INFLUENCES ON DRUG AND ALCOHOL-INVOLVED FEMALE OFFENDERS
ENGAGEMENT IN DESIRED COMMUNICATION WITH THEIR PROBATION/PAROLE
OFFICERS
By
Jennifer Cornacchione

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
For the degree of
Communication – Doctor of Philosophy
2014

ABSTRACT
INFLUENCES ON DRUG AND ALCOHOL-INVOLVED FEMALE OFFENDERS
ENGAGEMENT IN DESIRED COMMUNICATION WITH THEIR PROBATION/PAROLE
OFFICERS
By
Jennifer Cornacchione
This study investigates factors that affect female probationers' and parolees' decision to
engage or not engage in desired communication with their probation/parole officers (PO) about
difficult issues they face. The influence of perceived norms, conversational goals, and various
secondary variable considerations (such as criminogenic needs, perceptions of PO characteristics
and influences, and social network qualities) on communication behavior are examined.
Interviews were conducted with 402 women on probation and parole across the state of
Michigan, and both quantitative and qualitative data were collected. Of these 402 women, 127
stated that there was a time when they wanted to talk to their PO about a difficult need or issue
they were facing. Issues they wanted to talk about included housing, relationship issues, and
illegal activity. Results indicated that social norms did not predict whether or not communication
occurred, but that personal resource goals were a significant predictor. Women who were
concerned about threats to their freedom and resources were less likely to have talked to their PO
about the issue. Additionally, women who reported that they felt anxiety and psychological
reactance during and after their meetings with their PO were also less likely to engage in the
conversation. These findings illuminate possible reasons that impact whether or not individuals
engage in difficult conversations.

Copyright by
JENNIFER CORNACCHIONE
2014

This dissertation is dedicated to the memory of my mom and dad. You always supported,
encouraged, and praised me, and those words have stuck with me every day, pushing me through
many days of self-doubt. I couldn't have done this without you, knowing you've always been my
biggest fans.

iv

ACKNOWLEDGEMENTS

I would like to thank Dr. Sandi Smith for your support, guidance, and friendship over the
last several years. Because of your encouragement and expertise, I have achieved many goals
and had incredible experiences that I would not have had otherwise. I am grateful for the
professional and personal relationship we have built.
I would also like to thank Drs. Merry Morash, Amanda Holmstrom, and Kami Silk for
serving on my committee, taking the time to provide me with insightful feedback, and always
challenging me to think more critically. Additionally, I would like to thank my former mentors,
Drs. Joanne Gilbert and Micheal Vickery, who initiated my interested in the field of
communication and encouraged me to attend graduate school.
There are several individuals and agencies affiliated with this project I would like to
acknowledge. Funding for this project is provided by the National Science Foundation under
Grant No. 1126162 and by a Strategic Partnership Grant from the Michigan State University
Foundation. The opinions, findings, and conclusions or recommendations expressed in this paper
are those of the author and do not necessarily reflect those of the funding agencies. Thanks to
Sarah Sheff and Lauren Zanger for coding, Beth Adams for transcription, and Mim Northcutt
Bohmert for project management and coffee breaks. Lastly, thank you to the women who
participated.
I would also like to thank my friends during my time at MSU who helped keep me sane,
provided support, and reminded me of life outside of school. I've enjoyed every moment of
watching TV and eating horrible food with you all.

v

Finally, I would like to thank my family. Thank you, Jon, Tony, Nicole, and Jill for
always being proud of your little sister. Thank you to Beth and Gary for being constant sources
of support. Thank you to Gus, Oliver, and Farkas for being such great companions, especially on
stressful days. And thank you to Dennis for putting up with more than anyone, and always
finding a way to make me laugh. Most of all, I'd like to thank my mom and dad for raising
someone who never quits.

vi

TABLE OF CONTENTS

LIST OF TABLES

ix

INTRODUCTION

1

CHAPTER 1
LITERATURE REVIEW
Women on Probation & Parole
Voice & Silence
Conversational Goals
Social Norms & Communication
Context of Study & Additional Factors Influencing Desired
Communication with POs
Supervision status
Criminogenic needs assessment
Social networks & social support
Perceived PO characteristics and interactions

4
4
8
9
13
17
18
19
20
23

CHAPTER 2
METHOD
Participants
Measures
Goals
Social Norms
Open-Ended Data Analysis
Motivations and constraints
Topics
Secondary Data Analysis
Supervision status
Criminogenic needs assessment
Social networks & social support
Perceived PO communication style
Elicited emotions & reactance

26
26
28
28
29
30
31
32
32
32
33
33
35
35

CHAPTER 3
QUANTITATIVE DATA ANALYSIS OVERVIEW
Sample Selection Bias

37
37

CHAPTER 4
RESULTS
Sample Description
Tests of Hypotheses & Research Questions

39
39
40

vii

Goals and constraints
Social norms
Offender characteristics
Social networks and social support
Perceptions of PO characteristics & interactions
Post Hoc Analyses

40
43
43
44
44
46

CHAPTER 5
DISCUSSION
Overview of Findings & Practical Implications
Conversational goals
Social norms
Offender characteristics
Social networks & social support
Perceptions of PO characteristics & interactions
Limitations & Directions for Future Research

49
49
49
51
54
55
56
59

CHAPTER 6
CONCLUSION

62

APPENDICES
Appendix A Interview Items for Norms and Goals
Appendix B Secondary Data Analysis Variable Items
Appendix C Criminogenic Needs Subscales
Appendix D Tables
Appendix E Footnotes

63
64
69
74
75
88

REFERENCES

90

viii

LIST OF TABLES

Table 1

Psychometric Properties of Continuous Measures

75

Table 2

Correlations Among Variables

76

Table 3

Open-Ended Categories and Conceptual Definitions

77

Table 4

Logistic Regression Predicting Communication

80

Table 5

Topics and Engagement in Communication

81

Table 6

Logistic Regression of Goals Predicting Communication

82

Table 7

Logistic Regression of Social Norms Predicting Communication

83

Table 8

Logistic Regression of Risk Score and Parole Status Predicting
Communication

84

Table 9

Logistic Regression of Social Network and Support Qualities
Predicting Communication

85

Table 10

Logistic Regression of Perceptions of PO on Predicting
Communication

86

Table 11

Logistic Regression for Impact of Perceptions of PO Relationship
Style on Communication

87

ix

INTRODUCTION
Over one million women in the United States are on probation or parole (Bonczar &
Glaze, 2009). Women offenders are unique from male offenders because of their past
experiences, which are often characterized by physical and sexual abuse both in childhood and as
adults. As a result, many of their offenses are often substance-centered because they turn to
drugs and alcohol to cope (Morash, 2010). Previous research has indicated female offenders
report their probation/parole officers (PO) as a main source of support and an important member
of their social network (Morash, 2010). However, almost no research has been reported on the
interaction between female offenders and their POs. This study examines the factors that impact
whether or not female offenders who want to talk to their POs about a difficult issue or problem
they are facing actually engage in communication. Some of these difficult issues or problems
could be criminogenic needs, which are risks and predictors of crime, and include things such as
the stress that comes from parenting, being abused by a partner, or not having access to safe
housing (Van Voorhis, Wright, Salisbury, & Bauman, 2010). Other topics of conversation could
include wanting to go to substance treatment and moving into a new home. Goals and social
norms are used to examine this phenomenon.
The multiple goals perspective (Caughlin, 2010) may help to explain whether or not
women on probation or parole engage in communication with their POs about difficult issues or
needs they are facing about which they want talk to their POs. Goals, defined as desired end
states (Wilson, 2007), often have been distinguished as primary or secondary (i.e., constraints)
(Dillard, Segrin, & Hardin, 1989). Primary goals define the interaction, whereas constraints are
generally derived from the primary motivations of the interaction, and can include things such as
maintaining positive relationships and not offending the other interactant. Women on probation

1

or parole may face multiple, conflicting goals when thinking about engaging in communication
with their PO about any of their needs or issues that they may wish to discuss. Palomares noted
(in press) that individuals may have certain goals, but no behavioral indicators may ever be
exhibited. For example, a woman may have a goal to seek help, but the presence of a behavioral
indicator (talk or not talk) can vary depending on a variety of reasons, including constraints.
Wanting to conform to perceived social norms may facilitate or constrain the offender's
actual engagement in communication. Social norms are defined as beliefs about the expected or
ideal behavior within a situation (Sherif, Harvey, White, Hood, & Sherif, 1961). The three main
types of norms often used in research are descriptive (what is done), injunctive (what ought to be
done), and subjective (what one thinks important others want one to do) (e.g., Anderson,
Cornacchione, & Maloney, 2013; Kam, Matsunaga, Hecht, & Ndiaye, 2009; Park & Smith,
2007). Although there is a great deal of social norms research, social norms research about
communication is limited. Specifically, most research examines social norms about behavior,
such as recycling or using condoms, but not about communication as the behavior. In fact, much
of the research examining communication within the social norms framework has studied it as a
vehicle for norms transmission. This paper examines normative influences on communication
(i.e., whether or not a female offender engages in communication with her PO) between female
offenders and their PO about difficult issues or topics, including criminogenic needs.
This paper will first provide an overview of the need to study communication between
POs and their female clients, as well as factors that make women on probation and parole a
unique population to study. The multiple goals approach and social norms will be explained, and
an overview of their predicted relationships with engaging in communication will be provided.
Additional factors that may also predict whether a woman communicates to her PO about her

2

needs will also be discussed. Research questions and hypotheses will be advanced, the study
method and analyses will be outlined, and the findings will be discussed. Implications and
directions for future research will also be provided.

3

CHAPTER 1: LITERATURE REVIEW
Women on Probation and Parole
Women on probation and parole face unique needs that males often do not. Some of these
needs or issues include providing for children and dealing with issues of custody. Additionally,
in comparison to male offenders and women in the general population, women offenders are
often the victims of domestic violence both as children and adults (Morash, 2010). These
experiences often lead them to cope by using controlled substances. Substance use and abuse is
one of the most common offenses committed by women (Morash, 2010; Salisbury & Van
Voorhis, 2009). In one study, 91% of women re-entered into the community after prison release
reported using crack, cocaine, or heroin six months prior to their most recent arrest
(Freudenberg, Daniels, Crum, Perkins, & Richie, 2005).
When women offenders have been convicted of a felony or misdemeanor crime, they are
assigned to a term of supervision called parole (for felony-level offenders who have completed
their prison term) or probation (for felony- or misdemeanor-level offenders supervised in the
community instead of being incarcerated). Although probation and parole are two distinct types
of community supervision, the rules and conditions that are mandated by the court for both are
quite similar, such as being required to attend substance abuse treatment, go to educational and
employment services, obtain housing, attend medical appointments, and obey all supervising
agent‟s instructions. The offenders must also regularly report to an assigned probation or parole
officer (USBJS, n.d.). Failure to comply with these conditions can result in being placed in
prison or jail.
One of the main goals of community supervision is to prevent recidivism, which is
defined as a relapse in criminal behavior. Recidivism rates among female parolees are especially

4

high. In a sample of a female release cohort in 15 states, approximately 60% of paroled women
were rearrested. Additionally, 30% returned to prison within three years (Deschenes, Owen, &
Crow, 2007). Several factors have been identified that can affect female offenders‟ future
recidivism rates, and they are called criminogenic needs. Criminogenic needs are factors that
have been found to predict or cause crime. These factors include parental stress; self-efficacy;
trauma, victimization, and abuse; mental health; self-esteem; and intimate relationships, among
others (Van Voorhis, Salisbury, Wright, & Bauman, 2008). Having antisocial attitudes, including
criminal thinking, and associating with antisocial friends are some of the strongest predictors of
future criminal behavior (Andrews & Bonta, 2003). Other needs and problems faced by female
offenders include educational challenges, financial and employment problems, housing safety,
family conflict (i.e., with family of origin), stresses associated with parenting, substance abuse
and treatment, and mental health problems, including anxiety and depression (Van Voorhis et al.,
2008). These factors may be important topics of conversation between women and their POs.
Talking to their POs about these things may be important for avoiding future arrests and
recidivism, as well as having successful outcomes, such as finding a job or getting an education.
Because of the significant role these challenges can place on women who are living in the
community and are currently serving probation or parole sentences, it is important to examine
whether the women initiate conversations about these, or other, topics (such as relapsing) with
their PO.
Female offenders' relationships with their POs can be especially important because they
often receive little family support. In fact, women substance users receive less tangible and
emotional support than non-substance users (Mallik & Visher, 2008). Other factors also
complicate their social networks. For example, many family members themselves have broken

5

the law, or have been the ones to abuse the women (Morash, 2010). Furthermore, uneducated
and low-income women often have smaller social networks upon which to rely (Reisig,
Holtfreter, & Morash, 2002). The offenders' reliance on their POs has been suggested in several
studies. For example, a recent study found that a greater percentage of women on parole received
memorable messages from their PO compared to women on probation (Cornacchione et al.,
2013). This may be because women on parole have spent time in prison, and, thus, their social
networks have been severed, making communication with their PO even more important and
salient. Other research (e.g., Cobbina, 2010) indicates that the nature of the relationship between
female parolees and their supervising officers impacts reentry. One study, for example, found
that female offenders who were more submissive, isolated, and friendly were more likely to
develop strong alliances with a therapist compared to women who were hostile and aggressive
(Ross, Polaschek, & Ward, 2008). These traits could make women feel more comfortable talking
with their PO about issues they are facing compared to women who are more hostile and
aggressive. Women who are isolated may tend to develop closer relationships with their PO.
Female offenders tend to have limited access to information and social support (Skeem,
Eno Louden, Manchak, Vidal, & Haddad, 2009). Because of their diminished access to support
and a minimal social network, the relationship with their POs could play a significant role in
their outcomes, as the PO might be their only source of information and support. Research
conducted by Skeem and colleagues (2009) found that female probationers with mental illness
and substance abuse problems tended to have better probation outcomes (e.g., fewer violations)
when they had a satisfying relationship with their POs, which also included participatory
decision-making. Skeem, Eno Louden, Polaschek, and Camp (2007) found that female
probationers who perceived their POs to be tough had a higher number of violations. Taxman

6

and Ainsworth (2009) noted similar findings for offenders: perceptions of fairness and caring
lead to more positive outcomes. Although research has demonstrated the positive outcomes
related to successful supervision, there is minimal literature on studying the communicative
relationship between female offenders and their POs to determine what successful supervision
entails. Communicating with their POs may be the one of the keys to successful supervision and
this communication might be influenced by the women‟s conversational goals and constraints, as
well as the perceived norms that exist within their relationships or networks. Women on
probation and parole may only be able to obtain the necessary resources they need if they
communicate with their POs about these issues.
However, the relationship between offenders and their supervising officers can be quite
complicated. POs serve dual-roles in relation to their clients. Although research indicates
numerous positive outcomes for offenders when POs are caring and supportive, POs must also
enforce the law and protect public safety (Kennealy, Skeem, Manchak, & Eno Louden, 2012;
Skeem et al., 2007). Additionally, POs have a lot of discretion in their supervision. For example,
whereas some POs might throw a woman into jail if she has started using drugs again, other POs
may instead refer her to rehabilitation services. Thus, engaging in communication with their POs
about difficult issues or needs could have negative consequences for the women, even if the PO
perceives the outcome to be positive (e.g., the woman does not want to go to jail, but being put in
jail may save her life and protect the public). Furthermore, some POs might take on more of the
counselor role and be open to talking about various issues with their clients, while others might
be more interested in just asking standard questions related to the women's supervision
requirements. There are no standardized protocols for the meetings between POs and their

7

clients. Although engaging in communication with their PO can lead to better outcomes, there
may be various goals and constraints driving whether the conversation even occurs.
Voice & Silence
Traditionally, offenders have not had voice. Community supervision has a history of
being dominantly characterized by toughness. However, recent research shows that certain POclient relationships (those characterized by trust and caring) can lead to improved probation and
parole outcomes (Skeem et al., 2003). Thus, as the shift in the nature of supervision occurs,
including the instantiation of gender-specific caseloads, it might be the case that women are
learning to have a voice, are engaging in more decision-making, and, thus, are using their voice
to help improve their outcomes.
Previous research in the context of organizations (e.g., Rusbult et al., 1988) and romantic
relationships (e.g., Rusbult, Zembrodt, & Gunn, 1982) has examined the concept of voice,
particularly in reference to dissatisfaction. More recent research has identified voice and silence
as two separate continuous constructs (Van Dyne, Ang, & Botero, 2003). Voice is defined as
people expressing their ideas, opinions, and information. Silence, on the other hand, implies
withholding information. After a thorough review of the literature, Taxman and Ainsworth
(2009) found that more positive outcomes are likely to occur when offenders perceive they have
a voice in decisions. Having a voice, or contributing to decision-making, has also been found to
increase perceptions of procedural justice (e.g., Korsgaard & Roberson, 1995; Lind, Kanfer, &
Earley, 1990). If women are able to talk to their PO and work together to find a solution, then
they may have better supervision outcomes, including a decreased number of violations and
decreased recidivism. However, factors that may impact women's tendency to express voice or
silence about particular topics could be the various goals and constraints that exist.

8

Conversational Goals
Models of message production may facilitate understanding of whether women who want
to talk to their POs about difficult issues actually engage in the conversation. Goals have been
defined as, “future end states of affairs that an individual desires to attain or maintain” (Wilson,
2007, p. 74), and have been acknowledged as important components of communication (Berger,
2005). Within relationships, individuals‟ goals and constraints may influence whether topics are
discussed, why they are discussed, and the nature in which they are discussed. This study
examines how goals and constraints impact whether the conversation actually occurs.
In 2010, Caughlin traced the development of the multiple goals perspective in
communication research. Three assumptions exist in the goals research tradition: communication
is strategic and purposeful, individuals most often pursue multiple goals at the same time, and
communication goals often conflict. Individuals typically communicate with purpose, such as to
gain compliance, provide support, enhance positive face, persuade another to do something, and
to give advice. Because communication is purposeful, it is often the case that individuals pursue
multiple goals simultaneously. Goals can be distinguished as primary or secondary (Dillard et
al., 1989). Secondary goals are actually constraints (Palomares et al., 2013). Whereas goals are
desired end states individuals wish to attain, constraints are ongoing concerns, considerations,
and behavioral expectations. Specifically, "constraints...influence the behaviors enacted in
pursuit of a goal" (Palomares, in press, p. 84; 2013). Goals refer to the main reason or drive for
engaging in communication, and often involve influence goals to persuade the other interactant
(Dillard, 1989). The primary goal, “brackets the situation. It helps segment the flow of behavior
into a meaningful unit; it says what the interaction is about” (Dillard et al., 1989, p. 21).
Constraints, on the other hand, refer to other considerations about the interaction that might

9

constrain or alter the conversation, such as ongoing concerns and considerations. Specifically,
they “derive from more general motivations that are recurrent in a person‟s life” (Dillard et al.,
1989, p. 20). The desire to avoid conflict and refrain from offending the other person are two
concrete examples of constraints because they serve as concerns about having the conversation
or how the conversation is ultimately enacted (Dillard et al., 1989). In general, constraints shape
and refine message production, as well as denote the range of available behavioral options
(Schrader & Dillard, 1998).
A number of goal and constraint types have been discussed in the literature, and Dillard‟s
(1990) Goals-Planning-Action model specifies five specific types of constraints. These include
identity, interaction, relational resource, personal resource, and arousal management (Dillard et
al., 1989; Schrader & Dillard, 1998). Identity refers to concerns about behaving in ways
consistent with personally held beliefs and values. Dillard and colleagues (1989) define this as
“internal standards of behavior” that “derive from one‟s moral standards…and personal
preferences concerning one‟s own conduct” (p. 20). For example, a probationer may want to
maintain her identity as being strong and tough, which might inhibit her from seeking help.
Interaction refers to the desire to engage in social appropriateness and impression management.
They also represent one‟s desire to avoid threatening face of the interaction partner (Dillard et
al., 1989). For example, an offender may simply ask for help rather than demanding that her PO
do something to fix the situation. Relational resource refers to concerns about preserving one‟s
relationship with the other interactant because of the rewards and gratifications associated with
that relationship, such as an offender not wanting to tell her PO about engaging in illegal activity
because she does not want to damage the positive relationship they currently have. Personal
resource refers to the desire to avoid consequences for one‟s own resources, such as going back

10

to prison. Arousal management refers to concerns about avoiding feelings of embarrassment or
nervousness, and generally staying within preferred arousal levels. For example, a female
parolee may experience a lot of anxiety regarding her recent relapse.
Dilemmas often occur during interactions when deciding which goal to pursue and how
to pursue it (i.e., constraints), and these dilemmas can affect both communication behaviors and
relationship outcomes. For example, a woman on probation who has relapsed and begun using
drugs again may want to discuss this relapse with her PO to receive the treatment she needs
(goal). Ultimately, however, she may decide to avoid talking about this because she fears her PO
would think negatively of her or initiate action to incarcerate her (constraint). An inherent
conflict exists within the PO-offender relationship. Although a woman may want to be open and
honest with her PO, being open and honest could result in legal action, thus making facilitating
open and honest communication problematic. Women under community supervision already face
many barriers and threats to freedom (e.g., difficulty finding a job because of a felony conviction
and a risk for (re)incarceration); thus, it is anticipated that personal resource constraints will be
more influential than other considerations in the women's decision to engage in communication
with their POs. Therefore, it is predicted that:
H1: As ratings of personal resource increase, women will be less likely to engage in
communication with their POs about a difficult need/issue they are facing.
Using the multiple goals approach, scholars have sought to identify people‟s motivations
for engaging or not engaging in communication, along with the factors that serve as both the
independent and dependent variables, in a variety of contexts and situations. Donovan-Kicken
and Caughlin (2010) examined the relationship between topic avoidance and relationship
satisfaction for women recently diagnosed with cancer. The authors specifically wanted to

11

examine the influence that motivations of avoidance had on the outcomes. They found that the
motivations for avoidance moderated the relationship between topic avoidance and relationship
satisfaction. When social constraints were high, the relationship between satisfaction and
avoidance was negative. Conversely, when low social constraints were present, topic avoidance
predicted greater relationship satisfaction. Cornacchione (2013) found that when concerns for
maintaining the relationship were high, individuals were less likely to engage in communication
to persuade a family member about a health issue. Scott (2010) found that in end-of-life
discussions, competing goals and constraints often caused dilemmas. While some individuals
wanted to avoid communicating about end-of-life during this time, it had implications for other
goals, such as relational goals. For instance, relational distancing frequently occurred because
having end-of-life conversations could potentially hurt the family member. In the context of
topic avoidance in families after a lung cancer diagnosis, it was found that multiple reasons
explained the avoidance of communication about the health issue, including maintaining hope
and optimism, along with protection of the family members (Caughlin, Mikucki-Enyart,
Middleton, Stone, & Brown, 2011). Different reasons for avoiding conversations were also found
to have different outcomes. For example, avoiding was dissatisfying when it was done out of
denial (Caughlin et al., 2011). The current study seeks to understand female offenders' reasons
for engaging and not engaging in communication with their POs even though they want to have
the conversation. Thus, the following research questions ask:
RQ1: What are the reported reasons (i.e., goals) for engaging in communication with
POs about the difficult issue?
RQ2: What are the reported reasons (i.e., constraints) for not engaging in the desired
conversation with POs about the difficult issue?

12

Social Norms & Communication
Conforming to perceived social norms may serve as a goal or a constraint when
individuals decide to engage or not engage in communication. Social norms are defined as “a
system of anchors around which all social relationships are organized” (Sherif, 1936, p. 126) that
“regulate members‟ behavior in terms of the expected or the ideal behavior” (Sherif, Harvey,
White, Hood, & Sherif, 1961, p. 9). Norms denote a range of acceptable behavior. Research has
found that people define themselves in the context of their relationships. Sherif (1936) stated that
“one feels oneself in definite relationships with others, and one acquires definite expectations as
to their responses, as well as definite responsibilities toward them” (p. 182). Norms exist within
relationships and groups, and these norms guide attitudes and behaviors.
Behaviors are driven, in part, by a desire to avoid social sanctions. Behaviors can also be
driven by a desire to receive social approval. Asch (1956) stated “the compliant
person…translates social opposition into a reflection of his personal worth” (p. 52). Individuals
who strongly identify with their group are more likely to conform to their group norms (Kwon &
Lease, 2009), and acceptance into groups is a function of engaging in socially accepted behavior
(Shulman & Levine, 2012). Individuals are motivated to conform to the normative behavior to
remain a member of their referent groups without receiving any social sanctions and to achieve
approval. The relationship between the individual and the referent group is directly related to
conformity, such that distal peer norms tend to be less influential than proximal peer norms,
(e.g., Cho, 2006; Yun & Silk, 2011). For example, a female offender may refer to other female
offenders about what behavior should or should not occur, as opposed to referring to other
people in the community. Individuals rely on others for a variety of resources, which facilitates
conformity. Lapinski and Rimal (2005) stated "because people are dependent on others to meet

13

their needs, they are concerned about others' evaluation of their behaviors" (p. 131). These
potential dependencies may have significant implications for women when they are trying to
decide whether or not to communicate with their POs to address needs or difficult issues. It is
possible that the problems they are facing could result in further supervision or punishment. For
example, a woman may be unable to abstain from illicit drug and alcohol use, and talking to her
PO about relapsing could cause her to get into more legal trouble.
Three types of norms have been identified in the literature: subjective, injunctive, and
descriptive, and exist at both the societal and personal level (Cialdini, Kallgren, & Reno 1991;
Park & Smith, 2007; Perkins & Berkowitz, 1986). Subjective norms, as defined in the theory of
planned behavior, are perceptions about what important others expect one to do; thus, they fall at
the personal level of norms (Azjen & Fishbein, 1980). Individuals are motivated to comply with
these important referents. Injunctive and descriptive norms refer to beliefs about what ought to
be done and beliefs about what is actually done (i.e., prevalence of a behavior), respectively
(Cialdini, Kallgren, & Reno, 1990; Lapinski & Rimal, 2005). At the personal level, descriptive
norms refer to an individual‟s belief about the prevalence of a behavior among people whom the
individual considers important or whose opinion is valued by the individual. Personal-level
injunctive norms refer to an individual‟s belief about the approval for that behavior among
people who the individual considers important, or whose opinion they value (Park & Smith,
2007). Because this is a study of relational communication, it is focused on personal-level
perceived norms as opposed to societal level norms. The perceived norms, or cognitions about
norms, examined in this study are those that have emerged through the women's interactions with
her referents.

14

Extensive theoretical work has addressed the role of social norms in predicting individual
behaviors. The theory of planned behavior (TPB; Ajzen, 1991), focus theory (Cialdini, Reno, &
Kallgren, 1990; Reno, Cialdini, & Kallgren, 1993), and the theory of normative social behavior
(Lapinski & Rimal, 2005; Rimal & Real, 2005) all address the influence of social norms on
behavior. Subjective norms (i.e., perceptions about what important others expect one to do) in
the TPB have been found to be a strong predictor of intentions to engage in health behaviors,
such as to use condoms (Albarracin, Johnson, Fishbein, & Muellerliele, 2001). Subjective norms
have also been found to be a significant predictors of communication initiation in a variety of
contexts, including intent to engage in family discussion about organ donation (Park & Smith,
2007) and intent to tell family members about genetic test results (Barsevick et al., 2008). Park
and Smith found that subjective norms moderated the effect between both perceived behavioral
control for, and attitudes about, engaging in family discussion about organ donation. Those with
stronger perceived subjective norms (perceptions that important others wanted them to engage in
the conversation) were more likely to intend to engage in family discussion about organ
donation, but descriptive norms (perceptions of how much others did engage in communication)
were positively related to behavioral intent when subjective norms were weak. However, one
study found no relationship between subjective norms and behavioral intent (e.g., Bran & Sutton,
2009). Other research (e.g., Rivis & Sheeran, 2003) found descriptive norms to be a significant
predictor of behavioral intent. Extensive work focused on drinking on college campuses has
demonstrated that inflated perceptions of descriptive norms lead individuals to engage in
consumption patterns that are perceived to be normative (e.g., Perkins & Berkowitz, 1986).
Studies have found that perceptions of a normative behavior occurring (descriptive norms) lead

15

to engagement in the behaviors of condom use and homophobic communication (Elwood,
Greene, & Carter, 2003; Hall & LaFrance, 2012).
Research conducted by Lapinski, Anderson, Shugart, and Todd (2013) found that
injunctive norms (beliefs about what ought to be done ) moderate the effect of descriptive norms
on behaviors, such that when perceptions of approval were high, descriptive norms were
positively associated with engaging in the behavior. Park and Smith (2007) documented that as
personal injunctive norms increased in strength, subjective norms became a significant predictor
of behavioral intent. Cornacchione (2013) found that as perceptions of approval increased (high
injunctive norms), the likelihood of engaging in communication with a family member about a
health issue also increased. Based on previous research, it is expected that perceived norms about
engaging in communication will be positively associated with actual engagement in
communication.
H2a: There will be a positive relationship between a woman's perception of other women
on probation and parole engaging in communication with their PO about difficult issues
and actual engagement in communication. (descriptive norms)
H2b: There will be a positive relationship between a woman's perceptions of approval
from normative referents about engaging in communication with her PO about difficult
issues and actual engagement in communication. (injunctive norms)
H2c: There will be a positive relationship between a woman's perceptions of normative
referents expecting her to engage in communication with her PO about a difficult issue
and actual engagement in communication. (subjective norms)
It is possible that the role of norms for women on probation and parole in predicting
communication behavior may be nonexistent. As previously mentioned, women on probation

16

and parole have smaller social networks, thus, potentially making normative impacts irrelevant.
If this is the case, then it might be difficult to determine the normative referents for the women.
Additionally, there may be no perceived norms because she might feel isolated. In fact, many
women have reported staying inside their home and away from other people as ways to stay safe
in their neighborhood and avoid criminal activity (Cobbina, Morash, Kashy, & Smith, in press).
Research has also found that social norms can be intentionally violated (Ross &
Mirowsky, 1987). One study noted that adolescent sex offenders and juvenile offenders were
more normless than nondelinquent youths. Normlessness is defined as the “subjective
dissociation from socially approved norms governing behavior and…acceptance of socially
unapproved behavior as a way to achieve conventional goals" (Miner & Munns, 2005, p. 493).
This means that individuals are aware of the norms that exist, but find it socially acceptable to
deviate from these norms to achieve a greater goal that would not be possible to achieve while
adhering to social norms. Additionally, in committing the crimes, women already violated
societal norms for proper conduct and behavior, partially due to desperate situations and trying to
fulfill needs. Given the complexity of social norms for offenders, there may be additional factors
other than goals and norms that influence whether or not offenders who want to talk to their PO
actually engage in communication.
Context of Study & Additional Factors Influencing Desired Communication with POs
This study is situated within a three-wave National Science Foundation-funded study of
402 women on probation and parole that examines the communicative relationship between
women on probation/parole and their probation/parole officers to see what predicts successful
outcomes, including reduced recidivism. Interviews were spaced approximately three months
apart.

17

Additional data collected through this NSF study might help explain what impacts
whether or not women on probation and parole engage in desired communication with their POs
about difficult issues or needs. In fact, it is possible that these factors might predict the outcome
above and beyond the goals and norms, especially since, as previously discussed, norms may not
be well-formed for the women. Factors unique to the offenders, such as their supervision status
(probation/parole), how many criminogenic needs they have, as well as the quality of their
relationships might impact their "dependence" on and the importance of their relationship with
their POs. This might, in turn, have an effect on whether or not the conversation occurs. Factors
unique to the POs, such as the women's perception of their (PO) communication style, could also
impact whether the conversation actually occurs. Analyses will be conducted with these
additional variables that could potentially predict whether or not individuals talk to their PO.
These variables are all at the personal level and based on the women‟s perceptions/self-reports.
Supervision status. Whether a woman is on probation or parole may impact whether or
not she talks to her PO. It has been documented that women on probation and parole cite their
PO as a main source of support (Morash, 2010). However, parolees may face greater challenges.
Returning home from prison is often laden with difficulties. Many returning prisoners face
barriers and obstacles that reduce their rights as citizens, such as being ineligible for public
assistance, food stamps, and education loans (Petersilia, 2001). Additional research has
demonstrated that women on parole tended to recall memorable messages from their PO more
than women on probation (Cornacchione et al., 2013), indicating that women on parole may rely
more on their PO than women on probation. The nature of the women's community supervision
(i.e., probation vs. parole) will be examined to see if it is related to whether or not women talk to
their PO about a difficult issue.

18

H3: Women on parole are more likely to engage in communication with their PO than
women on probation.
Criminogenic needs assessment. Women on probation and parole have various needs
and challenges that need to be dealt with while on community supervision, and they may need to
communicate with their PO to address these issues. Criminogenic needs are factors that have
been found to predict or cause crime. Factors unique to female offenders have been identified,
and are measured using the gender-responsive needs assessment developed by Van Voorhis and
colleages (e.g., Salisbury, Van Voorhis, & Spiropoulos, 2009; Van Voorhis et al., 2008; Van
Voorhis, Wright, Salisbury, & Bauman, 2010). The predictors of crime (i.e., criminogenic needs)
for women include parental stress; self-efficacy; trauma, victimization, and abuse; mental health;
self-esteem; intimate relationships; unsafe housing; few educational assets; and financial issues,
among others (Van Voorhis et al., 2008).
Besides just identifying risk factors, various strengths have been identified in the
literature, and these strengths help women to avoid future criminal behavior. These strengths
include self-esteem, self-efficacy, family and relationship support, and financial and educational
assets (Van Voorhis et al., 2008). It is expected that if women have a large number of
criminogenic needs (or more needs than strengths), then they will be more likely to talk to their
POs about addressing these needs or dealing with difficult issues. Needs are conceptualized as a
total risk score, which subtracts the women's total strengths from their total risk score. Women
who have a larger criminogenic risk score have more to gain from communicating with their POs
compared to women who have fewer needs (i.e., a lower criminogenic risk score).
H4: There will be a positive relationship between the women's total risk score (needs
minus strengths) and engaging in communication with their PO.

19

Social networks & social support. Women on probation and parole may have smaller
and weaker social networks, especially for those who are uneducated and low-income (Reisig et
al., 2002). It is expected that if women have small social networks, they will be more likely to
communicate with their POs to achieve their needs. Female offenders who have used drugs and
alcohol often receive little support from friends and family (Mallik & Visher, 2008). Previous
research found that a greater proportion of women on parole compared to women on probation
recalled memorable messages they received from their PO, and this finding may be due to
parolees having smaller social networks after spending time in prison (Cornacchione et al.,
2013). Women who have been to prison and who are on parole often have weakened support
networks due to their physical absence from their communities of origin compared to women on
probation who have committed a crime but has been sentenced to a period of correctional
supervision in lieu of imprisonment; thus, their social networks remain intact.
The quality of the relationships (i.e., social support received) within the women's social
networks might also influence their behaviors, including talking to their PO about difficult issues
they are facing. Estroff and Zimmer claim that "social support characteristics...might either
constitute risk factors for violence by creating the opportunity or need for violence or serve to
prevent or decrease the opportunity and need for threatening or assaultive behaviors" (1994, p.
260). Cullen (1994) also noted the importance of social support in criminology, acknowledging
that it can serve as a protective factor for offenders to help individuals refrain from engaging in
criminal activity and help them reintegrate into society. With difficult and uncomfortable
situations abounding for those on probation and parole, the effects of social support have been
found to mediate (and dampen) the relationship between discomfort and hostility compared to
those lacking social support (Hochstetler, DeLisi, & Pratt, 2010). Because of the significant

20

impact that support has on offenders and their criminal behavior and feelings, it is reasonable to
expect that social support characteristics may also be predictive of behaviors other than violence,
such as communication. If women perceive that they are lacking social support from individuals
in their network, then they may be more likely to talk to their PO about needs and difficult issues
to obtain the support needed.
Social support is a communicative process (Burleson, Albrecht, Goldsmith, & Sarason,
1994) whereby individuals in ongoing relationships seek and provide support to one another
(Barnes & Duck, 1994). Traditionally, social support is communicated in five ways:
informational, tangible, esteem, emotional, and social network. Informational and tangible
support are considered types of action-facilitating support that help the distressed individual
solve the problem through both information support and tangible aid (Cutrona & Suhr, 1992).
Esteem, emotional, and network support are considered types of nurturant support that provides
comfort to the stressed individual without providing any direct support to solve the problem
(Cutrona & Suhr, 1992). The three types of support examined in this study are emotional
support, social network support, and instrumental support, which are derived from work
conducted by Estroff and Zimmer (1994).
Estroff and Zimmer conceptualize emotional support as whether the woman can talk to
anyone in her social network if she is upset or worried, and if she perceives that someone truly
cares about her. Emotional support is important for female offenders who are under community
supervision because they likely face multiple challenges such as finding a job, remaining sober,
and trying to regain custody of children. Because these are stressful life events that are likely to
induce stress and worry, it is important for offenders to have individuals in their network that
they can talk to when they are upset or worried to help cope with these stressors.

21

Social network support is conceptualized as an individual's having someone with whom
she spends time. If women indicate that they do not spend time with individuals within their
network, that could impact their sense of support, and subsequently, their decisions to talk to
their POs. Research indicates that spending time with other individuals can help buffer against
reoffending, and this finding is especially strong for individuals who received visitors during
incarceration (e.g., Cochran, 2014). However, research also indicates that many offenders avoid
crime in their neighborhood by staying alone (e.g., Cobbina et al., in press). Depending on the
social network composition, it can result in either positive or negative behaviors (Portes, 1998).
Finally, instrumental support is determined by whether the woman would ask members in
her social network for help, including assistance in obtaining goods and services. Having
services provided by social network members (such as transportation and childcare) can help
prevent offenders from turning to crime to satisfy their needs, for example by driving a car
without a license (Wright, DeHart, Koons-Witt, & Crittenden, 2013).
One reason a woman may not receive the support she needs is because of the quality of
relationships she has with her family, friends, and romantic partners. Women have frequently
experienced physical and sexual abuse as children and adults, and oftentimes, family members
are the ones who abuse them (Leverentz, 2006). Additionally, women on parole may be more
detached from their previous relationships, so support may be especially hard to receive from
family, friends, or romantic partners. Thus, it is expected that women with small social networks
will be more likely to talk to their POs. Additionally, women who report receiving little support
from their network will also be more likely to talk to their POs about difficult issues. Thus, it is
predicted and asked that:

22

H5: Women with smaller social networks will be more likely to engage in
communication with their PO compared to women with larger social networks.
H6: As the percentage of individuals within their social network who provide support
(instrumental, emotional, social) increases, women will be less likely to engage in
communication with their PO.
In addition to these offender-specific characteristics, it is important to consider whether
there are PO characteristics that might also affect whether or not communication occurs.
Perceptions of PO characteristics and interactions. Women's perceptions of their PO
may also impact whether they talk to their POs about issues they face. PO-specific characteristics
could include communication style and elicited emotions. The perceived PO communication
style could have substantial effects on whether or not women engage in desired communication
with their PO. The measure of the PO communication style is based on the family
communication patterns measure in family communication literature. The family communication
patterns (FCP) framework focuses on the ways in which individuals perceive their
communication with other family members (Koerner & Fitzpatrick, 2002). The scale was
originally developed to address the degree to which parents' discussion of ideas and concepts
(concept-orientation) or social roles/relationships, such as authority (socio-orientation)
influences children's information processing, decision making, and behaviors (McCleod &
Chaffee, 1972; Schrodt et al., 2008). The roles within the original measure, and subsequent
versions (e.g., Ritchie & Fitzpatrick, 1990) examine the relationship between individuals of
authority (parents) and their subordinates (children). In this context, the framework focuses on
the ways in which offenders (subordinates) perceive their communication with their PO (an
authority figure).

23

Previous research on the FCP has found that some communication patterns lead to
desirable outcomes (Schrodt et al., 2008). The FCP specifies two communication orientations
that idividuals in authority can adopt: conversational and conformity (Koerner & Fitzpatrick,
2002a; Ritchie & Fitzpatrick, 1990). Conversation orientation refers to how open communication
is within the relationship or group. This communication pattern encourages discussions that
recognize the contributions and perspectives of both the PO and the offender. The conversational
communication orientation encourages everyone to be independent, share feelings, and
contribute to the decision-making process (Koerner & Fitzpatrick, 2002; Ritchie & Fitzpatrick,
1990), which would be exceptionally important for offenders (Taxman & Ainsworth, 2009).
Conformity orientation refers to the extent to which communication reflects obedience (e.g.,
Ritchie & Fitzpatrick, 1990). Conformity orientation generally deals with how much emphasis is
placed on obedience to authority (Fitzpatrick & Ritchie, 1994). In a meta-analysis examining the
FCP, the greatest effect sizes between communication patterns and outcomes were for
conversation orientation, and those being associated with many positive outcomes, such as less
alcohol consumption (Schrodt et al., 2008). Women who perceive that their PO engages in a
more conversational communication orientation may be more likely to actually engage in the
conversation. Conversely, women who perceive that their PO engages in a more conformity
communication orientation may be less likely to talk to her PO.
H7: There will be a positive relationship between the women's perceptions of their PO's
conversation communication orientation and engagement in communication.
H8: There will be a negative relationship between the women's perceptions of their PO's
conformity communication orientation and engagement in communication.

24

Negative feelings and reactions elicited by the PO during meetings with their clients may
impact whether or not women decide to talk with their POs about the needs or issue they are
facing. Research on psychological reactance (Brehm, 1966; Brehm & Brehm, 1961) explains that
when people feel like their freedom is being restricted, they are likely to do the opposite of what
is told of them. For example, if the PO makes her client feel angry, guilty, or ashamed during
interactions, the woman may not do what the PO tells her to do in order to establish her freedom.
She may ignore the PO and do nothing, or do the opposite of what the PO suggests. Thus, it
would be expected that if women had these negative feelings when interacting with their PO,
then they would be less likely to engage in communication with her.
H9: There will be a negative relationship between women's reported a) psychological
reactance, b) emotional reactance and c) anxiety during meetings with their PO and
engagement in communication.

25

CHAPTER 2: METHODS
Participants
The focus of this study is understanding the relationship between female offenders and
their PO by examining the factors that impact whether or not women on probation and parole
who desire to talk to their PO about difficult issues or needs they face actually engage in the
conversation. To examine this, the effects of goals and social norms on communication are
examined, along with secondary variable considerations. Study respondents at the Time 1
interview consisted of 402 women on probation and parole in Michigan who participated in the
interview. Of these, 305 (75.87%) were on probation, 93 (23.13%) were on parole, and 4 (<1%)
reported being on both probation and parole. The sample age ranged from 18 to 60 with a mean
of 33.87 (SD = 10.53). The sample was ethnically and racially diverse and included 198 White
women (49%, n = 12 of these reported being both White and Hispanic), 147 Black women
(36.6%, n = 10 reported being Black Hispanic), one Pacific Islander (<1%), and four Native
American (<1%). Forty-four (11%) additional women identified multiple racial group
memberships. An additional three women reported other racial groups, two refused to answer,
one did not know her racial group, and two others did not provide information. Approximately
half of the women reported having children (51.7%, n = 208), 52 were married (12.9%), and 232
were in dating relationships (57.7%).
Data were collected from the women in 3-waves, each spaced approximately 3 months
apart. Women were initially recruited using various methods. Interviewers spent periods in the
parole and probation office sites, and if women agreed, agents (probation/parole officers)
introduced them to an interviewer to hear about the project. Alternatively, if women gave their
agents permission, agents provided phone contact information and research staff called women

26

to arrange a time to hear about the study. For the T2 and T3 interviews, women were contacted
by project staff, which could include their T1 interviewer. The women were contacted through
various methods such as text messaging, phone calls, email, letters mailed to their last home
address, and asking their PO for help. Women consented to being contacted for subsequent
interviews during the T1 interview. The T1 interview lasted approximately 90 minutes, and the
women received a $30 gift card to their choice of Walmart, Target, or CVS. After completing the
T2 interview, which lasted approximately 45 minutes, women received a $50 gift card. The T3
interview lasted approximately 90 minutes, and women received a $75 gift card.
Goals and social norms data were collected during the T3 interview. Measures from T1
and T2 are also used in the data analysis. At the T3 interview, 394 female offenders on probation
or parole were interviewed face-to-face using a structured interview conducted by trained
interviewers. During the difficult conversations portion of the interview, the women were asked
a series of questions to generate quantitative measures of goals and norms. Additionally, openended questions were asked by interviewers so that more in-depth responses could be obtained.
This portion of the interview was audio recorded.
At the beginning of the difficult conversations portion of the interview, women were
asked if there was a time when they faced a difficult issue about which they wanted to talk to
their PO. If the woman responded no, she skipped over this portion of the interview. If the
woman responded yes, she was then asked to describe this difficult issue (topic). For those
indicating that there was a time when they wanted to talk to their PO, the women were next
asked if they actually engaged in the conversation with their PO (yes/no). If they did engage in
communication with their PO, they were hen asked to describe what they said to their PO, as
well as their main goal for having the conversation. If the women did not engage in the

27

conversation with their PO, they were then asked to explain why the conversation did not occur
(i.e., the main concerns that prevented them from talking to their PO). All women who wanted to
talk to their PO, regardless of whether the conversation actually occurred, next answered a series
of qualitative and quantitative questions regarding perceived personal level norms (descriptive,
injunctive, and subjective) as well as multiple goals and constraints. Women were given the
option to read the questions and select the answers themselves or have them read to by the
interviewer. Quantitative items were modified to be more understandable by this population,
since many have low literacy rates. The Flesch-Kincaid readability is at an 8th grade reading
level. Additionally, for the purpose of keeping the interview as short as possible, each scale was
reduced to three items.
Measures
Preliminary data analyses included checks for reliable measures. SPSS was used to
compute Cronbach‟s alpha coefficient for each scale with three or more items in order to check
for scale reliability. Psychometric properties (mean, standard deviation, range, reliability) from
each scale are presented in Table 1. Correlations among all variables are presented in Table 2.
Goals. Measures to assess goals constructs were adapted from Dillard et al. (1989) and
were scored such that higher scores indicate greater perceptions of the construct being measured.
All items were measured using 7-point Likert-type scales ranging from strongly disagree to
strongly agree. Three items were used to measure each norm type. Thus, the dimensionality of
these measures cannot be assessed.
Influence goals were measured with three items, such as “I was very concerned about
getting what I wanted in this conversation.” After removing one item (diff_talk5: "the outcome

28

of this conversation had important personal consequences for me"), the scale exhibited
acceptable reliability, α = .64.
Three items were used to measure interaction constraints. An example question to
measure this variable is, “I was careful to avoid saying things which were socially
inappropriate.” After removing one item (diff_talk8: "I was concerned with making or keeping a
good impression"), the scale exhibited acceptable reliability, α = .63.
Identity was measured using three items, such as “In this conversation, I was concerned
with not violating my own standards for what is right and wrong.” The scale exhibited poor
reliability, α = .27. Thus, this variable is excluded from data analysis.
Three items were used to measure relational resource constraints/considerations,
including “I was not willing to risk possible damage to my relationship with the agent to get
what I wanted.” The scale exhibited poor reliability, α = .14. Thus, this variable is excluded from
data analysis.
Three items were used to measure personal resource constraints, such as “I was afraid to
have the conversation because it could take away my freedom or lead to more supervision.” The
scale exhibited acceptable reliability, α = .72.
Arousal management was measured using three items, including “I was worried about the
possibility this conversation would make me uncomfortable.” The scale exhibited acceptable
reliability, α = .80.
Social norms. Measures to assess social norms (subjective, personal level descriptive,
personal level injunctive) were adapted from Park and Smith (2007). Items were measured using
7-point Likert-type scales ranging from strongly disagree to strongly agree, and were scored so

29

that higher values indicated a greater perception of the norm existing. Three items were used to
measure each norm type. Thus, the dimensionality of these measures cannot be assessed.
Three items were used to measure personal level descriptive norms, including "People
who are important to me and in a situation similar to mine have talked with their POs about this
issue.” The scale exhibited acceptable reliability, α = .80.
Three items were used to measure personal level injunctive norms, including “Most
people who are important to me approve of talking to my PO about this issue.” The scale
exhibited acceptable reliability, α = .84.
Three items were used to measure subjective norms, such as “My peers expect me to talk
with my PO about this issue.” The scale exhibited acceptable reliability, α = .83.
Open-Ended Data Analysis
Responses to open-ended questions were audio recorded and transcribed into Microsoft
Word for qualitative data analysis. Two research assistants (one undergraduate and one master's
student) were independently trained to code the open-ended responses. The unit of measurement
was the complete response the woman provided for each question section. The trained coders
were given approximately 20% of the open-ended responses and coded them to establish initial
coding reliability and to assess discrepancies that might exist. Once adequate reliability was
established and all discrepancies resolved between the coders, the remaining responses were
coded independently. Data were then input into SPSS from Microsoft Excel.
Categories to code the data were first determined by the primary researcher after reading
through the transcripts. Once the coders began coding for initial reliability, they identified any
other common themes they saw emerging. The coders and the researcher then discussed whether
to add or remove any of the categories. This happened multiple times throughout the coding

30

process to determine if the coding scheme needed to be modified. The coders went through up to
five rounds of coding and codebook revision before acceptable levels of reliability were
achieved. Revision and training occurred after each round of reliability analysis. The researcher
documented discrepancies and identified any common errors that occurred to assist in training
the coders or editing the codebook. Additionally, coders were asked their input on any additional
categories to add, change, or eliminate. The variables coded were the main reason/goal for
having the conversation (RQ1), the constraints/reasons reported for not having the conversation
(RQ2), and the topics of conversation (descriptive information). Initial Cohen's Kappa for the
main goal for having the conversation was .29, and progressed sequentially to .24, .60, and .76.
Initial Cohen's Kappa for constraints to having the conversation was .68. Initial Cohen's Kappa
for topics of the conversation was .75. Final Cohen's Kappa reliabilities are listed below. The
coding scheme with conceptual definitions is located in Table 3.
Motivations and constraints. Women who engaged in the conversation were asked what
was driving them, or their main goal, in having the conversation with their POs. Eleven
categories were identified: inform PO, inform PO specifically to avoid violation/trouble, initiated
by PO, confession, seek informational support/advice, seek tangible support, seek help generally,
emotional expression, and other. Two of the 11 categories were interview errors: not asked by
interviewer and asked incorrectly. Cohen's Kappa for coding primary goals is .88. Categories and
conceptual definitions are located in Table 3.
Women who did not engage in the conversation with their PO were asked the main
reason why the conversation did not occur. Seven categories were identified: prevent restricted
freedom, PO lacks resources, PO communication issues, negative PO relationship, maintain
positive relationship with PO, other, and interviewer didn't ask. Cohen's Kappa for coding the

31

reported reasons for not engaging in the conversation is .87. Categories and conceptual
definitions are located in Table 3.
Topics. All women who desired to have the conversation, regardless of whether or not
they engaged in the conversation, were asked to report the topic or issue they wanted to talk to
discuss with their PO. Twenty-one categories were identified: housing issues, paying back
restitution fees, finding a job/job issues, both restitution and job issues, need services (financial,
support), transportation issues, relapsing generally, illegal activity--using substances, illegal
activity--non-substances, contact with police, talk about a third party (relationship issues), abuse,
associating with others on probation/parole, children issues, relationship with PO, end
supervision early, mental health, previous behavior, and other. Two of the 21 categories entailed
interview errors: interviewer didn't ask and woman didn't answer. Cohen's Kappa for coding the
topics is .87. Topics and conceptual definitions are listed in Table 3.
A copy of the difficult conversations portion of the interview is located in Appendix A.
Secondary Data Analysis
A number of other predictor variables were outlined earlier in the paper. These interview
items are included in Appendix B. They are classified as secondary variables because they were
variables already collected in the larger data set, and are examined in addition to the goals and
norms.
Supervision status. At the beginning of the Time 1 interview, women were asked
whether they were on probation or parole. A dummy-coded variable was created to indicate that
0 = parole and 1 = probation to include into the regression model with parole as the reference
category.

32

Criminogenic needs assessment. Several variables were examined to measure the
women's total needs assessment using the Van Voorhis scale (Van Voorhis et al., 2008).
Variables included in the needs assessment are: criminal history, education,
financial/employment status, housing safety, having antisocial friends, anger, history of mental
illness, current mental illness, abuse and trauma, substance abuse, family and romantic
relationship quality, parenting, and self-efficacy. A total needs score was calculated at the T1
interview that took into account the women‟s needs and strengths. This variable was created by
subtracting a woman's total strengths score (education, family support, self efficacy) from her
total risk score (anger, anti-social attitude, criminal history, financial/employment, unsafe
housing, anti-social friends, current depression/anxiety, current psychosis, current and history of
substance abuse, family conflict, parental stress). The greater the score, the more needs that exist
for the women. More details regarding the specific subscales' measurement are located in
Appendix C. The mean total risk score for the sample is 24.40 (SD = 8.03, Min = 7, Max = 44)
and the mean total strengths score is 2.98 (SD = 1.54, Min = 0, Max = 6).
Social networks & social support. During all three interviews, women were given a
social network inventory sheet. Each woman filled in the initials of each person over the age of
17 she regularly spends time with. For each individual, she supplied information regarding
his/her gender, age, and relationship. In addition, dichotomous social support questions were
asked for each of the individuals listed on the social network inventory that measures perceived
availability of social support, and the percentage of people within each woman's social network
who she perceived that she could get that type of support from was calculated. For perceived
availability of emotional support, or whom the woman perceives she could talk to if she was
upset or worried, included questions such as “When you are upset or worried about something,

33

who would you talk with?” For perceived availability of social network support, which is the
woman's perception of availability to spend time with someone, the following question was
asked: “Is this someone you hang out with, or spend time with?” Finally, “Would you ask this
person if you needed something, like help doing something, money, or help caring for family
members?” was used to measure instrumental support, or the perceived availability of
individuals in their network whom the women could ask for help. The sum of the number of
people listed on the social network inventory is used to analyze the data to indicate the size of
their network.
Descriptive information on the social network was obtained by running descriptive
statistics. For the current sample, the mean size of the women's social network at the T1
interview was 4.99 (SD = 2.69, Min = 1, Max = 15). Many women (41%) listed their boyfriends
in their social network as well as their mothers (41%) with fewer reporting that their fathers were
members of their social network (14%). Friends were listed in the social network inventory (M =
1.67, SD = 1.65, Min = 0, Max = 7) more frequently than siblings (M = 0.71, SD = 0.98, Min = 0,
Max = 5). The majority of individuals listed within the social network inventories were female
(62.4%). The average number of individuals who had committed an offense within a woman's
social network was approximately one (M = 1.36, SD = 1.50, Min = 0, Max = 6). With those who
reported a lover within their network (n = 52), 59.62% of those lovers had committed an offense.
Additionally, of those who reported their mother within their network (n = 51), 10% of those
mothers had committed an offense, while 38.89% of fathers in the network had committed an
offense (n = 18).
The percentage of perceived availability of the three types of support was calculated by
summing each type of support that was reported in the social network inventory, then dividing

34

that sum by the total number of people listed in each woman's network. Overall, high levels of
perceived availability of support were reported. On average, each social network comprised of
high levels of perceived availability o instrumental support (M = .76, SD = .26), emotional
support (M= .92, SD = .16), and social network support (M = .73, SD = .27).
Perceptions of PO communication style. To measure women‟s perceptions of their
PO‟s communication style, the revised family communication patterns scale was used (Ritchie &
Fitzpatrick, 1990). As previously mentioned, the roles within the original measure and
subsequent versions (e.g., Ritchie & Fitzpatrick, 1990) examine the relationship between
individuals of authority (parents) and their subordinates (children). In this study, the measure
focuses on the ways in which offenders (subordinates) perceive their communication with their
PO (authority).
Twenty-six items were used to measure conformity and conversation communication
orientation on a 7 point Likert scale ranging from 1 = very strongly disagree to 7 = very strongly
agree. Example conversation orientation items include “My PO frequently asks my opinion when
we are talking” and “I can tell my PO almost anything.” Example conformity orientation items
include “Sometimes my PO says something like „my ideas are right and you should not question
them‟” and “My PO expects me to obey him/her.” Both exhibited acceptable levels of reliability,
as evidenced by Cronbach's alpha of .89 for conversation and .82 for conformity.
Elicited emotions & reactance. To measure psychological reactance, 7 questions were
asked to determine how strongly women agreed with statements regarding the reactance process.
Seven items were adapted from the Therapeutic Reactance Scale (Dowd, Milne, & Wise, 1991)
to reflect reactance in response to interactions with the probation or parole agent. Women
responded to 7-point scales ranging from 1 = very strongly disagree to 7 = very strongly agree to

35

items such as “I get very irritated when my PO tells me what I must or must not do” and “Often I
lose enthusiasm for doing something just because my PO expects me to do it." The scale
exhibited an acceptable level of reliability, α = .91.
A measure of emotional reactance was also used to examine negative emotions elicited
by the PO. This measure was adapted from previous work examining reactance (e.g., Dillard &
Shen, 2005; Quick & Stephenson, 2007). Women were asked the extent to which they felt guilty,
ashamed, annoyed, irritated, and angry during and after talking with their POs on a scale from 1
= not at all to 5 = very much. The scale exhibited acceptable reliability, α = .85.
To measure anxiety, items from the anxiety subscale of the Brief Symptom Inventory,
shortened version (Derogatis & Melisarotos, 1983), were adapted for this study. This measure
asked women to evaluate how much they experienced six anxiety-related states after interacting
with their supervising agent, including “Nervousness or shakiness inside”, “Feeling scared for no
reason”, and “Feeling tense or keyed up”. Women reported using 5-point scales ranging from 1 =
not at all to 5 = very much. The scale exhibited acceptable reliability, α = .94.

36

CHAPTER 3: QUANTITATIVE DATA ANALYSIS OVERVIEW
Ideally, one binomial logistic regression would be conducted to test hypotheses H1-H9
because it would be parsimonious (see Table 4 for test results). Generally, there should be 20
cases per predictor, with some researchers suggesting 30-50 to avoid Type II error rates (Hart &
Clark, 1999). With 17 predictors, this requires a sample size of at least 340 to run a full model.
Thus, separate logistic regressions will be run (goals, norms, offender traits [status, risk score],
social network characteristics [size, social support], and reactions to and perceptions of the PO
[anxiety, reactance, communication style]). The dependent variable in all statistical analyses will
be whether or not the women engaged in communication with their PO (1 = yes, 0 = no).
Predictor variables are continuous measures of social norms (descriptive, injunctive, subjective),
goals (influence, identity, interaction, relational resource, personal resource, arousal
management), anxiety, reactance (emotional and psychological), perceived PO communication
style, social support, and social network size. Supervision status (probation vs. parole) is
dummy-coded (0 = parole, 1 = probation) to be entered into the logistic regression, with parole
as the reference category. To further examine the relationship between supervision status and
engagement in the conversation, a chi-square test is used to look at the associations between
variables.
Sample Selection Bias
The nature in which the social norms and goals items were collected introduced a
selection bias. A subset of women answered those questions only if they answered "yes" to the
question of whether there was ever a time when they wanted to talk to their PO about a difficult
need or issue they were facing. If the women responded "no", a skip pattern took them past this
section. Thus, it is possible that there might be systematic differences between women who did

37

and did not self-select into this sample. Independent sample t-tests were conducted with all
secondary predictor variables to compare the differences between women who reported there
was a time when they wanted to talk to their PO and women who reported there was never a
time; this analysis revealed two significant differences. The independent variables entered into
the equation were probation/parole status, total needs score, size of social network, three ratio
support variables, anxiety, psychological reactance, emotional reactance, perceptions of PO
conversation orientation, and perceptions of PO conformity orientation. Differences emerged
between the two samples for total needs score, t(360) = -2.68, p = .008, η2=.02, with those who
said they wanted to talk reporting, on average, more needs (M = 21.42, SD = 8.48) than those
who reported not wanting to talk to their PO (M = 18.80, SD = 8.92). Additionally, significant
differences emerged for PO-elicited anxiety, t(356) = -2.31, p = .02, η2 = .01. Those who wanted
to talk to their PO reported a greater anxiety score (M = 1.86, SD = 0.99) than those who did not
want to talk to their PO (M = 1.59, SD = 1.11). However, the effect sizes are small. This study
examines the factors that facilitate or inhibit communication with POs for women who want to
talk to their POs. Thus, the findings of this study can only be generalized to those women who
wanted to talk to their PO.1

38

CHAPTER 4: RESULTS
Sample Description
Out of the original sample of 402, 127 women on probation and parole indicated that
there was a time when they wanted to talk to their POs about a difficult need/issue they were
facing. Of the 127, 84 (67.9%) talked to their PO, while 41 (32.8%) did not engage in the
conversation. One individual did not know whether she talked to her PO, and another refused to
answer the question. The majority of women were on probation (70.9%, n = 90), while 35 were
on parole (27.6%). Two individuals indicated they were on both probation and parole (1.6%). In
the data analysis, these two individuals were grouped into the parole category since they had
recently been incarcerated. The majority of women reported being white (52.7%, n = 67, n = 3
reporting Hispanic), followed by black (31.4%, n = 40, n = 2 reporting Hispanic). Fifteen women
indicated they were multiracial (11.8%, n = 15). Other women were Indian (0.02%, n = 2, n = 1
reporting Hispanic) and Pacific Islander (0.01%, n = 1).
The most frequently reported topic of conversation was regarding a third party (20.3%, n
= 25), such as "my mom and her drinking." Many women also reported housing issues as a
frequent topic about which they wanted to talk to their POs (17.8%, n = 22), such as "I want my
own place." The remainder of topics were fairly evenly distributed among the women. Several
women reported topics related to illegal behavior, such as relapsing (2.4%, n = 3), using illegal
or controlled substances (4.1%, n = 5), engaging in illegal behavior not related to substances
(1.6%, n = 2), such as shoplifting, and having police contact (1.6%, n = 2). Topics related to
financial issues and other support services needed by women were also reported, such as
restitution (4.9%, n = 6), finding a job (4.1%, n = 5), job and restitution issues (1.6%, n = 2),
need services (4.1%, n = 5), and transportation issues (2.4%, n = 3). Topics related to being on

39

community supervision were also reported, such as wanting to end supervision early (3.3%, n =
4), associating with others on probation/parole (3.3%, n = 4), the women's relationship with PO
(1.6%, n = 2). Other topics reported by women were partner abuse (1.6%, n = 2), previous
behavior (3.3%, n = 4), children issues (2.4%, n = 3), mental health (2.4%, n = 3), and other
(6.5%, n = 2). In nine instances, the interviewer did not ask the question (7.3%) and there were a
few times when the woman did not answer the question (3.3%, n = 4). See Table 5 for a
breakdown of topics by whether the women engaged or not in communication with their POs.
Tests of Hypotheses and Research Questions
Goals and constraints. H1 predicted that personal resource considerations would be a
significant predictor of whether or not the women engaged in communication. A binomial
logistic regression was conducted with the influence, interaction, personal resource, arousal
management variables entered as predictors. The overall model was statistically significant, χ2
(4) = 15.21, p = .004. The model was able to correctly classify 64.8% of the participants as
engaging or not engaging in communication based on the predictor variables. Personal resource
was found to be the only significant predictor, with those reporting higher levels of personal
resource considerations (i.e., concerns about losing freedom) being less likely to engage in
communication with their PO,  = -0.36, Wald‟s 2 = 6.08, p = .014. For every one-unit increase
in personal resource ratings, the odds of engaging in communication increased by 0.72. See
Table 6 for the logistic regression coefficient, Wald test, and odds-ratio for all predictors.
RQ1 asked the reasons reported by women on probation/parole for engaging in
communication with their supervising officer about a difficult need/issue they were facing.
Collectively, the most frequently reported reasons for engaging in the conversation with their
POs were to obtain some form of support (37.4%). Specifically, many reported wanting to

40

receive informational support (15.7%, n = 13), such as obtaining advice on how to find a job
with a conviction. One woman reported "Just to get a better understanding of everything that I
can do." Women also reported their main reason for engaging in the conversation was to receive
tangible support (13.3%, n = 11), such as getting into a rehabilitation program, or "maybe even
coming up with some kind of funding to pay for it." Several women were seeking help generally
(8.4%, n = 7) without specifying whether tangible or informational support were desired. For
example, one woman stated
"Because I wanted help. I wanted help, and I pretty much didn‟t know … I can‟t say I
didn‟t know where to go I just … you know, I hadn‟t talked to anyone else about it, you
know. I just knew I‟d get up in the morning and I wanted vicodin, you know, it‟s pain
pill. And it wasn‟t because I was in pain so much as the injuries; it was … I had got
addicted to the pain pill."
Many women also reported that their main reason for having the conversation was for
emotional expression (14.5%, n = 12). One woman said that her drive in having the conversation
with her PO was "To look at it more positive, you know, just to be, like not let it get to me as
much, I guess."
Several women indicated that they had the conversation to avoid getting into trouble
(13.3%, n = 11). For instance, one woman stated:
"When I went to her and told her it was to cover myself. To make sure that I wasn‟t
gunna be doing something that got me in trouble or violated my parole by dealing with,
which this person wasn‟t a felon or nothing like that, but I mean, just dealing with them
on the police side of things where it had got out of control like that where I was thinking,
you know, now I have to go to the police about it."
Similarly, some women stated that they wanted to "come clean", or confess, to their POs
(3.6%, n=3). One woman reported the main reason for having the conversation was, "A
conscious because it is wrong. And I think that at somewhere in my mind that if I bring it up in
even a shaded color, you know, that it‟s getting it off my chest, so…"

41

Other women wanted to inform their POs about the issue (13.3%, n =11). For example,
one woman said "To make her aware of it." A few women indicated that the conversation
occurred because the PO initiated it (3.6%, n = 3), such as "The fact that she asked what was
going on. She just took the time and asked me." Although this indicates that it was not a
motivation to have the conversation per se, it alludes to the idea that the conversation might not
have happened if the PO did not initiate the discussion.
Four responses were coded as other (4.8%), and several were coded as interview "errors",
such as the interviewer not asking the question (7.2%, n = 6) or asking the question incorrectly
(2.4%, n = 2).
RQ2 asked about the reported constraints on engaging in communication. Many women
reported that they did not have the conversation because they did not want their PO to restrict
their freedom as a result of having the conversation (27.5%, n = 11), such as, "„Cause I don‟t
want any retaliation from her, or repercussions from her." Another large percentage of women
did not engage in the conversation because of issues related to their PO. Many reported PO
communication issues (25%, n = 10). For example, one woman stated:
"I don‟t know, like I just feel … like when I go see her, it‟s … you know she asks me the
usual questions like, “Have you had any police contact or any changes in your address?”
And then she gives me my new report date and then sends me on my way. It‟s not … it‟s
almost like you get in there to … for her to just ask you about this. There‟s like no time to
talk, which I just feel like it‟s a waste."
Many women also reported a negative relationship with PO (17.5%, n = 7) as a reason
for not engaging in the conversation, such as "Because she don‟t make me feel like I could come
to her with that kind of stuff, you know how some people just don‟t seem like you can just come
over and talk to them?" Only one woman reported that she did not want to have the conversation
because she was afraid of disappointing her PO and wanted to maintain the relationship with her

42

PO (2.5%). Several women indicated that the main reason why they did not have the
conversation was because they perceived it to be pointless--that the PO lacked resources to help
even if they wanted to help (10%, n = 4). One woman reported, "Because they just don‟t have
any resources [laughs]. You know so I kind of knew that from the beginning so it‟s just a lost
cause." One reason was coded as other (2.5%). In six cases, the interviewer did not ask the
question (15%).
Social norms. H2 made predictions about the influence of (a) descriptive, (b) injunctive,
and (c) subjective norms on engaging in communication. Specifically, it was expected that norms
would predict a woman's engagement in communication. A binomial logistic regression was
conducted with the three norms types entered as the predictors. The model was not statistically
significant, χ2 (3) = 1.63, p = .65. Thus, H2 was not supported. See Table 7 for logistic regression
coefficient, Wald test, and odds-ratio for all predictors.2
Offender characteristics. Two hypotheses stating the relationship between the offenderspecific characteristics and engagement in communication were tested. H3 predicted that women
on parole would be more likely to engage in communication than women on probation, and H4
predicted that women would be more likely to engage in communication as their total risk score
(needs minus strengths) increased. A logistic regression was conducted with the total risk score
and status (parole as reference category) entered as the predictor variables. The overall model
was not statistically significant, χ2(2) = 2.95, p = .23. Thus, hypotheses H3 and H4 were not
supported. See Table 8 for logistic regression coefficient, Wald test, and odds-ration for all
predictors. A 2x2 chi-square test was conducted to examine the relationship between status
(probation/parole) and the engagement in communication (yes/no). The relationship was not
found to be statistically significant, χ2 (1, n=125) = 2.60, p = .11.

43

Social networks & social support. Two predictions were made about the women's social
network and the impact on whether or not they engaged in communication with their PO about a
difficult issue or need. H5 stated that women with smaller social networks would be more likely
to engage in communication. H6 predicted that the greater percentage of social support types
within their social network, the less likely they would engage in communication with their POs.
A binomial logistic regression was conducted with the three ratio support variables and the size
of the social network entered as predictors. The overall model was not statistically significant,
χ2(4) = .81, p = .94. Thus, the size of the social network and percentage of support within their
social network were not predictors of whether or not the women engaged in communication with
their POs about a difficult need or issue, and the data do not fit the hypotheses. See Table 9 for
logistic regression coefficient, Wald test, and odds-ratio for all social network and support
predictors.
Perceptions of PO characteristics and interactions. Finally, several predictions were
made regarding the women's perceptions of and reported results of interactions with their POs.
H7 predicted a positive association between perceptions of PO conversation orientation and
engaging in communication. H8 predicted a negative association between perceptions of PO
conformity orientation and engaging in communication. H9 predicted a negative relationship
between (a) psychological reactance, (b) emotional reactance, and (c) anxiety. A binomial
logistic regression was conducted with these five variables entered. The overall model was
statistically significant, χ2(5) = 12.94, p = .024. When looking at the individual predictors, none
were statistically significant at p <. 05.
One potential cause of no significant predictors when an overall model is statistically
significant is multicollinearity, which occurs when two or more predictor variables are correlated

44

with one another (Cohen, Cohen, West, & Aiken, 2003). A linear regression was conducted with
these predictors to determine whether multicollinearity was present. Three measures were used
to determine the existence of multicollinearity: the variance inflation factor (VIF), tolerance, and
condition number (Cohen et al., 2003). The VIF "provides an index of the amount that the
variance of each regression coefficient is increased relative to a situation in which all...predictor
variables are uncorrelated" (Cohen et al., 2003, p. 423). A common rule of thumb is that
predictors with a VIF greater than or equal to 10 indicates potential multicollinearity. The VIF
values for the predictors in the current model are 1.81 or smaller. Second, the tolerance for each
predictor was examined, which explains how much variance in each predictor is independent of
the other predictors (Cohen et al., 2003). Any value less than or equal to 0.10 indicates potential
multicollinearity problems. The tolerance values in this regression model are 0.55 or larger. A
third diagnostic test is examining the condition number, which involves a principle component
analysis, resulting in eigenvalues (kappa) for each predictor (Cohen et al., 2003). A general rule
of thumb is that values over 30 indicate serious problems of multicollinearity (Cohen et al.,
2003). κ's for this model are all less than 1. Therefore, it is concluded that multicollinearity is not
present.
Two indicators in the regression analysis trend towards statistical significance. Anxiety
( = -0.36, Wald‟s 2 = 6.08, p = .14) and psychological reactance ( = -0.36, Wald‟s 2 = 6.08,
p = .17) were both negatively associated with engaging in communication. This indicates that if
individuals reported feeling anxiety and psychological reactance during and after meetings with
their POs, then the odds of engaging in communication with their POs about the difficult need or
issue they were facing decreased. However, perceptions of the PO's communication style had no
impact on whether or not the women engaged in communication. Thus, H7 and H8 were not

45

supported, and H9 was trending towards support. See Table 10 for logistic regression coefficient,
Wald test, and odds-ratios for all predictors.
Post Hoc Analyses
Additional analyses were conducted to probe other potential predictors of engaging in
communication, as well as relationships among variables. First, a series of logistic regressions
were conducted to examine the impact of the criminogenic needs subscales on engagement in
communication. None of the risk subscales predicted engagement in communication, χ2 (12) =
5.55, p = .55. Additionally, none of the strengths subscales predicted engagement in
communication, χ2 (3) = 2.81, p = .42. Next, a chi-square test of association was conducted to see
if various levels of risk (low, moderate, high) were associated with engagement (yes/no) in
communication. Again, the result was not statistically significant, χ2 (2, n = 125) = 1.09, p = .58.
A logistic regression was also conducted to see if the women's personality type predicted
engagement in communication. Using five personality traits (extroversion, agreeable,
conscientious, neuroticism, openness) as the predictors, the result was not statistically
significant, χ2 (5) = 0.76, p = .98.
Next, analyses were conducted to see if the women's report of their POs' relationship
style predicted whether or not they engaged in the conversation. The two relationship styles
identified in this study from Skeem et al.'s (2007) dual-role relationship inventory are supportive
and punitive. Supportive POs are characterized as being fair/caring and trustworthy, while
punitive POs are characterized by toughness. A binomial logistic regression was conducted, and
the findings were statistically significant, χ2 (2) = 12.49, p = .002. Results indicate that women
who reported their POs as being more supportive were more likely to engage in the conversation.
Specifically, for every one-unit increase in ratings of PO supportive relationship style, the odds

46

of engaging in the conversation increase by 1.56,  = 0.45, Wald‟s 2 = 9.36, p = .002. See Table
11 for the logistic regression coefficient, Wald test, and odds ratio for both predictors.
Additional analyses were conducted to examine differences in women's ratings of various
scales in the study. A paired-samples t-test was performed to see if there was a statistically
significant difference in women's ratings of their POs' communication styles. The test revealed a
significant difference, t (126) = -6.01, p < .001, η2 = .22. Overall, ratings of PO conversation
orientation (M = 4.28, SD = 1.10) were significantly greater than ratings of PO conformity
orientation (M = 3.46, SD = 0.95).
The data were also probed to see if women's ratings of norms were significantly different
from one another. To examine this, a repeated measures ANOVA was conducted. The analysis
revealed a significant finding, F (1.75, 116) = 23.77, p < .001, partial η2 = .17. Multiple
comparisons using the Bonferroni method revealed that injunctive norms were rated highest (M
= 4.94, SD = 1.95), followed by subjective norms (M = 4.36, SD = 2.07) and descriptive norms
(M = 3.71, SD = 1.96). All norms were significantly different than the others, p < .05. Similar
trends were found for those who did not engage in the conversation, F (1.71, 37) = 7.38, p =
.001, partial η2 = .16, as well as those who did engage in the conversation, F (1.74, 78) = 22.06,
p < .001, partial η2 = .22. In all three tests, injunctive norms were rated significantly higher than
the other norms, p < .05, indicating that perceptions of approval about engaging in
communication were rated highest by all participants, regardless of whether or not the
conversation occurred.
Data were similarly examined to see if ratings of Dillard's measures of goals and
constraints (influence, interaction, personal resource, arousal management) were significantly
different. A repeated measures ANOVA was conducted, and the analysis did not reveal a

47

significant finding, F (2.66, 115) = 1.21, p =.31, partial η2 = .01. Thus, ratings of the four goal
types retained for analysis were not significantly different from one another.

48

CHAPTER 5: DISCUSSION
The purpose of this study was to examine factors that impacted whether or not female
probationers/parolees who wanted to talk to their POs about a difficult issue or need they were
facing actually did or did not engage in the conversation. To study this, data were collected
regarding communication norms and conversational goals. Additionally, secondary analyses
were conducted with measures already included in the larger dataset. This chapter will overview
the findings and practical implications of the results. Additionally, directions for future studies
and limitations of the current study are highlighted.
Overview of Findings & Practical Implications
Conversational goals. One of the biggest findings in this study was that personal
resource considerations were a significant negative predictor of engaging in the conversation.
Quantitative analyses demonstrated that female offenders who reported greater concerns for
losing personal resources, such as freedom, were less likely to engage in the conversation with
their PO, despite the conversation being desired. This finding is interesting and important both
theoretically and practically.
Many recent studies have not found personal resources to be a concern when deciding
whether or not to engage in difficult conversations (e.g., Cornacchione, 2013). Oftentimes,
relational resource considerations (concerns about maintaining the relationship) are rated as more
important. This is likely because many studies examining the multiple goals approach have
applied it to the family or other intimate relationships, which are different than an assigned,
social relationship (e.g., probationer and supervising officer). Schrader and Dillard (1998) noted
that personal resource considerations tend to be more important when 1) relational contexts were
distant and 2) stakes were high for the communicator, including considerable risk, high

49

importance, and complexity. For the women in this study, many of the conversational topics
reported were considerably risky, important, and complex. For example, someone who relapsed
(risk) may have wanted to talk to her PO to obtain treatment (importance). However, the
situation was complex because she both wanted to talk to her PO, but knew that there might be
legal repercussions. Thus, personal resource goals were of high importance, and the offenders
were less likely to have the conversation.
Research on superior-subordinate relationships within organizations has found that outgroup members tend to use more regulative tactics when communicating (Waldron, 1991). These
tactics include things such as distortion of information or avoidance, and are often used because
out-group members report low levels of trust (Waldron, 1991). It is possible that female
offenders perceived themselves as out-group members in reference to their supervising officers,
and, therefore, may have engaged in these regulative tactics because they did not trust their PO
to refrain from punishing them when talking about the difficult issue/need. There is evidence
here for the reverse in that when women perceived their PO as being trustworthy (i.e., supportive
relational style), they were more likely to have the conversation. With these results, POs could be
trained in how to interact with their clients to create perceptions of trustworthiness and being
supportive so that their clients are less likely to fear restrictions to personal freedom, and, thus,
more likely to talk to their POs about needs or issues they are facing. This may be of crucial
importance in preventing recidivism.
The findings from the qualitative data were similar to those found with the quantitative
analyses. The most frequently reported reason for not engaging in the desired conversation with
their POs was because the women feared that having the conversation could result in restricted
personal freedom, such as getting into more legal trouble and going back to prison. Several other

50

women reported issues with their POs as a constraint to having the conversation, such as a
negative relationship and communication issues (e.g., PO did not want to discuss the issue, or
usually rushed through conversations). Somewhat similar to the reasons reported for not having
the conversation, many women stated that the main reason they wanted to have the conversation
was to avoid trouble, such as notifying the PO about a criminal behavior before a third party
notified them. Overall, however, most women reported their primary goals as just wanting to
seek help and support, either through information acquisition, tangible assistance, or just wanting
help without specifying the type. Because of the generally positive perceptions women had of
their POs in this sample, it can be inferred that the women trusted their POs and viewed them as
a source of support for information regarding issues with which they are currently dealing.
Several individuals who did not engage in the conversation thought that, even if they did talk to
their POs, it would be pointless because the POs lacked the resources to help. Knowing that
seeking support and help is a common reason for talking to POs about difficult issues, it would
be useful to provide POs with the resources necessary to help women deal with the common
issues reported, or to communicate to the women that they do, in fact, have resources to help.
Social norms. The social norms approach is traditionally used in communication and
other social science research to examine behavior change. In this study, social norms were
examined to see how perceptions of personal-level norms impacted behavior regarding
engagement in communication. Social norms did not predict whether or not the women engaged
in desired communication with their PO. Although predictions were made based on previous
social norms research, there could be multiple explanations for this study's findings. One reason
norms may not have predicted whether or not communication occurred is that perceived norms
about prevalence of offenders communicating with their PO about a difficult need/issue do not

51

exist. Descriptive norms were the only norm type that was not significantly different than the
midpoint of the 7-point scale, which was "neither agree nor disagree". This could indicate the
non-existence of this behavior, and, therefore, the norm for how to behave. Individuals under
community supervision are not supposed to interact with other offenders, which might be a
possible explanation for the lack of descriptive norms; without interacting with similar
individuals, it would be hard to perceive the prevalence of a behavior. However, as seen in the
social network descriptive data, many of these women listed individuals in their network who
had committed an offense, including boyfriends, friends, and family members. Injunctive and
subjective norms, however, were rated as significantly higher than the midpoint of the scale.
Although the women may not have known how other women on probation and parole behaved
(descriptive norms), they may have perceived that engagement in the behavior may be expected
of, and approved for, them. However, it is also possible that these perceptions of approval and
expectations may not be based on other offenders, but instead other individuals within their
social network, such as family and friends. This makes sense given that probationers/parolees are
not supposed to associate with other offenders. Thus, it might be difficult to make inferences
about the socially approved/accepted behavior among offenders. As a result, they might make
inferences regarding what would be approved/accepted among non-offenders with whom the
women associate.
Despite the logistic regression results not predicting engagement in communication, an
ANOVA indicated that injunctive norms were rated statistically significantly higher than both
descriptive and subjective norms. Women seemed to be more concerned with perceptions of
approval rather than doing what they thought others wanted them to do (subjective), as well as
doing what they thought others actually were doing (descriptive). It is possible that these women,

52

as a result of breaking the law, have been facing disapproval for a long time, and, thus, try to do
what would gain approval in other situations in which they may be able to control. It might be
the case that social norms are not the most useful construct to study in relation to offenders for
various reasons already outlined, including that they have already intentionally violated societal
norms in committing the crime.
Similar to the ideas outlined in Daly and Bouhours (2008), it might be possible for
criminal justice officials, such as POs and judges, to communicate norms to offenders. This
could include not only norms about legal behavior, but also encouraging open and honest
communication with their POs. Additionally, other offenders, or those who have been released
from community supervision, could "teach" norms to women currently on community
supervision that could help them to engage in difficult conversations. This would also help
socialize women as they make the transition to community supervision. Research has
demonstrated that normative information can be internalized, even if coming from acquaintances
or strangers. In fact, normative information can be internalized without the presence of a
reference group (Lapinski & Rimal, 2005; Sherif, 1936).
There are also theoretical implications to these findings. The traditional social norms
approach for campaigns may not be as useful for marginalized populations, where campaigns
may sometimes be most critical. This is because marginalized populations may lack a social
network, resulting in a lack of perceived descriptive norms, as was found here. Many social
norms campaigns rest on the assumption of a discrepancy between descriptive and injunctive
norms (e.g., Michigan State University Social Norms Program Model). Thus, this lack of norms
within a population makes it difficult to aim a campaign at reducing this discrepancy. Social
norms approaches may not be the best approach for campaigns with marginalized populations.

53

This also highlights the importance of conducting formative research prior to creating and
disseminating any campaign or intervention (Atkin, 1994).
Offender characteristics. Neither the women's status (probation vs. parole) or their total
risk score impacted whether or not they engaged in desired communication with their POs. In the
quantitative analysis overview chapter, it was found that women with a greater needs score were
more likely to want to talk to their POs. Thus, the level of risk/needs might have more to do with
wanting to talk to their PO rather than with actual engagement. This is an important finding
because it indicates that women who have greater needs are more likely to want to talk to their
PO. Knowing that women with a greater level of needs want to talk to their POs can help
facilitate PO-initiation of these conversations. Offenders might lack communication efficacy,
which deals with the ability of the individual to "complete successfully the communication tasks
involved in the information-management process" (Afifi, 2009, p. 178). Therefore, it would be
important to train POs to bring up these common topics of conversation, many of which are
common needs that female offenders have, to help transfer the burden of information sharing
from solely the offender onto the PO to facilitate conversation about these issues or needs. POs
should be taught how to elicit greater detail regarding these issues rather than just asking generic
questions about specific topics. The topics reported in this study were abundant; however, many
of them fell within the criminogenic needs categories. This is promising given that many women
want to talk to their POs about issues that have been found to be predictive of subsequent
criminal behavior. Not only would it be important to facilitate POs' initiation of the conversation,
but to also make sure that they are connecting women to the resources they need. Previous
research with this sample has found that the POs are not doing an adequate job in connecting

54

their clients to needed services (Morash, Northcutt Bohmert, Kashy, Cobbina, & Smith, in
progress).
Social networks & social support. Contrary to predictions (H4, H5), characteristics of
the women's social networks did not predict engagement in communication. These findings are
not surprising given the perceived amount of support availability (emotional, instrumental,
network) reported by women in regard to their social networks. On average, women perceived
social support availability by 73% of network members (such as spending time together),
instrumental support by 76% (such as providing help with childcare), and emotional support by
92% (such as talking when upset/worried). Thus, women in this study might perceive their
quality of relationships to be positive, and this impact might be more important than network
size. It was expected that smaller social networks would lead to engagement in communication
with POs. However, it might be the case that perceived quality of the their social networks,
rather than the size, might be more important to women. Nevertheless, neither social support
availability or size predicted engagement in communication. One reason might be that there was
no measure of social support quality. Although the women perceive that support is available, it
might not actually occur, or if it does, it might not be good or useful support. Another
explanation for the null findings might be that the women have other individuals with whom they
can talk to about their difficult issues or needs, such as a case manager or therapist, rather than
their POs or other individuals within their social network.
The amount of perceived availability of support provided to women by their social
network members is not surprising given previous research on networks, support, and relational
maintenance. Within personal relationships, providing support is expected. A series of studies
conducted by Argyle and Henderson (1984) found that informal rules exist within friendships,

55

and that in order for relationships to be maintained, they must be rewarding. Common rules that
emerged in these studies included trust, confiding in the other, and helping in times of need. If
these rules are violated, then relationships are usually terminated. Some of the reasons reported
for deteriorating friendships is not providing support or volunteering help when needed (Argyle
& Henderson, 1984). Therefore, it would be expected that individuals within the women's social
networks would be providing some form of support, or be perceived to be able to provide it,
otherwise, the relationship would be terminated. Overall, it is interesting that social network
qualities did not impact women's decision to engage in communication with their POs about
difficult issues or needs they were facing. Another explanation for this finding is that, regardless
of size and quality of the network, women may think that their PO may be the only individuals
who could help them with the particular issue. Related to this, and indicating the benefits of
talking to POs about issues, research has found that when POs discuss more topics with their
clients, women experience greater feelings of efficacy related to finding work and avoiding
criminal activity (Morash, Kashy, Smith, & Cobbina, in progress).
Perceptions of PO characteristics & interactions. Several factors related to the
women's perceptions of their POs were also examined. Women's perceptions of their supervising
officers were entered into the regression model to determine if they predicted whether or not
engagement in desired communication occurred. The overall regression model was statistically
significant, indicating that women's perceptions of their POs impacted whether the conversation
occurred. It was discovered that the more anxious women felt during or after meeting with their
POs, the less likely they were to engage in the conversation. Similarly, the more psychological
reactance that was experienced, the less likely the women were to engage in the conversation.
However, the impact of these two predictors must be interpreted with caution as they were not

56

statistically significant, but only trended in this direction. One reason that women may feel
anxious while meeting with their POs is because they have a difficult conversation that they want
to engage in with their POs. It could also be the nature of the interactions (i.e., mandated
meetings) that induces anxiety, and not necessarily the PO's treatment of the women during their
meetings. Examining the mean scores of the perceived communication patterns illuminates this
possible explanation. The mean score of conformity orientation communication hovers around
the midpoint, while the mean score of conversation orientation communication is significantly
greater than the midpoint of the scale. This suggests that the women perceived their POs to
display open communication styles, yet women still reported experiencing anxiety.
Despite the high conversation and low conformity orientation reported, it is interesting
that psychological reactance was present, and served as a potential predictor of non-engagement
in communication. Although the questions were framed in a way to elicit state reactance, it is
possible that women offenders exhibited or reported on trait reactance. Thus, their reports of
reactance (albeit low) may be more a function of their personality rather than induced by their
POs.
Overall, the data indicated that women perceived their POs as being more conversational
in nature. However, it just might have been the case that the threat of a loss of freedom and
personal resources was much more important to the women, thus resulting in non-engagement in
desired communication regardless of perceptions of their PO's communication style.
A post hoc analysis was conducted to see if the women's perceptions of their PO's
relationship style impacted whether or not the conversation occurred. Perceiving their PO as
more supportive resulted in being more likely to engage in the conversation. Perceiving their PO
as tough did not impact whether or not the conversation occurred. This is interesting given that

57

perceptions of openness in terms of communication style did not predict engagement in
communication, but perceptions of fairness and caring in relating generally did impact whether
or not the women engaged in the conversation. It appears that relationship characteristics as a
whole were more important than just communication styles. In fact, within the larger data set, it
has been found that women who rate their POs as more supportive experienced less anxiety and
reported higher levels of self-efficacy (Morash et al., in progress). These findings further support
the argument that certain PO relationship styles impact probation/parole outcomes. Thus, it
would be useful to train supervising agents on how to have a more supportive relationship with
their clients so that the women engage in communication with their POs to address issues they
are facing, which, in turn, could result in more positive supervision outcomes for the women.
However, as previously discussed, there is a lot of variability in the mandated meeting
with POs. Additionally, POs serve as both a law enforcer and a counselor in their dual-role
relationship with their clients. Thus, for some women, it might be more useful to talk to their
POs about difficult needs or issues because the PO will be more supportive and helpful.
However, for other women, it might not be useful to talk to their POs about these issues because
it might not result in any assistance, or it could restrict the women's freedoms. Although the
latter example is a negative outcome for women, it might be actually be a positive outcome, and
is perceived to be one by the PO. Thus, engaging in communication with their POs might not
always be something that is perceived to be, or is actually, a useful thing to do. However,
developing meeting guidelines and training POs to manage their the dual-role relationships might
be one way to make meetings more beneficial. For instance, meetings should be comprised of
both the tough/firm/law enforcer aspect of the PO, such as making sure the women are not
breaking the law and following the conditions of their sentence, as well as the caring/supportive

58

aspect of the PO relationship style, such as making sure the women are receiving the resources
that are needed to help them navigate supervision and make positive progress.
Limitations and Directions for Future Research
This study has several limitations. First is the selection bias introduced, which occurred
because of the way in which the data were collected for the goals and norms measures. Women
self-selected into the sample by answering "yes" to whether there was ever a time when they
wanted to talk to their PO about a difficult issue or need. Thus, the findings of this study should
be interpreted with caution, and can only be generalized to women on probation and parole who
want to talk to their POs. Future research should examine all individuals to see what factors
impact their desire to talk to their POs, and compare outcomes of the two groups (those who
want to talk vs. those who do not).
Having a dichotomous dependent variable is also limiting in terms of data analysis.
Including a continuous measurement, such as intent to engage in communication, may have
provided further insight into the underlying processes occurring when offenders are deciding
whether to engage in a conversation with their PO. Additionally, adding a continuous measure
would have been a good way to examine not just the normative and goal impacts on intent, but
also whether intent lead to actual behavior. This could also facilitate longitudinal analysis of the
process, from intent to actual behavior to outcomes of having (or not having) the conversation.
Measurement was also an issue with this study. The scale reliabilities for goals items
were not as high as anticipated, and some subscales could not be used. The reliability of the four
scales that were retained, however, were relatively consistent with existing literature (e.g.,
Cornacchione, 2013; Dillard, 1989). Future research should explore ways to enhance the
measurement of conversational goals and constraints in deciding whether or not one engages in

59

communication. Additionally, measurement of goals and norms were retrospective, and the
actual processes and thoughts going on may be different than what was reported in this study.
Instead, the associations among variables are examined. Further studies should examine these
processes to determine causal order, which would provide further insight into the relationship
between norms and goals, and their subsequent impact on engaging in communication.
A similar measurement issue is that women were not asked to report on who their
normative referents were. An assumption was made that women would perceive themselves to be
in-group members with other female offenders. However, this may not be the case. Future
research in this area should explicitly ask participants who they look to when deciding how to
behave in various situations. Additionally, future research should inquire as to whether the
offenders have other individuals in their network, such as case managers and therapists, who
might serve as a more influential factor in whether or not communication with the PO occurs.
Additionally, interviewer errors were present throughout the difficult conversations
portion of the study. One problem was that not all of the questions were asked by the
interviewers that would have solicited more qualitative information from the women.
Additionally, questions were sometimes asked incorrectly, making the women's responses
unusable. Although data quality checks were conducted throughout data collection, it was not
possible to ensure 100% accuracy with all interviewers and interviews.
As previously mentioned throughout this section, an important area for future research is
to examine the outcomes of women who did vs. did not engage in the conversation with their
POs. Outcomes could include recidivism, satisfaction, reduction in risk score, and improvement
in strength score. It would be interesting to see if outcomes are better for women who actually
engaged in the conversation compared to the women who did not. Similarly, it would be

60

interesting to compare these same outcome variables for women who wanted to talk to their PO
compared to those who did not.
Theoretically, a critical next-step would be to examine the message structure, and see
how ratings of goals and norms impact the message sent. This would help in building a model to
understand individuals' motivations for engaging in communication by illuminating how goals
and norms shape message production.

61

CHAPTER 6: CONCLUSION
This research demonstrates an important step in understanding factors that impact
whether or not women on probation and parole, who have a number of needs and issues with
which to deal, talk to their POs about these issues. It largely addresses theoretical ideas within
communication (goals and social norms) while incorporating factors that have been documented
to impact women's probation/parole outcomes (e.g., criminogenic needs) to see how they
influence whether or not women who want to talk to their POs about difficult issues actually
engage in the conversation.
Community supervision is a major national investment in preventing recidivism
(Maruschak & Parks, 2012). Research shows that certain types of PO-client relationships
(marked by trust and caring, but not toughness) lead to better probation and parole outcomes
(Skeem et al., 2003). Additionally, a working alliance between offenders and supervising officers
also promotes positive outcomes (Taxman & Ainsworth, 2009). This study sought to address the
working alliance between offenders and supervising officers to determine how women are
potentially using their voices to address their needs and the factors that impact their ability to do
so. Findings point to the importance of personal resource considerations and women's
perceptions of their PO's relationship style in whether or not a desired conversation occurs. This
study extends previous research that has sought to develop a model to understand how
individuals decide whether to engage in communication about difficult or sensitive issues.

62

APPENDICES

63

APPENDIX A
Interview Items for Norms and Goals
diff_talk_1 (time 3 interview)
Was there a time when you wanted to talk to your current PO about a difficult issue? That issue
could include any needs you have or problems you‟ve faced since you‟ve been
supervised by this agent?
_____yes
_____no
If no: skips over this entire portion of the interview, and moves on to the next section
If yes: What was this difficult issue?
Did you actually talk to your PO about this issue?
______yes ________no
If yes, what did you say? What was driving you? In other words, what was your main
goal when trying to talk to your PO?
If yes, how did you start this conversation?
If no, why not? In other words, what was your main goal or concern when deciding not to
talk to your PO about this issue?

In a similar situation, would you talk to your PO about this or similar issues again? What was the
result of trying to talk to your PO?
How many other people who are important to you and in a similar situation would talk to their
POs about these issues? Would people who are important to you believe you should talk
to your PO about this issue?
Would people who are important to you encourage you to talk to your PO about this issue?

Ask following questions if they did or did not engage in the conversation with their PO
diff_talk_x
The next set of questions are about this time that you just described, when you talked to or
thought about talking to your PO about a difficult issue. You can either answer these on your
own on the computer or we can do it together.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

diff_talk_4

64

Strongly Agree

People whose opinions I value and who are in a situation similar to mine have talked with their
POs about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_5
The outcome of this conversation had important personal consequences for me.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_6
When talking [or thinking about talking], I was concerned about being true to myself and my
values.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_7
Most people who are important to me think that I should talk to my PO about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_8
I was concerned with making (or keeping) a good impression in this conversation.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_9
I was not willing to risk possible damage to my relationship with the agent to get what I wanted.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_10
In this situation, I wanted to avoid saying things that might have made me afraid or nervous.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_11
People who are important to me and in a situation similar to mine have talked with their POs
about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_12
In considering having this conversation, it was very important to me to convince the agent to do
what I wanted her or him to do.

65

Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_13
In considering having this conversation, I wanted to behave in a mature, responsible manner.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_14
Most people who are important to me approve of talking to my PO about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_15
Getting what I wanted was more important to me than preserving the relationship with the agent.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_16
In considering this conversation, I was concerned with not violating my own standards for what
is right and wrong.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_17
Most people whose opinion I value think I should talk to my PO about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_18
I was very concerned about getting what I wanted in this conversation.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_19
In considering this conversation, I was careful to avoid saying things which were socially
inappropriate.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_20
The agent could have made things very bad for me after I started the conversation.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_21
Most people whose opinions I value believe that I should talk to my PO about this issue.

66

Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_22
I was worried about the possibility this conversation would make me uncomfortable.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_23
I felt that having the conversation with the agent would be a waste of my time.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_24
My peers expect me to talk with my PO about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_25
I was concerned with putting myself in a “bad light” in this conversation.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_26
I didn‟t really care if I made the agent mad or not by having this conversation.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_27
Most people who are important to me support other women talking to their PO about this issue
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_28
I was afraid to have the conversation because it could take away my freedom or lead to more
supervision.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

Strongly Agree

diff_talk_29
I was afraid of being uncomfortable or nervous.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

diff_talk_30

67

Strongly Agree

People who are important to me and in a situation similar to mine engage in discussions with
their POs about this issue.
Strongly Disagree

(1) (2) (3) (4) (5) (6) (7)

68

Strongly Agree

APPENDIX B
Secondary Data Analysis Variable Items
Perceived PO Communication Style (asked during T2 interview)
Please use the following scale for each item, and circle the number that best represents your
answer. These questions are about how you typically interact with your PO. By PO we mean
your probation or parole officer.
1
Very
Strongly
Disagree

2
Strongly
Disagree

3
Disagree

4
Neither Agree
Nor Disagree

5
Agree

6
7
Strongly
Very
Agree
Strongly
Agree

My PO
commpat_1
…..and I often talk about things where she/he and I disagree.
commpat_2
…. makes it clear to me, “every person should have some say in making their own decisions.”
commpat_3
….. frequently asks me my opinion when we are talking.
commpat_4
….. encourages me to challenge her/his ideas and beliefs.
commpat_5
….. often tells me something like “you should always look at both sides of an issue.”
commpat_6
I usually tell my PO when I am thinking about things.
commpat_7
I can tell my PO almost anything.
commpat_8
With my PO, I often talk about my feelings and emotions.
commpat_9
When I see my PO, I often have long, relaxed conversations about nothing in particular.
commpat_10
I really enjoy talking with my PO, even when we disagree.
commpat_11
My PO likes hearing my opinions, even when she/he does not agree with my opinions.
commpat_12
… encourages me to express my feelings.
commpat_13
… is very open about her or his emotions when we interact.
commpat_14
Sometimes I talk to my PO about my day-to-day experiences more than once a week.
commpat_15
I talk to my PO about my plans and hopes for the future.

69

commpat_16
At times my PO says something like, “you should know better.”
commpat_17
Sometimes my PO says something like, “my ideas are right and you should not question them.”
commpat_18
Sometimes my PO says something like, “a person under supervision should not argue with the
PO or the judge.”
commpat_19
My PO often says something like, “there are some things that just shouldn‟t be talked about.”
commpat_20
One type of advice that my PO often gives me is something like, “you should give in on
arguments rather than risk making people mad.”
commpat_21
When anything really important is involved, my PO expects me to obey without question.
commpat_22
When I interact with my PO, she/he always ends the conversation with a final statement.
commpat_23
My PO thinks it is important to be seen as an authority in charge.
commpat_24
My PO becomes irritated with my views that are different from hers/his.
commpat_25
If my PO does not approve of something, she does not want to know about it from me.
commpat_26
My PO expects me to obey her/him.

Anxiety (T2)
During and after recent conversation or being with the PO, how much did you experience:
anx_1
Nervousness or shakiness inside
1
Not at all

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

anx_2
Feeling scared for no reason
1
Not at all
anx_3
Feeling fearful
1
Not at all

70

anx_4
Feeling tense or keyed up
1
Not at all

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

3
Somewhat

4
Quite a bit

5
Very much

anx_5
Panic or Terror
1
Not at all

anx_6
Feeling so restless you could not sit still
1
Not at all

2
Just a little

Psychological Reactance (T2)
How much do you disagree or agree with these statements?
Rating:
1
2
3
4
5
Very
Strongly
Disagree
Neither Agree
Agree
Strongly
Disagree
Nor Disagree
Disagree

6
7
Strongly
Very
Agree
Strongly

preact_1
1. It makes me angry when my PO points out something that I already know
preact_2
2. Suggestions and advice from my PO often make me do the opposite
preact_3
3. When my PO pushes me to do something, I often tell myself, “For sure I won‟t do it.”
preact_4
4. Often I lose enthusiasm for doing something just because my PO expects me to do it
preact_5
5. I get annoyed when my PO puts someone else up as an example to me
preact_6
6. I get very irritated when my PO tells me what I must or must not do
preact_7
7. When my PO gives me advice, I take it more as a demand

71

Emotional Reactance (T2)
During and just after talking to your probation or parole officer, how often are you feeling
remeas_1
Guilty
1
Not at all

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

2
Just a little

3
Somewhat

4
Quite a bit

5
Very much

remeas_2
Ashamed
1
Not at all
remeas_3
Annoyed
1
Not at all
remeas_4
Irritated
1
Not at all
remeas_5
Angry
1
Not at all

Needs Assessment (T1). For measures and information regarding the needs assessment, contact
Van Voorhis et al. (2008).

72

Social Network Size & Social Support (T1)
Expanded Questions:
When you are upset or worried about something, who would you talk with?
Who do you spend time with, hang out with?
Who would you ask if you needed something, like help doing something, money, or help caring for family members?
If you needed help with school work, or help with your job, who would you ask?
How frequently do you have contact? – yearly, a few times a year, monthly, weekly, daily, unknown
How important is this person to you?
What is the most serious offense committed by this person?
First
name
and
last
initial

Sex
M/F

Age

Relation

Upset /
worried
Who
Talk To

Who
Really
Cares
About
You

Spend
Time
With

Y/N

Y/N

Y/N
1.

2.
3.
4.
5.
6.
7.

73

Who Ask
if You
Needed
Something
Y/N

Who Ask
for Help
w/ work
or job
Y/N

How
Often In
Contact

Importance
1 not;
2 somewhat;
3 very

Most
serious
offense

APPENDIX C
Criminogenic Needs Subscales
Questions for criminal history aimed to understand the current conviction as well as prior
offenses and those committed after the current conviction.
Questions assessing educational needs included a number of dichotomous items (yes/no)
to understand current learning problems as well as educational levels. Questions assessing
financial/employment needs included a number of yes/no questions regarding their current
employment situation as well as their level of income and ability to pay bills. Questions
assessing housing safety included items such as the safety of the neighborhood in which the
woman lived and how safe she felt in her home with the people with whom she lived.
A number of questions were also asked to determine the woman‟s history of mental
illness as well as current symptoms. A series of yes/no questions were asked, and if the woman
responded yes, then she was asked if she experienced them recently. Questions were also asked
to determine anger/hostility.
A number of items were asked to assess whether women had experienced physical and/or
sexual abuse as a child and as an adult. Later in the interview, more specific abuse questions
were asked about their childhood and adulthood.
Whether a woman is currently using alcohol and/or drugs and how it impacts her life
was also assessed.
Several scales were used to examine the woman‟s relationship quality with romantic
partners, family of origin, and parenting/parental involvement. Questions assessing whether the
woman has antisocial friends were also asked to see if she interacts with individuals who commit
crimes. Finally, self-efficacy was also measured.

74

APPENDIX D
Tables
Table 1
Psychometric Properties of Continuous Measures
M

SD

Range



Descriptive Norms

3.74

1.96

1-7

.80

Subjective Norms

4.40

2.07

1-7

.83

Injunctive Norms

4.97

1.94

1-7

.84

Influence Goal

3.86

1.80

1-7

.64

Interaction Goal

3.79

1.72

1-7

.63

Personal Resource Goal

3.28

1.94

1-7

.73

Arousal Management Goal

3.64

2.02

1-7

.80

PO Conformity

3.46

0.95

1-7

.82

PO Conversation

4.28

1.10

1-7

.89

Anxiety

1.86

1.11

1-5

.94

Psychological Reactance

2.72

1.16

1-7

.89

Emotional Reactance

1.37

0.70

1-5

.85

Social Network Size

4.99*

2.69

1-15

Percent Instrumental Supp

0.76*

0.26

0-1

Percent Emotional Supp

0.92*

0.16

0-1

Percent Social Support

0.73*

0.27

0-1

Total Needs Score

21.42*

8.48

3-41

Variable

Note. Bolded values are significantly above or below the midpoint of the scale
*Not a scale, but a number reported by the women, and calculated by researcher

75

Table 2
Correlations Among Variables
Infl

Intr

PR

AM

DN

SN

IN

Risk

SNS

Inst

Emo

Soc

PO v

PO f

Anx

Psyc
hR

Em
oR

Infl
1
Intr
1
.43
PR
1
.28
.57
AM .46
.67
1
.64
DN
.18
-.07 -.01 .01
1
SN
.00
.02
.01
1
.22
.53
IN
.17
-.05 -.15 .05
1
.55
.77
Risk .11
.11
.13
-.02 -.03 -.03 1
.19
SNS .10
.14
-.11 .02
.16
.13
.17
.03
1
Instr .02
.04
.02
.05
.07
-.08 -.01 -.01 -.20 1
Emo -.10 -.23 -.05 -.03 .07
.08
.04
-.12 -.28 .28
1
Soc
-.04 -.00 -.13 -.11 .07
-.10 -.05 .19
-.14 .40
1
.22
PO v -.05 -.13 -.28 -.20 .14
-.16 .24
.06
-.05 .05 1
-.21 .23
PO f .11
-.03 .11
-.00 -.04 -.07 -.02 .08
.05 -.12
1
.21
.26
.23
Anx .10
-.12 -.08 -.14 .22
.06
-.13 .07
.08 -.41
1
.26
.28
.31
.39
Psyc .13
1
.19
.31
.20
-.21 -.15 -.25 .07
-.23 -.15 -.07 .17 -.46
.51
.40
hR
Emo .07
.17
-.04 -.09 .14
-.02 -.04 .08
-.01 -.42
1
.25
.20
.20
.24
.53
.42
R
Note. bold indicates p < .05
Infl = influence; Intr = interaction; PR = personal resource; AM = arousal management; DN = descriptive norm; SN = subjective
norms; IN = injunctive norm; Risk = criminogenic needs risk score; SNS= size of social network; Instr = instrumental support in
network; Emo = emotional support; Soc = social support; PO v = perceptions of PO conversation orientation; PO f = perceptions of
PO conformity orientation; Anx = anxiety; PsychR = psychological reactance; EmoR = emotional reactance

76

Table 3
Open-Ended Categories and Conceptual Definitions
Variable

Category

Conceptual Definition

Inform PO

Woman just generally informs PO about
the issue.
Woman explicitly states that she told her
PO because she did not want to get in
trouble.
Woman states that she PO asked her a
question about this issue or something
related to the issue.
Woman states that she wanted to come
clean and be honest with her PO. No
mention of woman wanting to tell her
PO before someone else. No reason
otherwise listed in this codebook
provided.
Woman states that she wanted the PO's
advice or information about how to go
about doing something.
Woman states that she wanted the PO to
get her something tangible, such as
sending her to rehab or counseling.
Woman states that she wanted PO to
help her, but isn't specific.
Woman states something about
expressing emotions.
Does not fit into other categories.
Interviewer did not ask the question.
Interviewer asked goal for the issue and
not the goal for having the conversation.

Main Goal_Talk

Inform PO to avoid
trouble/violation
Initiated by PO
Confession

Seek informational
support/advice
Seek tangible support
Seek help generally
Emotional expression
Other
Not asked by interviewer
Asked incorrectly
Constraints_NotTalk
Prevent restricted freedom

Woman states she did not want to get in
trouble, go to prison, rehab, etc.
Pointless [PO lacks
Woman states that the PO could not
resources]
have done anything to help. PO lacks
resources
Current PO Communication Woman states that the PO isn't returning
issues
her phone calls, communicating with the
woman, or would rush the conversation
if it actually occurred (too busy).

77

Table 3 (cont'd)
Negative relationship with
PO
Maintain positive
relationship with PO
Other
Interviewer didn't ask

Woman makes comment about the
nature of the relationship not facilitating
the conversation.
Woman mentions it could have hurt her
otherwise good relationship with her
PO.
Does not fit into any other category
already listed
Interviewer skips over questions--does
not ask

Topics/Issue
Housing issues
Paying back restitution fees
Finding a job/job issues
Restitution and job
Need support/financial
services
Transportation issues
Relapsing generally
Illegal activity--substances
Illegal activity--non
substances
Contact with police
Negative relationships/talk
about 3rd party
Abuse (sexual or physical
or both)
Associating with others on
probation/parole
Issues with children

78

Wants to talk about housing, such as
changing or not having housing
Woman owes money as part of her
sentence, and is having trouble
Woman is having trouble finding a job
Woman is having trouble paying back
her fees because she can't find a job
Woman states that she needs services,
such as financial help
Woman talks about problems with
driving, getting places
States that she relapsed, but doesn't
specify
She's using drugs or alcohol
She's engaging in illegal activity, but not
drugs/alcohol
Woman states that she had contact with
police
Woman talks about another individual
about something he/she did to the
woman (excluding abuse)
Woman states that she is being sexually,
physically abused
Woman states that she has or wants to
be able to spend time with family
members, romantic partners, or other
individuals who are on probation/parole
Woman is facing difficulties with
children, such as being able to see them
or custody issues

Table 3 (cont'd)
Relationship with PO

Woman wants to talk to PO about their
relationship
Woman wants to talk to her PO about
ending her probation/parole sentence
early
Woman wants to talk about things
related to her mental well-being
Woman wants to talk about things she
did in her past (i.e., years ago)
Does not fit into any other category
already listed
Interviewer didn't ask or asked
incorrectly
Question avoided and not actually
answered

End supervision early
Mental Health
Previous behavior
Other
Interviewer Error
Woman didn't answer

79

Table 4
Logistic Regression Predicting Communication
Predictors
Desc Norms
Subj Norms
Inj Norms
Influence Goal
Interaction Goal
Personal Res Goal
Arousal Mgt Goal
Status (dummy coded)
TotalNeedsScore
SocialNetwork Size
PercentInstru
PercentEmot
PercentSocial
Anxiety
Psych Reactance
Emot Reactance
PO Conversational
PO Conformity
Likelihood Ratio



SE

-0.04
-0.12
0.07
0.10
0.04
-0.31
-0.06
0.34
0.03
-0.03
-1.07
0.43
-0.48
-0.47
-0.28
0.28
0.01
-0.08

0.30
0.24
0.21
0.15
0.17
0.18
0.18
0.54
0.03
0.10
1.10
1.64
1.01
0.27
0.29
0.38
0.27
0.30

Wald‟s
2
0.06
0.32
0.12
0.40
0.06
2.90
0.11
0.40
0.70
0.08
0.96
0.07
0.22
3.00
0.88
0.53
0.002
0.08

120.57

80

E
(odds ratio)
0.96
0.89
1.07
1.10
1.04
0.74
0.94
1.41
1.03
0.97
0.34
1.54
0.62
0.63
0.76
1.32
1.01
0.92

Table 5
Topics and Engagement in Communication

Topic
Change housing
No housing
Housing generally
Relapsing
Illegal activity
Talk about 3rd party/rel
Abuse
Restitution
Job issues
Restitution & job
Police contact
End supervision early
Associate w/ offenders
Need services/transpo
Relationship w/PO
Previous behavior
Child custody issues
Other

Engaged in Communication
with PO
No
Yes
1
8
2
1
2
8
1
2
4
2
5
20
0
2
2
4
1
4
0
2
1
1
2
2
2
2
2
6
2
0
1
3
0
3
2
5

81

Total
9
3
10
3
6
25
2
6
5
2
2
4
4
8
2
4
3
7

Table 6
Logistic Regression of Goals Predicting Communication



SE

Influence Goal
Interaction Goal
Personal Res Goal
Arousal Mgt Goal

0.10
-0.03
-0.33
-0.09

0.12
0.14
0.15
0.16

Likelihood Ratio
*p< .05

134.83

Predictors

Wald‟s
2
0.71
0.04
4.88*
0.35

82

e
(odds ratio)
1.11
0.97
0.72
0.91

Table 7
Logistic Regression of Social Norms Predicting Communication
Predictors
Descriptive Norms
Subjective Norms
Injunctive Norms
Likelihood Ratio



SE

-0.04
-0.15
0.21

0.13
0.16
0.17

Wald‟s
2
0.08
0.94
1.54

134.83

83

e
(odds ratio)
0.96
0.86
1.23

Table 8
Logistic Regression of Risk Score and Parole Status Predicting Communication



SE

Total Needs Score
Parole

-0.15
-.68

0.02
0.41

Likelihood Ratio

155.24

Predictors

Wald‟s
2
0.41
2.75

84

e
(odds ratio)
0.99
1.98

Table 9
Logistic Regression of Social Network and Support Qualities Predicting Communication



SE

Social Network Size
Instrumental Support
Emotional Support
Social Support

0.02
-0.54
0.002
-0.14

0.08
0.89
1.40
0.82

Likelihood Ratio

150.53

Predictors

Wald‟s
2
0.08
0.36
0.00
0.03

85

e
(odds ratio)
1.02
0.59
1.00
0.87

Table 10
Logistic Regression of Perceptions of PO on Predicting Communication
Predictors
Anxiety
Psych Reactance
Emot Reactance
PO Conversational
PO Conformity
Likelihood Ratio (-2 Log L)
*p < .15



SE

-0.37
-0.33
0.29
0.08
-0.19

0.23
0.24
0.35
0.23
0.26

143.00

86

Wald‟s
2
2.71*
1.85
0.68
0.12
0.51

e
(odds ratio)
0.69
0.72
1.33
1.08
0.83

Table 11
Logistic Regression for Impact of Perceptions of PO Relationship Style on Communication
Predictors
PO Supportive
PO Punitive
Likelihood Ratio (-2 Log L)
*p < .01



SE

0.45
0.13

0.15
0.22

150.53

87

Wald‟s
2
9.36*
0.34

e
(odds ratio)
1.56
1.14

APPENDIX E
Footnotes
1

A procedure used for sample selection bias is the Heckman Two-Step Correction

(Heckman, 1979). When selection bias occurs, two models are present. The first is the selection
model, which is the gateway question or filter (skip pattern). In this study, the selection model
would use whether or not the women ever wanted to talk to their POs about a difficult issue or
need would be the dependent variable. The second model is the outcome model, which is a
subset of the model. In this study, the outcome model is whether or not the women actually
engaged in communication with their POs. The Heckman estimation attempts to explain how the
selection model impacts the findings of the outcome model. This is done by "estimation of a
probit model for selection, followed by the insertion of a correction factor...into the second
model of interest" (Bushway, Johnson, & Slocum, 2007, p. 152). Although commonly used, the
Heckman Correction has many drawbacks and can be quite problematic. One issue is that the
addition of the correction factor introduces collinearity into the model (e.g., Bushway et al.,
2007). Another problem is that the Heckman Correction inflates standard errors when the
covariates are identical in the selection and outcome model, as evidenced by Monte Carlo
simulations (e.g., Puhani, 2000), which would be the case in this study. Probably most important
for this study is that the Heckman Correction is used solely for a probit model first (for the
selection model) followed by a linear regression (for the outcome model) (Bushway et al., 2007;
Dubin & Rivers, 1990). Because the outcome variable of this study is dichotomous, the
Heckman correction cannot be used. Overall, even if some degree of sample selection bias is
present, Heckman procedures perform no better than uncorrected estimators and, if a sample

88

only has a few hundred cases, there is substantial risk that the Heckman procedures will make
the estimates worse (DeMaris, 2004; Stolzenberg & Relles, 1997).
2

Social network size was added as a control variable. Adding this control did not change

the findings, χ2 (4) = 1.66, p = .80. Perceived PO relationship style (supportive & punitive) was
added as a control variable. Adding this control changed the overall model, χ2 (5) = 11.94, p =
.04, but due to the effect of perceived supportive PO relationship style, p = .006.

89

REFERENCES

90

REFERENCES

Afifi, W. A. (2009). Uncertainty and information management in interpersonal contexts. In S.
W. Smith & S. Wilson (Eds.). New directions in interpersonal communication research
(pp. 94-114), Thousand Oaks, CA: Sage.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood Cliffs, NJ: Prentice-Hall.
Anderson, J., Cornacchione, J., & Maloney, E. K. (in press). Normative Beliefs and Social
Support in Weight Loss Communication. The International Journal of Communication
and Health.
Andrews, D. A., & Bonta, J. (2003). The psychology of criminal conduct (3rd ed.). Cincinnati,
OH: Anderson.
Argyle, M., & Henderson, M. (1984). The rules of friendship. Journal of Social and Personal
Relationships, 1, 211-237.
Asch, S. E. (1956). Studies of independence and conformity: I. A minority of one against a
unanimous majority. Psychological Monographs: General and Applied, 70, 1-70.
Atkin, C. K. (1994). Designing persuasive health messages. In L. Sechrest, T. E. Backer, E. M.
Rodgers, T. F. Campbell, & M. L. Grady (Eds.), Effective dissemination of clinical and
health information (pp. 99-109). Rockville, MD: U.S. Dept. of Health and Human
Services.
Barnes, M. K., & Duck, S. (1994). Everyday supportive contexts for social support. In B. R.
Burleson, T. L. Albrecht, & I. G. Sarason (Eds.) Communication of social support:
Messages, interactions, relationships, and community (pp. 175-194). Thousand Oaks,
CA: Sage.
Barsevick, A.M., Montgomery, S. V., Ruth, K., Ross, E. A., Egleston, B. L., Bingler, R., Malick,
J., … & Daly, M. B. (2008). Intention to communicate BRACA1/BRCA2 genetic test
results to the family. Journal of Family Psychology, 22, 303-312. doi:10.1037/08933200.22.2.303
Berger, C. R. (2005). Interpersonal communication: Theoretical perspectives, future prospects.
Journal of Communication, 55, 415-447. doi:10.1111/j.1460-2466.2005.tb02680.x
Bonczar, M. A., & Glaze, L. E. (2009). Probation and parole in the United States, 2008.
Washington, DC: Bureau of Justice Statistics.

91

Brann, M., & Sutton, M. L. (2009). The theory of planned behavior and college students‟
willingness to talk about smoking-related behaviors. Communication Research Reports,
26, 198-207.
Browning, D. M., Meyer, E. C., Truog, R. D., & Solomon, M. Z. (2007). Difficult conversations
in health care: Cultivating relational learning to address the hidden curriculum. Academic
Medicine, 82, 905-913.
Burleson, B. R., Albrecht, T. L., Goldsmith, D. J., & Sarason, I. G. (1994). Introduction: The
communication of social support. In B. R. Burleson, T. L. Albrecht, & I. G. Sarason
(Eds.), Communication of social support: Messages, interactions, relationships, and
community (pp. xi –xxx). Thousand Oaks, CA: Sage.
Bushway, S., Johnson, B. D., & Slocum, L. A. (2007). Is the magic still there? The use of the
Heckman two-step correction for selection bias in criminology. Journal of Quantitative
Criminology, 23, 151-178.
Caughlin, J. P. (2010). A multiple goals theory of personal relationships: Conceptual integration
and program overview. Journal of Social and Personal Relationships, 27, 824-848.
doi:10.1177/0265407510373262
Caughlin, J. P., Mikucki-Enyart, S. L., Middleton, A. V., Stone, A. M., & Brown, L. E. (2011).
Being open without talking about it: A rhetorical/normative approach to understanding
topic avoidance in families after a lung cancer diagnosis. Communication Monographs,
78, 409-436.
Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). A focus theory of normative conduct: A
theoretical refinement and reevaluation of the role of norms in human
behavior. Advances in Experimental Social Psychology, 24, 201-234.
doi:10.1016/S0065-2601(08)60330-5
Cobbina, J. E., Morash, M., Kashy, D., & Smith, S. W. (in press). Race, neighborhood danger,
and coping strategies among female probationers and parolees. Race and Justice.
Cochran, J. C. (2014). Breaches in the wall: Imprisonment, social support, and recidivism.
Journal of Research in Crime & Delinquency, 51, 2, 200-229.
doi:10.1177/0022427813497963
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation
analysis for the behavioral sciences (3rd Ed.). Mahwah, NJ: Lawrence Erlbaum.
Cornacchione, J. (2013). Normative influences on engaging in communication to persuade a
family member about a health behavior. Unpublished Preliminary Study, Michigan State
University.

92

Cullen, F. T. (1994). Social support as an organizing concept for criminology: Presidential
address to the Academy of Criminal Justice Sciences. Justice Quarterly, 11, 527-560.
Cutrona, C. E., & Suhr, J. A. (1992). Controllability of stressful events and satisfaction with
spouse support behaviors. Communication Research, 19, 154-176.
Daly, K., & Bouhours, B. (2008). Judicial censure and moral communication to youth sex
offenders. Justice Quarterly, 25, 496-522.
Demaris, A. (2004). Regression with social data: Modeling continuous and limited response
variables. Hoboken, NJ: Wiley.
Derogatis, L. R., & Melisaratos, N. (1983). The Brief Symptom Inventory: An introductory
report. Psychological Medicine, 13, 595-605.
DiClemente, R. J. (1991). Predictors of HIV-preventive sexual behavior in a high-risk adolescent
population: The influence of perceived peer norms and sexual communication on
incarcerated adolescents‟ consistent use of condoms. Journal of Adolescent Health, 12,
385-390.
Dillard, J. P. (1989). Types of influence goals in personal relationships. Journal of Social and
Personal Relationships, 6, 293-308. doi:10.1177/0265407589063004
Dillard, J. P. (1990). A goal-driven model of interpersonal influence. In J. P. Dillard (ed.),
Seeking compliance: The production of interpersonal influence messages (pp. 41-56).
Scottsdale, AZ: Gorsuch Scarisbrick.
Dillard, J. P. (2008). Goals-plans-action theory of message production: Making influence
messages. In L. A. Baxter & D. O. Braithwaite (Eds.), Engaging theories in interpersonal
communication: Multiple perspectives (pp. 65-76). Thousand Oaks, CA: Sage.
Dillard, J. P., & Shen, L. (2005). On the nature of reactance and its role in persuasive health
communication. Communication Monographs, 72, 144-168.
Dillard, J. P., Segrin, C., & Harden, J. M. (1989). Primary and secondary goals in the production
of interpersonal influence messages. Communication Monographs, 56, 19-38.
doi:10.1080/03637758909390247
Donovan-Kicken, E., & Caughlin, J. P. (2010). A multiple goals perspective on topic avoidance
and relationship satisfaction in the context of breast cancer. Communication Monographs,
77, 231-256. doi:10.1080/03637751003758219
Dowd, E. T., Milne, C. R., & Wise, S. L. (1991). The therapeutic reactance scale: A measure of
psychological reactance. Journal of Counseling & Development, 69, 541-545.

93

Dubin J. A., & Rivers, D. (1990) Selection bias in linear regression, logit and probit models. In J.
Fox & J. S. Long (Eds.), Modern methods of data analysis (pp. 411-443). Thousand
Oaks, CA: Sage Publications.
Elwood, W. N., Greene, K., & Carter, K. C. (2003). Gentlemen don‟t speak: Communication
norms and condom use in bathhouses. Journal of Applied Communication Research, 31,
277-297.
Estroff, S., & Zimmer, C. (1994). Social networks, social support, and violence among persons
with severe, persistent mental illness. In J. Monahan & H. Steadman (Eds.), Violence and
Mental Disorder: Developments in Risk Assessment (pp. 259-295). Chicago, IL:
University of Chicago Press.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Fitzpatrick, M. A., & Ritchie, L. D. (1994). Communication schemata within the family:
Multiple perspectives on family interaction. Human Communication Research, 20, 275301.
Freudenberg, N., Daniels, J., Crum, M., Perkins, T., & Richie, B. E. (2005). Coming home from
jail: The social and health consequences of community reentry for women, male
adolescents, and their families and communities. American Journal of Public Health, 95,
1725-1736. doi:10.2105/AJPH.2004.056325
Hall, J., & LaFrance, B. (2012). “That‟s gay”: Sexual prejudice, gender identity, norms, and
homophobic communication. Communication Quarterly, 60, 35-58.
Hart, R. A., & Clark, D. H. (1999). Does size matter? Exploring the small sample properties of
maximum likelihood estimation. Paper presented at the Annual Meeting of the Midwest
Political Science Association.
Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153-161.
Hochstetler, A., DeLisi, M., & Pratt, T. C. (2010). Social support and feelings of hostility among
released inmates. Crime & Delinquency, 56, 588-607. doi:10.1177/0011128708319926
Hu, L., & Bentler, P. M. (1999). Cutoff criteria indexes on covariance structure analysis:
Conventional criteria vs. new alternatives. Structural Equation Modeling, 6, 1-55.
Kallgren, C. A., Reno, R. R., & Cialdini, R. B. (2000). A focus theory of normative conduct:
When norms do and do not affect behavior. Personality & Social Psychology Bulletin,
26, 1002-1012.
Kam, J. A., Matsunaga, M., Hecht, M. L., & Ndiaye, K. (2009). Extending the theory of planned
behavior to predict alcohol, tobacco, and marijuana use among youth of Mexican
heritage. Prevention Science, 10, 41-53.

94

Keating, D. K., Russell, J., Cornacchione, J., & Smith, S. W. (2013). Family communication
patterns and difficult family conversations. Journal of Applied Communication Research,
41, 160-180.
Kennealy, P. J., Skeem, J. L., Manchak, S. M., & Eno Louden, J. (2012). Firm, fair, and caring
officer-offender relationships protect against supervision failure. Law and Human
Behavior, 36, 496-505.
Koerner, A. F., & Fitzpatrick, M. A. (2002a). Toward a theory of family communication.
Communication Theory, 12, 70-91. doi:10.1111/j.1468-2885.2002.tb00260.x
Koerner, A. F., & Fitzpatrick, M. A. (2002b). Understanding family communication patterns and
family functioning: The roles of conversation orientation and conformity orientation.
Communication Yearbook, 26, 37–69. doi:10.1207/s15567419cy2601_2
Korsgaard, M. A., & Roberson, L. (1995). Procedural justice in performance evaluation: The role
of instrumental and noninstrumental voice in performance appraisal discussions. Journal
of Management, 21, 657-670.
Kwon, K., & Lease, A. M. (2009). Children‟s social identification with a friendship group: A
moderating effect on intent to conform to norms. Small Group Research, 40, 694-719.
doi:10.1177/1046496409346578
Lapinski, M. K., & Rimal, R. N. (2005). An explication of social norms. Communication
Theory, 15, 127-147. doi:10.1093/ct/15.2.127
Lapinski, M. K., Anderson, J., Shugart, A., & Todd, E. (2013). Social influence in child care
centers: A test of the theory of normative social behavior. Health Communication, (ahead
of print), 1-14.
Leverentz, A. M. (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 & Delinquency, 43,
459-488.
Lind, E., Kanfer, R., & Earley, P. (1990). Voice, control and procedural justice: Instrumental and
non-instrumental concerns in fairness and judgments. Journal of Personality and Social
Psychology, 59, 952-959.
Maruschak, L. M., & Parks, E. (2012). Probation and parole in the United States, 2011
(Bulletin). Washington, D.C.: Bureau of Justice Statistics.
McCleod, J. M., & Chaffee, S. R. (1972). The social construction of reality. In J. Tedeschi (Ed.),
The social influence process (pp. 50-99). Chicago, IL: Aldine Atherton.
Miner, M.H., & Munns, R. (2005). Isolation and normlessness: Attitudinal comparisons of

95

adolescent sex offenders, juvenile offenders, and nondelinquents. International Journal of
Offender Therapy and Comparative Criminology, 49, 491-504.
Mollen, S., Rimal, R. N., & Lapinski, M. K. (2010). What is normative in Health
Communication research on norms? A review and recommendations for future
scholarship. Health Communication, 25, 544-547. doi:10.1080/10410236.2010.496704
Morash, M. (2010). Women on probation and parole. Lebanon, NH: Northeastern University
Press.
Morash, M., Kashy, D., Smith, S. W., & Cobbina, J. (in progress) The effects of supervising
agent relationship style. Working Paper from the Study of Women on Probation and
Parole.
Morash, M., Northcutt Bohmert, M., Kashy, D., Cobbina, J., & Smith, S. W. (in progress).
Women felons at the nexus of prison downsizing, social welfare reductions, and collateral
consequences of conviction.
Park, H. S., & Smith, S. W. (2007). Distinctiveness and influence of subjective norms, personal
descriptive and injunctive norms, and societal descriptive and injunctive norms on
behavioral intent: A case of two behaviors critical to organ donation. Human
Communication Research, 33, 194-218. doi:10.1111/j.1468-2958.2007.00296.x
Park, H. S., Klein, K. A., Smith. S. W., & Martell, D. (2009). Separating subjective norms,
university descriptive norms and injunctive norms, and U.S. descriptive and injunctive
norms for drinking behavior intentions. Health Communication, 24, 746-751.
doi:10.1080/10401230903265912
Perkins, H.W. & Berkowitz, A.D. (1986). Perceiving the community norms of alcohol use
among students: Some research implications for campus alcohol education
programming. International Journal of the Addictions, 21, 961-976.
Petersilia, J. (2001). Prisoner reentry: Public safety and reintegration challenges. The Prison
Journal, 81, 360-375.
Portes, A. (1998). Social capital: Its origins and applications in modern sociology. Annual
Review of Sociology, 24, 1-24.
Puhani, P. A. (2000). The Heckman correction for sample selection and its critique. Journal of
Economic Surveys, 14, 53-68.
Quick, B. L., & Stephenson, M. T. (2007). The reactance restoration scale (RSS): A measure of
direct and indirect restoration. Communication Research Reports, 24, 131-138.

96

Reisig, M. D., Holtfreter, K., & Morash, M. (2002). Social capital among women offenders:
Examining the distribution of social networks and resources. Journal of Contemporary
Criminal Justice, 18, 167-187. doi:10.1177/1043986202018002004
Reno, R. R., Cialdini, R. B., & Kallgren, C. A. (1993). The transsituational influence of social
norms. Journal of Personality and Social Psychology, 64, 104-112. doi:10.1037/00223514.64.1.104
Ritchie, L. D., & Fitzpatrick, M. A. (1990). Family communication patterns: Measuring
intrapersonal perceptions of interpersonal relationships. Communication Research, 17,
523-544.
Rimal, R. N., & Real, K. (2003). Understanding the influence of perceived norms on behaviors.
Communication Theory, 13, 184-203. doi:10.1111/j.1468-2885.2003.tb00288.x
Rimal, R., & Real, K. (2005). How behaviors are influenced by perceived norms.
Communication Research, 32, 389-414. doi:10.1177/0093650205275385
Ross, C. E.,&Mirowsky, J. (1987). Normlessness, powerlessness, and trouble with the law.
Criminology, 25, 257-278.
Ross, E. C., Polaschek, D. L. L., & Ward, T. (2008). The therapeutic alliance: A theoretical
revision for offender rehabilitation. Aggression and Violent Behavior, 13, 462-480.
doi:10.1016/j.avb.2008.07.003
Russell, J., Keating, D. M., Cornacchione, J., & Smith, S. W. (2012, November). An initial
investigation of difficult conversations in families from a multiple goals perspective.
Paper presented at the Annual Meeting of the National Communication Association,
Orlando, FL.
Salisbury, E. J., & Van Voorhis, P. (2009). A quantitative investigation of women probationers‟
paths to incarceration. Criminal Justice & Behavior, 36, 541-566.
doi:10.1177/00938454809334076
Salisbury, E. J., Van Voorhis, P., & Spiropoulos (2009). The predictive validity of a genderresponsive needs assessment: An exploratory study. Crime & Delinquency, 55, 550-585.
Scott, A. M. (2010). Family conversations about end-of-life health decisions (unpublished
doctoral dissertation). University of Illinois, Urbana-Champaign.
Schrader, D. C., & Dillard, J. P. (1998). Goal structures and interpersonal influence.
Communication Studies, 49, 276-293. doi:10.1080/10510979809368538
Schrodt, P., Witt, P. L., & Messersmith, A. S. (2008). A meta-analytical review of family
communication patterns and their associations with information processing, behavioral,

97

and psychosocial outcomes. Communication Monographs, 75, 248-269.
doi:10.1080/03637750802256318
Sherif, M. (1936). The psychology of social norms. New York, NY: Harper & Row.
Sherif, M., Harvey, J. O., White, B. J., Hood, W. R., & Sherif, C. W. (1961). The Robbers cave
experiment. Scranton, PA: Harper.
Skeem, J. L., Encandela, J., & Eno Louden, J. (2003). Perspectives on probation and mandated
mental health treatment in specialized and traditional probation departments. Behavioral
Sciences and the Law, 21, 429-458
Skeem, J., Eno Louden, J., Polaschek, D., & Camp, J. (2007). Assessing relationship quality in
mandated community treatment: Blending care and control. Psychological Assessment,
19, 397-410.
Skeem, J., Eno Louden, J., Manchak, S., Vidal, S., & Haddad, E. (2009). Social networks and
social control of probationers with co-occurring mental and substance abuse problems.
Law and Human Behavior, 33, 122-135. doi:10.1007/s10979-008-9140-1
Stolzenberg, R. M., & Relles, D. A. (1997). Tools for intuition about sample selection bias and
its correction. American Sociological Review, 62, 494-507.
Taxman, F. S., & Ainsworth, S. (2009). Correctional milieu: The key to quality outcomes.
Victims & Offenders, 4, 334-340. doi:10.1080/115564880903227347
United States Bureau of Justice Statistics (n.d.).Terms and definitions. Retrieved from
http://bjs.ojp.usdoj.gov/index.cfm?ty=tp&tid=15#terms_def
Van Dyne, L., Ang, S., & Botero, I. C. (2003). Conceptualizing employee silence and employee
voice as multidimensional constructs. Journal of Management Studies, 40(6), 1359-1392.
Van Voorhis, P., Salisbury, E., Wright, E., & Bauman, A. (2008). Achieving accurate pictures of
risk and identifying gender responsive needs: Two new assessments for women
offenders. Washington, DC: U.S. Department of Justice, National Institute of
Corrections. Retrieved from http://static.nicic.gov/Library/022844.pdf
Van Voorhis, P., Wright, E. M., Salisbury, E., & Bauman, A. (2010). Women‟s risk factors and
their contributions to existing risk/needs assessment: The current status of a genderresponsive supplement. Criminal Justice and Behavior, 37, 261-288.
Waldron, V. R. (1991). Achieving communication goals in superior-subordinate relationships:
The multi-functionality of upward maintenance tactics. Communication Monographs, 58,
289-306.
Wilson, S. R. (2007). Communication theory and the concept of “goal”. In B. B. Whaley & W.

98

Samter (Eds.) Explaining communication: Contemporary theories and exemplars, pp. 73104. Mahwah, NJ: Lawrence Erlbaum.
Wright, E. M., DeHart, D. D., Koons-Witt, B. A., & Crittenden, C. A. (2013). "Buffers" against
crime? Exploring the roles and limitations of positive relationships among women in
prison. Punishment & Society, 15, 71-95. doi:10.1177/1462474512466199
Yanovitzky, I., & Rimal, R. N. (2006). Communication and normative influence: An
introduction to the special issue. Communication Theory, 16, 1-6. doi:10.1111/j.14682885.2006.00002.x

99