and women participants.
The American Journal of Drug and Alcohol Abuse, 32(1), 7‐28.
Michalsen, V. (2011). Mothering as a life course transition: Do women go straight for their
children? Journal of Offender Rehabilitation, 50, 349-366.
Miller, J. M., Schrenk, C. J., & Tewksbury, R. (2011).Criminological theory: A brief
introduction. (3 eds.). Pearson Education, Inc.
Minow, M. (1990).Adjudicating differences: Conflicts among feminist lawyers. In Hirsch, M.
and Keller, E. F. (ed.) Conflicts in Feminism, (149-164) New York, NY: Routeledge.
Moloney, K.P., and Moller, L. F. (2009). Good practice for mental health programming in
prison: Reframing the parameters. Public Health, 123(6), 431-433.
Moffitt, T. (1993). Adolescent-limited and life-course-persistent antisocial behavior: A
developmental taxonomy. Psychological Review, 100, 674-701.
Morash, M. (2010). Women on probation and parole: a feminist critique of community programs
& services. Boston: Hanover [N.H.]: Northeastern University Press; University Press of
New England.
Muhammad, B.M. (2012). Exploring the silence among children of prisoners: A descriptive
study. Dissertation submitted to the Graduate School-Newark Rutgers, The State
University of New Jersey.
Mumola, C.J. (2000). Incarcerated parents and their children (NCJ 182335).
Washington, DC: Bureau of Justice Statistics.
Murray, J., & Farrington, D. P. (2005). Parental imprisonment: Effects on boys’ antisocial
behaviour and delinquency through the life-course. Journal of Child Psychology and
Psychiatry, 46(12), 1269–1278.
Murray, J., & Farrington, D.P. (2008). Parental imprisonment: Long-lasting effects on boys’
internalizing problems through the life course. Development and Psychopathology, 20(1),
273-290.
Osgood, D. W., Johnston, L.D., O’Malley, P. M., and Bachman, J. G. (1988). The Generality of
Deviance in Late Adolescence and Early Adulthood. American Sociological Review,
53(1): 81–93.
Paltrow, L. M. (1990). When becoming pregnant is a crime. Criminal Justice Ethics, 9(1), 41-47.
Petersilia, J. (2003). When prisoners come home: Parole and prisoner reentry. New York, NY:
Oxford University Press, USA.
105

Petersilia, J. (2005). Meeting the challenges on prisoner reentry. In What works and why:
Effective approaches to reentry. 175-192. American Correctional Association. Lanham,
Maryland: Library of Congress Cataloging-in-Publication Data.
Poehlmann, J. (2005b). Incarcerated mothers’ contact with children, perceived family
relationships, and depressive symptoms. Journal of Family Psychology, 19(3),
350-357.
Porporino, F. J. (2005). Revisiting responsivity: Organizational change to embrace evidencebased principles and practices. In What works and why: Effective approaches to reentry.
193-230. American Correctional Association. Lanham, Maryland: Library of Congress
Cataloging-in-Publication Data.
Richie, B. (2001). Challenges incarcerated women face as they return to their communities:
Findings from life history interviews. Crime and Delinquency, 47, 368-389.
Riehman K. S., Hser, Y., and Zeller, M. J. (2000). Gender differences in how intimate partners
influence drug treatment motivation. Drug Issues, 30(4), 823-838.
Salisbury, E., and Van Voorhis, P. (2009). Gendered pathways: A quantitative investigation of
women probationers’ paths to incarceration. Criminal Justice and Behavior, 36(6), 541566.doi:10.1177/0093854809334076
Schroeder, R. D., Bulanda, R. E., Giordano, P. C. and Cernkovich, S. A. (2011). Parenting and
adult criminality: An examination of direct and indirect effects by race. Journal of
Adolescent Research, 25(1), 64-98.
Sherer, M., Maddux, J.E., Mercadante, B., Prentice-Dunn, S., Jacobs, B., and Rogers, R.W.
(1982). The self-efficacy scale: Construction and validation. Psychological Reports, 51,
663-671.
Stringer, S. M. (2006).Breaking parole: A critical analysis of the New York State Division
of Parole's caseload management standards, New York, NY: Manhattan Borough
President.
Thompson, M., and Petrovic, M. (2009). Gendered transitions: Within-person changes in
employment, family, and illicit drug use. Journal of Research in Crime and Delinquency,
46, 377-408.
Tibbetts, S. G. (2002). Gender, emotions, and criminal offending. Paper presented at the annual
meeting of the Academy of Crime Justice Sciences, Anaheim, CA. Available from S. G.
Tibbetts at California State University, San Bernadino.
Uggen, C., Wakefield, S., and Western, B. (2005).Work and family perspectives on reentry.In
Travis, J., and Visher, C. Prisoner Reentry and Crime in America. New York, NY:
Cambridge University Press.
Van Voorhis, P., Baumann, A., and Brushett, R. (2013). Revalidation of the Women’s Risk Needs
Assessment: Probation Results. (Final Report) (pp. 1–177). University of Cincinnati.
106

Varriale, J. A. (2008). Female gang members and desistance: Pregnancy as a possible
exit strategy? Journal of Gang Research, 15, 35–64.
Visher, C. A., and Travis, J. (2003). Transitions from prison to the community: Understanding
individual pathways. Annual Review of Sociology, 29, 89-113.
Warr, M. (1998). Life course transitions and desistance from crime. Criminology, 36(2), 183216.
White, G. D. (2012). Gender-Responsive Programs in U.S. Prisons: Implications for Change.
Social Work in Public Health, 27(3), 283-300.
Wright, J. P., Tibbetts, S. G., and Daigle, L. E. (2008).Criminals in the making: Criminality
across the life course. Thousand Oaks, CA: Sage Publications, Inc.

107