Forgiveness, Mood, and Attachment Style

Home , Significant other

FORGIVENESS, MOOD, AND ATTACHMENT STYLE

Thesis

Submitted to

The College of Arts and Sciences of the

UNIVERSITY OF DAYTON

In Partial Fulfillment of the Requirements for

The Degree

Master of Arts in Clinical Psychology

Jennifer Ellen Farrell

UNIVERSITY OF DAYTON

Dayton, Ohio

August, 2010

FORGIVENESS, MOOD, AND ATTACHMENT STYLE

APPROVED BY:

______Mark S. Rye, Ph.D. Chairperson

______Catherine Lutz Zois, Ph.D. Committee Member

______Jack J. Bauer, Ph.D. Committee Member

Concurrence:

______Carolyn Roecker Phelps, Ph.D. Chair, Department of Psychology

Jennifer Ellen Farrell

2010

ABSTRACT

FORGIVENESS, MOOD, AND ATTACHMENT STYLE

Name: Farrell, Jennifer Ellen University of Dayton, 2010

Advisor: Dr. M. Rye

This study examined the role of attachment and mood when forgiving a romantic partner. Participants (N=164) were recruited from introductory psychology classes at a medium-sized Midwestern Catholic university and a Northeastern liberal arts college.

Participants completed self-report questionnaires assessing Forgiveness, Depression,

Attachment Style, Anger Rumination, and Forgiveness Perspectives. Consistent with hypotheses, Anger Rumination mediated the relationship between Forgiveness and

Depression, and the relationship between Emotional Benefits and Depression. This study also found that there were differences in several forgiveness measures across attachment styles. Compared to participants with fearful attachment, participants with a secure attachment were more likely to forgive a specific offender and less likely to view the offender’s behavior as being caused by personality defects. Participants with a secure attachment style were less likely to view the offense as having a major, negative impact on their lives, as compared to individuals with fearful attachment or preoccupied attachment. Contrary to hypothesis, dispositional forgiveness did not vary by attachment style. Implications and study limitations are discussed.

iii

ACKNOWLEDGEMENTS

My heartfelt thanks and appreciation go out to all of those individuals who assisted me in the process of completing my thesis. Dr. Rye spent hours helping to guide my research, editing drafts, and discussing ideas with me. Even after relocating to New

York, Dr. Rye has served as an exceptionally dedicated, enthusiastic, and supportive advisor. Thank you, Dr. Rye, for being so instrumental in helping me to complete this project—I truly could not have done it without all of your help! I would also like to thank Dr. Catherine Lutz Zois and Dr. Jack Bauer for their suggestions, feedback, and encouragement throughout this process.

Finally, I could not have completed this project without the support of my friends and family. I am endlessly appreciative of the time they spent motivating me and for their interest and questions about my work.

TABLE OF CONTENTS

ABSTRACT………………………………………………………………...……...... iii

ACKNOWLEDGMENTS…………………………………………………...…….…..iv

LIST OF TABLES…………………………………………………………………...... vii

CHAPTER

I. Introduction………………………………………………………….…...... …...1

Forgiveness Forgiveness and Depression Anger Rumination Adult Attachment Style and Forgiveness

II. Method………………………………………………………………...….…....23

Participants Measures Procedure

III. Results………………………………………………………………...….…….32

Preliminary Analyses Analyses of Major Study Questions

IV. Discussion………………………………………………………………….…..37

Major Study Questions Study Limitations Clinical and Research Implications

APPENDICES

Appendix A………………………………………………………...….…...... 48

Appendix B………………………………………………………….…...... ….49

Appendix C………………………………………………………………...….51

Appendix D………………………………………………………………....…53

Appendix E………………………………………………………………….…56

Appendix F……………………………………………………….……………58

Appendix G…………………………………………………….………...……60

Appendix H……………………………………………….....…………...... ….62

Appendix I………………………………………………….……………....….63

Appendix J……………………………………………….…………………….65

REFERENCES……………………………………………….………………...……...68

LIST OF TABLES

1. Demographic Characteristics/Relationship History of Participants…………………………..…………………..………...…………. 80

2. Means, Standard Deviations, and Cronbach Alphas for all Major Study Variables……………..………………...... …………...81

3. Contextual Variables Pertaining to Wrongdoing..…………………………………………………………..….. ….82

4. Correlations between Forgiveness and Forgiveness Perspectives…………..………………………………………………………..83

5. Correlations between Forgiveness Measures, Depression, and Anger Rumination……………………..……………………………………….……..84

6. Hierarchical Multiple Regression Examining the Prediction of Depression by Anger Rumination and Forgiveness…………………………...85

7. Hierarchical Multiple Regression Examining the Prediction of Depression by Forgiveness and Anger Rumination…………...…….……...... 86

8. Hierarchical Multiple Regression Examining the Prediction of Depression by Anger Rumination and Emotional Benefits………………...... 87

9. Hierarchical Multiple Regression Examining the Prediction of Depression by Emotional Benefits and Anger Rumination....……….……...... 88

10. Hierarchical Multiple Regression Examining the Prediction of Depression by Anger Rumination and Offense Impact..…………….……...... 89

11. ANOVA Results for Forgiveness Measures and Attachment Style..…………………………….……………...... 90

vii

CHAPTER I

INTRODUCTION

Involvement in a satisfying romantic relationship is among the most powerful

predictors of happiness and both mental and physical health (Cohen, Underwood, &

Gottlieb, 2000). However, conflicts, offenses, and transgressions are unavoidable in

close relationships, because no two individual’s interests, attitudes, and behaviors are

perfectly in synch all of the time. Eventually, many individuals feel frustrated, offended,

betrayed or wronged by a romantic partner. Conflicts and interpersonal transgressions

are a major source of negative feelings that have the potential to disrupt a relationship

(Fincham, 2000).

Common responses to a conflict or interpersonal transgression in a romantic

relationship include: avoiding the relationship partner, seeking revenge, wishing harm

upon the offender (McCullough, Worthington, & Rachal, 1997), or forgiving the

offender. Compared to other alternatives, forgiveness is associated with reduced negative

affect (Worthington & Scherer, 2004). Empirical studies have also shown that a failure

to forgive is related to mental health problems, such as depression (Maltby, Macaskill, &

Day, 2001).

Research is needed to better understand why forgiveness relates to depression.

There is evidence that anger rumination negatively relates to forgiveness (McCullough,

Bellah, Kilpatrick, & Johnson, 2001; Stoia-Caraballo et al., 2008) and positively relates to negative affect (Nolen-Hoeksema & Morrow, 1993; Stoia-Caraballo et al., 2008). It

may be that when an individual does not forgive, they are prone to increased anger

rumination. This rumination will stimulate memories and feelings associated with the

wrongdoing, and may result in the wrongdoing being relived over and over again.

Therefore, it is likely that rumination will contribute to depression. Given these

relationships, it is possible that anger rumination mediates the relationship between

forgiveness and depression.

Another variable relevant to romantic relationships that is related to forgiveness is

adult attachment style. Research has indicated that individuals who are securely attached

are more willing to remain in a relationship and work through conflicts, potentially

leading to forgiveness (Lawler-Row, Younger, Piferi, & Jones, 2006). Insecure attachment predicts more negative beliefs about love, less empathy, and a harsher attributional bias (Hazan & Shaver, 1987; Feeney, Noller, & Callan, 1994), all factors which make it less likely for an insecurely attached person to be forgiving (Burnette,

Taylor, Worthington, & Forsyth, 2007). However, research is needed to better

understand forgiveness perspectives that may account for the relationship between

attachment style and forgiveness. For instance, McCullough et al. (1997) point out that

forgiveness requires developing a more complex cognitive representation of one’s

partner—the offending partner may no longer be viewed as an idealized person, but

rather an individual who possesses both positive and negative capabilities (Flanigan,

1998). These complex representations may be more difficult for insecurely attached

individuals.

This study will address the following research questions: (1) Does anger

rumination serve as a mediator between forgiveness and depression? (2) Does attachment

style relate to both forgiveness and forgiveness perspectives? A review of the literature

will be organized in the following manner. First, a general conceptualization of

forgiveness will be discussed with a focus on common misconceptions of forgiveness.

The next section will consider how forgiveness of a transgression committed by a

romantic partner is related to depression. The third section will focus on how anger

rumination relates to forgiveness and depression. This section will also discuss research

on the role of anger rumination as a possible mediator between forgiveness and

depression. Next, an overview of attachment styles will be presented along with findings

relating attachment style to willingness to forgive a romantic partner. Finally, the present

study will be discussed.

Forgiveness

Types of Forgiveness

Forgiveness can be studied from multiple perspectives and the complexity of the

construct has been emphasized by researchers (Gorsuch & Hao, 1993). McCullough,

Hoyt, and Rachal (2000) developed a 3x2x4 taxonomy for categorizing measures of

forgiveness. The first dimension considers the following levels of specificity: offense-

specific, dyadic, and dispositional. Offense-specific measures are assessments of the extent to which a person has forgiven a specific offender for a specific offense. Dyadic measures represent an aggregate of the extent to which an individual forgives (or seeks forgiveness) in a single relationship across multiple offenses. Dispositional forgiveness represents a person’s tendency to grant (or seek) forgiveness across a variety of interpersonal offenses occurring in a variety of relationships. This study will focus on offense-specific forgiveness.

The second level of the taxonomy involves direction of forgiveness. Most

measures assess forgiveness in the direction of granting forgiveness (i.e. from the

perspective of the forgiver), while several other measures consider forgiveness in the

direction of seeking or accepting forgiveness (i.e., from the perspective of the transgressor). Feeling forgiven by God and forgiveness of self (intrapersonal forgiveness) are two other types of forgiveness that have been recognized by some researchers (Gorsuch & Hao, 1993; Maltby, Day & Barber, 2004). This study will examine forgiveness from the perspective of the person who was wronged.

The third dimension of the taxonomy deals with ways in which forgiveness is

assessed. Offense-specific forgiveness can be assessed through at least four methods.

First, using self-report measures, one can report the extent to which he or she has

forgiven the offending partner (or sought forgiveness from the offended partner).

Second, using partner-report methods, one can report the extent to which the offended

relationship partner has granted forgiveness (or the extent to which the offending partner

appears to feel forgiven). Third, an outside observer (such as a clinician) can assess the

extent to which a partner has forgiven the offending relationship partner (or has sought

the offended partner’s forgiveness). Fourth, measures of constructive or destructive

behaviors toward an offending relationship partner (which do not rely on verbal or

written reports) can be used to infer the extent to which an offended partner has forgiven

an offending partner (or the offender has sought forgiveness of the offended partner).

This study will assess forgiveness using self-report measures.

Definition of Interpersonal Forgiveness

The present study defines interpersonal forgiveness as an unjustly wronged person’s act of deliberately giving up resentment toward an offender while fostering the undeserved qualities of beneficence and compassion toward that offender (Enright & the

Human Development Study Group, 1991; North, 1987). This definition recognizes both the absence of negative emotions, motivations, and behaviors as well as the presence of positive emotions, motivations, and behaviors as important aspects of forgiveness. While most definitions recognize the need to let go of negative affect toward an offender, disagreement arises over whether or not it is necessary for positive affect to be present

(Rye et al., 2001). Worthington (2005) points out that most researchers who define forgiveness as a reduction in negative emotions, motivations, or behavior study

relationships between people who are strangers or others who do not expect their

relationship to continue. In contrast, in the case of ongoing relationships (such as

romantic partners), simply reducing negative emotions, motivations, or behavior is less

likely to produce forgiveness (Fincham, Beach, & Davila, 2004).

Baskin and Enright (2004) have distinguished between the forgiveness decision

and the forgiveness process. The forgiveness decision focuses the construct in the

cognitive domain and is reflected by the philosopher Neblett (1974) who said that the

essence of forgiveness is in the decision to forgive, along with the proclamation “I

forgive you” (as cited in Baskin & Enright, 2004). On the other hand, forgiveness as a

process views the construct as an extended cognitive and affective/empathy approach.

North (1987) stated that forgiveness is a process, with the defining-moment decision

embedded within it (as cited in Baskin & Enright, 2004).

What Forgiveness is Not

While there is some disagreement about the definition of forgiveness, there is

greater consensus regarding what forgiveness is not (McCullough, Pargament &

Thoresen, 2000). Many social scientists recognize that forgiveness is not the same as

reconciling. Reconciliation is a dyadic process with the goal of restoring a relationship

and establishing mutual trust (Brown, 2003). Forgiveness does not require the restoration of a relationship, especially when involvement in the relationship poses a risk of future harm. For instance, a victim of domestic violence may choose to forgive while refusing to maintain a relationship with a person who may cause future harm (Fincham, 2000).

Furthermore, forgiveness is not the same as pardoning, which can only be granted

by a representative of society, such as a judge (Enright & Fitzgibbons, 2002).

Forgiveness also differs from condoning, because condoning removes the offense and

hence makes forgiveness unnecessary (Freedman & Enright, 1996). In contrast,

forgiveness involves recognizing that an injury occurred. Forgiveness is also

distinguished from excusing, which involves a belief that no unfairness occurred. In

contrast, forgiveness allows for the recognition that an unfairness or injustice was

committed (Baskin & Enright, 2004).

Also, forgiveness is not the same as forgetting about the offense, which removes

awareness of the offense from consciousness. To forgive is more than never thinking

about the offense. Instead, forgiveness allows a victim to recall the transgression in a

way that minimizes distress (Enright, 2001). Forgiveness should be differentiated from

denial, which involves an unwillingness to perceive the injury. In order to forgive, there

must first be recognition that a hurtful act has occurred (Enright & the Human

Development Study Group, 1991). Finally, it is possible to forgive and still pursue justice through the legal system (Freedman & Enright, 1996).

Laypersons Conceptualization of Forgiveness

While the distinction between forgiveness and the concepts described above are widely accepted by social scientists, the layperson’s individual conceptualization of forgiveness may differ (Zechmeister & Romero, 2002). In order to determine how lay persons define forgiveness, Younger, Piferi, Jobe, and Lawler (2004) asked 196 undergraduate students as well as 83 community participants to define forgiveness.

Forty-two percent of undergraduates and twenty-nine percent of the community sample defined forgiveness as acceptance, dealing with the event, or getting over it, stressing the practical aspect of surviving the offense and continuing with life. Thirty-three percent of undergraduates and thirty-nine percent of the community sample defined forgiveness as letting go of negative feelings and grudges (i.e., focusing on the emotional component of forgiveness). Twenty-four percent (undergraduates) and sixteen percent (community sample) indicated that forgiveness meant continuing or resuming the relationship, emphasizing behavior change and reconciliation. The last two themes reflected opposing views of forgiveness: 10% (undergraduates) and 11% (community sample) claimed forgiveness meant forgetting about the event, while 6% (undergraduates) and 8%

(community sample) specifically mentioned that forgiveness does not mean forgetting.

A few differences emerged between community and undergraduate respondents’ definitions of forgiveness and professional researchers’ definition of forgiveness. Most notably, social scientists argue that forgiveness does not necessitate reconciliation, yet almost 25% of undergraduates (and 16% of the community sample) view forgiveness as a

process that involves reconciliation, placing forgiveness in the interpersonal domain.

This implies that, for many laypersons, forgiveness may be most relevant in the context

of a continuing relationship.

Forgiveness and Depression

The relationship between forgiveness and depression is an area of growing interest to many researchers. This relationship has been examined using both correlational and experimental designs.

For instance, Maltby et al. (2001) examined the relationship between un- forgiveness and mood. Three hundred and twenty-four undergraduate students participated in the study. Participants completed measures of forgiveness of self and others as well as a general health questionnaire containing a measure of depression.

Results showed that for men and women, failure to forgive others was positively related to depression.

Brown (2003) studied forgiveness at the level of global disposition, across situations and relationships. The tendency to forgive scale was designed as a brief, coherent measure of dispositional forgiveness to relate to depression. One aim of the study was to determine whether the tendency to forgive scale could predict mental health independently of attitudes about forgiveness and vengeance seeking constructs. The study included 70 undergraduate students from a Midwestern university. Data supported the distinctions among tendency to forgive, attitudes about forgiveness, and vengeance seeking constructs—although these constructs are related, they are not synonymous.

Results showed that scores on the tendency to forgive scale were negatively related to depression.

Another study which shows forgiveness resulting in a subsequent decrease in depression was conducted by Toussaint, Williams, Musick, and Everson-Rose (2008).

Their research examined the relationship between forgiveness, depression, and hopelessness. Participants responded to a telephone survey of adults age 18 and older, in an effort to obtain a large, representative sample. The survey employed a rotating panel designed to gather data from approximately 500 respondents per month. Each monthly sample consisted of about 300 new respondents and 200 respondents being re- interviewed six months after their initial interview. For five consecutive months, respondents were asked questions regarding depression, forgiveness, and hopelessness.

Analysis of results were based on 1,311 respondents with complete data and controlled for religiousness/spirituality and demographic factors. Models for depression showed that increased forgiveness of others leads to decreased depression. Overall, individuals who reported high levels of forgiveness of others also reported lower levels of hopelessness and had lower odds of being diagnosed as clinically depressed. This speaks to the importance of forgiveness of others in promoting good mental health and indicates that forgiveness may be an important predictor of depression. This study is particularly beneficial due to the fact that it is longitudinal in nature, i.e., a group of individuals were studied over a relatively long period of time. The study provides perspective on how forgiveness and depression may be related over an extended period.

Reed and Enright (2006) compared the effect of forgiveness therapy to an alternative treatment among a group of women who had experienced spousal emotional abuse but had been permanently separated from the abusive partner for at least two years.

Forgiveness therapy aimed to help the women relinquish resentment and revenge and

develop goodwill. Forgiveness therapy did not encourage reconciliation in any way.

Alternative treatment, on the other hand, was a more standard therapeutic procedure that

did not directly target the amelioration of resentment. The forgiveness therapy was more

effective in reducing depression for these women. This study is also important in

pointing out that forgiveness must be clearly distinguished from reconciliation in

instances when remaining in a romantic relationship would be emotionally or physically

harmful for an individual.

Similarly, Rye et al. (2005) examined how a forgiveness group intervention

affects adult adjustment to divorce. Divorced participants (N=149), ranging in age from

23 to 73 years old, were recruited from the community. Each participant was assigned to

either a secular or religiously-based intervention, or a comparison condition.

Interventions consisted of 8 weekly sessions lasting 90 minutes per session. Results

indicated that participants in both the secular and religious interventions improved

significantly more on forgiveness of an ex-spouse than comparison participants (who did

not participate in any intervention during the course of the study). The secular

intervention participants showed a larger decrease in depressive symptoms over time than

comparison participants. Interestingly, no treatment effects for depression were found for

the religious intervention.

A study by Rye, Folck, Heim, Olszewski, and Traina (2004) involved recruiting

individuals from community singles organizations and church-based recovery groups to determine the relationship between forgiveness of one’s ex-spouse and mental health.

Significant correlations between forgiveness and several measures of mental health were found, after controlling for the effects of demographic/background variables. However,

11 these results were mixed. Forgiveness (Absence of Negative) was negatively correlated with depression. It was hypothesized that perhaps forgiveness decreases rumination and provides individuals with a new way to think about their circumstances, resulting in an improvement in overall mood. Yet, forgiveness (Presence of Positive) did not relate significantly to depression.

Rye and Pargament (2002) investigated the effect of three conditions: secular forgiveness intervention, religiously integrated forgiveness intervention, and no- intervention comparison on a sample of Christian women from a medium-sized

Midwestern state university who had experienced a variety of types of wrongdoing during the course of a romantic relationship. Women in the secular and religiously integrated forgiveness groups each participated in six sessions of group therapy. In the religiously integrated group, participants were encouraged to draw upon their Christian beliefs in achieving forgiveness. Seventy percent of the secular participants and 68% of the religiously integrated participants volunteered that the group helped them to overcome anger and/or helped them to forgive their offender. In both interventions, participants improved on forgiveness and existential well-being, however, no treatment effects were found with respect to measures of depression. Despite the unexpected results of this study, many other studies have found forgiveness and depression to be negatively related.

As noted above, several studies using both correlational and experimental designs have shown that forgiveness is inversely related to depression. However, little is known about mechanisms that explain the relationship. This study will explore the possibility that anger rumination mediates the relationship between forgiveness and depression. For

instance, when wronged, people may tend to think about the wrongdoing a lot, and may

even enjoy ruminating on the transgression. A mediator is a variable that “accounts for

the relation between a predictor and criterion” (Baron & Kenny, 1986, p. 1176). In a

mediation model, the predictor variable has an indirect relationship with the criterion

variable through a third variable (mediator). Variables that are possible candidates for

mediators are related to both the independent variable and the dependent variable (Baron

& Kenny, 1986).

Anger Rumination

Definition of Anger Rumination

Recently, researchers have begun to look at the constructs of anger and

rumination together. A large scale, factor-analytical investigation provided support for the independence and possibly complex structure of the anger rumination factor (Miller,

Jenkins, Kaplan, & Salonen, 1995). While anger is viewed as an emotion, anger rumination is defined as thinking about this emotion. Anger rumination can be defined as

“unintentional and recurrent cognitive processes that emerge during and continue after an episode of anger experience” (Sukhodolsky, Golub, & Cromwell, 2001, p. 690).

Research described below has found that anger rumination relates to both forgiveness and depression.

Anger Rumination and Forgiveness

Barber, Maltby and Macaskill (2005) examined the relationship between forgiveness of others and anger rumination, utilizing the Sukhodolsky et al. (2001) Anger

Rumination Scale. Two hundred undergraduate students were given questionnaires relevant to these variables. The Anger Rumination Scale was used to explore which

13 dimensions of anger rumination best predict scores in forgiveness of self and others. The four dimensions of the Anger Rumination Scale include: anger afterthoughts, angry memories, thoughts of revenge, and understanding of causes. Results of the study showed the subscales of the ARS share a significant negative correlation with forgiveness of others. Thoughts of revenge were found to uniquely predict forgiveness of others.

Anger Rumination and Depression

Research also has found a positive relationship between depression and anger rumination. For instance, a study by Gilbert, Cheung, Irons and McEwan (2005) investigated the relationship of anger-focused rumination to depression. Participants included 166 undergraduate students (138 females, 28 males) and utilized the ARS developed by Sukhodolsky et al. (2001). The study found that all forms of anger rumination (angry memories, understanding of causes, angry afterthoughts, and thoughts of revenge) were significantly associated with depression, apart from thoughts of revenge. Thoughts of revenge were not correlated with depression, possibly because revenge rumination is empowering in some way. Interestingly, it appears that thoughts of revenge may prevent forgiveness of others, but these thoughts do not seem to result in depressive symptoms.

Anger Rumination as a Mediator between Forgiveness and Depression

Stoia-Caraballo et al. (2008) hypothesized that anger rumination mediates the relationship between forgiveness and sleep quality. Researchers recruited 277 undergraduate students and had them fill out a variety of measures, including: forgiveness, sleep quality, negative affect, and anger rumination. Results showed that in

14 one mediating pathway, forgiveness was related to less anger rumination (and less negative affect), which in turn related to improved sleep quality.

Rye et al. (2008) showed that a similar relationship exists between forgiveness, anger rumination, and depression. The study looked at anger rumination and gratitude as mediators of the relationship between forgiveness and depression. Participants, who consisted of 196 undergraduate students, were asked to describe an experience in which they had been wronged and a time in which they had wronged another person.

Participants were also asked to fill out questionnaires related to forgiveness, depression, anger rumination, and gratitude. Anger rumination was found to mediate one indirect pathway from forgiveness to depression. Specifically, increased forgiveness predicted lower anger rumination which in turn predicted lower depression.

In theory, an individual who chooses not to forgive may become extremely focused on the wrongdoing and replay the event over in their mind multiple times. It may be that this anger rumination helps the individual justify their decision not to forgive. The negative feelings and memories stimulated by anger rumination are likely to contribute to feelings of depression. Although the research cited above points to anger rumination as a possible mediator between forgiveness and depression, this needs to be tested in the context of romantic relationships. Research is also needed concerning the possible role of attachment style when forgiving a romantic partner.

Adult Attachment Style and Forgiveness

Definition of Adult Attachment Style

When researchers discuss attachment style, they are referring to the propensity of individuals to form affectional bonds to particular others (Bowlby, 1980). Attachment

style is a systematic pattern of relational expectations, emotions, and behaviors that

results from internalization of a particular history of attachment experiences and working

models of interpersonal relationships (Fraley & Shaver, 2000). An individual’s

attachment style is a powerful predictor of emotion-regulation and reactions to relationship conflict (Eisenberg & Fabes, 1992; Mikulincer & Shaver, 2005).

The presence of attachment bonds can be inferred based on the extent to which a romantic partner, sibling, parent, or friend serves in any of several known attachment functions: proximity seeking (i.e., behaviors designed to bring the self nearer to the attachment figure), separation distress (i.e., negative affect exhibited when the attachment figure is unavailable), safe haven (i.e., desire to seek out the attachment figure for comfort and support), and secure base (i.e., use of the attachment figure as a base from which one can explore the world). (Tancredy & Fraley, 2006). Three classes of situations tend to activate the attachment system during a relationship (Kobak & Duemmler, 1994): fear-provoking situations, challenging situations, and conflictual interactions. The present study is particularly interested in how attachment style relates to forgiveness of a romantic partner following a conflict/transgression.

Attachment and intimacy are believed by many scholars to play a relatively meager role in comparison with sexual desire in the early stages of romantic relationships

(Hazan & Shaver, 1994). This is mainly due to the finding that, on average, it takes two years for a full-fledged attachment bond to develop (Fraley & Davis, 1997; Hazan &

Ziefman, 1994). However, Hazan and Ziefman (1994) did find that proximity-seeking and separation distress functions in particular were pronounced among romantic partners well before couples reached the two-year mark, suggesting that fledgling relationships are

16 attachment-relevant contexts. After all, bonds between romantic partners have to start somewhere, and therefore attachment concerns may frequently be salient during a relationship’s early stages (Hazan & Shaver, 1994).

Eastwick and Finkel (2008) propose that a) attachment concerns may be salient in relationships that are still developing and b) attachment anxiety is not only a dispositional orientation but also a state-like, normative, functional experience in establishing romantic relationships. In other words, attachment anxiety may motivate approach and other attachment behaviors in efforts to establish a romantic relationship. It was concluded from data collected in a series of four studies that the experience of attachment anxiety regarding romantic partners appears to be pronounced in the very early stages of a romantic relationship, before that relationship is “official.” Across studies, the predictive effects of partner-specific attachment anxiety were robust above and beyond the effect of sexual desire. The authors point out that if sexual attraction were the sole motivational force underlying romantic relationship initiation, it is unclear a priori why losing an opportunity for a new partner would be so excruciating. Losing a potential attachment figure is certainly more likely to evoke protest and despair than is losing a potential sexual partner. This suggests that the attachment system may be activated fairly quickly in a relationship, even if complete attachment requires approximately two years.

Models of Attachment

Several different methods of conceptualizing attachment have been proposed.

One of these is a three category model, developed by Ainsworth, Blehar, Waters, and

Wall (1978). This tripartite typology consists of secure, avoidant, and anxious- ambivalent attachment styles. Secure attachment is defined by confidence in the

availability of attachment figures in times of need and by comfort with closeness and

interdependence. Avoidant is characterized by insecurity concerning others’ intentions

and preference for emotional distance. Finally, Anxious-ambivalent attachment involves insecurity concerning others’ responses combined with a strong desire for intimacy and high fear of rejection.

Bartholomew and Horowitz (1991) developed a four category model of adult attachment based on two dimensions: model of self and model of others. Secure attachment reflects a positive model of self and others; it is characterized by freely experiencing both intimacy and autonomy in relationships. Preoccupied attachment involves a negative model of self and a positive model of others. This attachment style is distinguished by feelings of unworthiness and an exaggerated concern with seeking approval from others. The fearful attachment style results from a negative model of self and others; it is characterized by feeling unworthy, fearing rejection from others, and consequently avoiding interpersonal closeness. The fourth attachment category, dismissing, is the result of a positive model of self and negative model of others. An individual with a dismissing attachment style does not feel unworthy, but also does not value interpersonal closeness.

Attachment Style and Forgiveness of a Romantic Partner

A number of studies have supported a relationship between attachment style and willingness to forgive a romantic partner following a transgression. For instance, Lawler-

Row et al. (2006) used the four-category model to examine the role of attachment style and forgiveness of a transgressor within a close relationship. Results showed that securely attached young adults reported more trait and greater state forgiveness in

response to two discrete betrayals than insecurely attached young adults. Insecurely

attached young adults reported a greater desire to avoid the offender after the event.

Another study using the four-category attachment style approach considered three types of forgiveness: forgiveness of self, forgiveness of others, and forgiveness of a

situation (Webb, Call, Chickering, Colburn, & Heisler 2006). Subscale scores were

combined in order to derive a total forgiveness score. Those participants reporting a

secure attachment style produced significantly higher scores for self, situation, other, and

total forgiveness compared to all insecure attachment styles. No insecure attachment

style (fearful, preoccupied, dismissing) differed significantly from any other insecure

attachment category in forgiveness type.

Similarly, using a four-category model of attachment, a study found that

individuals in dating relationships who had a more positive model of self and a more

positive model of their partner had a greater tendency to forgive partner transgressions

(Kachadourian, Fincham, & Davila, 2004). For individuals with a more negative model

of self, there was no relationship between model of others and the tendency to forgive

partner transgressions. Thus, it appears that only those individuals who are securely

attached (have both a positive model of self and of their partner) are more likely to

forgive when their partners commit transgressions.

A study using the three-category attachment model recruited 140 Israeli

undergraduate participants (Mikulincer, Shaver, & Slav, 2006). Results showed that both

major forms of attachment insecurity (anxiety and avoidance) are related to reductions in

or distortions of forgiveness. Avoidant attachment is related to lower levels of

forgiveness. Anxious attachment, on the other hand, does not necessarily eliminate

forgiveness, but it apparently combines forgiveness with incompatible feelings.

Studies have also shown that attachment is related to forgiveness perspectives.

For instance, a study by Mikulincer (1998) examined attachment-style differences in attribution of hostile intent during anger-eliciting events. Participants, previously classified as anxious, avoidant, or secure, were exposed to hypothetical anger-eliciting scenarios differing in the intentions of a romantic partner (hostile, ambiguous, or nonhostile). Results indicated that proneness to attribute hostile intent was most accentuated among individuals with avoidant attachment, whose hostile attributional bias was observed across all conditions (hostile, ambiguous, nonhostile). Anxiously-attached people tended to attribute more hostile intent in both the hostile and ambiguous conditions while securely attached individuals attributed hostile intent only when there were clear contextual cues (in the hostile condition).

A study by Simpson, Rholes, and Phillips (1996) examined how people with different attachment orientations behave toward and perceive their romantic partners after conflict resolution. The attachment styles were broken down into: anxious-ambivalent, avoidant, and secure. Dating partners discussed and tried to resolve either a major or minor problem in their relationship. Individuals who were more anxiously attached reacted less positively toward their partners, especially when trying to resolve a major conflict. Highly anxiously attached individuals who discussed a major problem displayed greater stress and anxiety during their interactions and reported greater anger and hostility toward their partners. Unlike anxiously attached people, highly avoidant individuals neither reported nor displayed greater distress, they did not report feeling angry, and they

20 did not view their partner or relationship less positively following the discussion.

However, more avoidant men behaved in a less warm and supportive manner toward their partner; their discussions were also rated as poorer in quality. This study indicates that following relationship conflict, anxiously attached people are more likely than secure and avoidant individuals to display anxiety and stress.

Insecurely attached individuals may judge the impact of an offense to be more severe than those that are securely attached. Studies have shown that insecurely attached individuals, as compared to securely attached individuals, have difficulty with emotion regulation, especially anger, in times of threat (i.e., relationship conflict) (Mikulincer,

1998). In fact, insecurely attached individuals display greater depressive and anger rumination than secure individuals (Burnette et al., 2007, Mikulincer, 1998). Sroufe

(1984) emphasizes that securely attached individuals are able to tolerate increasing levels of tension, while maintaining organized behaviors in the face of such tension. Thus, when evaluating the impact of an offense, securely attached individuals may be less likely to perceive the offense as disrupting their emotional state.

Securely attached individuals may have more positive beliefs about forgiveness than those that are insecurely attached due to general differences on optimism. For instance, a study by Heinonen, Räikkönen, Keltikangas-Järrinen, and Strandberg (2004) found that insecurely attached individuals were more likely to hold a pessimistic life orientation, while securely attached individuals were more likely to be optimistic. In fact, attachment-related variables accounted for 48% of variance in dispositional optimism- pessimism. Further, Brissette, Scheier, and Carver (2002) found that pessimistic individuals have less adaptive coping strategies and may have a diminished ability to

21 generate supportive social networks, as compared to optimistic individuals. Additional research is needed to examine the relationship between attachment style and forgiveness perspectives.

Present Study

The present study examined the role of anger rumination, attachment, and mood when forgiving a romantic partner. Two questions were addressed. First, does anger rumination mediate between forgiveness and depression? It was hypothesized that increased forgiveness would predict decreased anger rumination, which in turn would predict decreased depression. Second, does attachment style relate to forgiveness and forgiveness perspectives? It was hypothesized that securely attached individuals would be more likely to forgive a romantic partner following an offense and would exhibit greater willingness to forgive across situations. It was also hypothesized that perspectives on forgiveness would vary depending on attachment style. Specifically, securely attached individuals were hypothesized to be more likely to view: 1) the offender as worthy of forgiveness, 2) forgiveness as a positive and beneficial process, 3) the offense as having a less important impact on their life, 4) the offender as unintentionally causing harm, and 5) the external factors related to the offender’s behavior as being relevant (i.e. the offender was going through a difficult time). 6) It was also hypothesized that securely attached victims would be more open to unconditional forgiveness. In contrast, insecurely attached individuals were hypothesized to 1) be more likely to believe that the offender does not deserve forgiveness, 2) view forgiveness as not beneficial, 3) judge the offense as having a significant and negative impact on their lives, 4) demonize the offender and view the offender’s behavior as malicious, and 5) be

22 less likely to attribute the wrongdoing to external factors (i.e. focus on factors internal to the offender), 6) forgive only under certain conditions (i.e. the offender apologizes).

CHAPTER II

METHOD

Participants

Participants consisted of 164 undergraduate psychology students from a medium-

sized Midwestern Catholic university (n=123) and a small, liberal arts college in the

Northeast (n=41) (see Table 1). Thirty-seven participants were eliminated from the study because they indicated they had never been wronged by a romantic partner or they did not select ‘romantic partner’ as the relationship they had to the person who mistreated them.

The final sample consisted of more female participants (67.7%) than male participants (32.3%). The majority of participants were Caucasian (84.7%). Other races represented in the sample included: African American (4.9%), Latino (3.7%), Asian

American/Pacific Islander (2.5%), and “other” (4.3%). Participants ages ranged from 18 to 22 (M=18.89, SD=.93). Most participants were in their first (53.7%) or sophomore

(36.0%) year of college, although there were also some participants in their junior (7.3%) or senior (3.0%) year of college.

Participants completed several questions pertaining to relationship history (see

Table 1). Almost half of participants indicated that they were currently in a romantic relationship (49.4%). Participants were also asked to select one of four paragraphs, each describing a different attachment style. Participants most commonly endorsed a secure

attachment style (40.9%). The remaining attachment styles included: fearful (29.9%),

preoccupied (16.5%), and dismissing (12.8%).

Measures

Participants completed several self-report questionnaires including measures of demographic/background information, relationship history, forgiveness (Forgiveness

Scale and Forgiveness Likelihood Scale), depression (CES-D scale), anger rumination

(Anger Rumination Scale), forgiveness perspectives (Offender Unworthy, Contingent

Forgiveness, Emotional Benefits, Offense Impact, Internal Attributions, External

Attributions) and adult attachment style (Relationship Questionnaire). These measures are described below.

Demographic/Background Information

A questionnaire evaluating demographic information was completed by each of the participants. Questions included items relating to age, gender, race, and educational level (Appendix A). Participants were asked if they are currently in a romantic relationship or have ever been involved in a romantic relationship. In addition, they completed several questions concerning the nature of a wrongdoing they experienced while in a romantic relationship (Appendix B).

Forgiveness

Forgiveness Scale. The Forgiveness Scale was used to measure forgiveness toward a particular offender (Rye et al., 2001; Appendix C). Participants were instructed to think about how they have responded to the person who wronged them. Participants rated their level of agreement on 15 Likert-type items, with response possibilities ranging from 1 (strongly disagree) to 5 (strongly agree). Factor analysis revealed the presence of

two subscales (Rye et al., 2001): Absence of Negative and Presence of Positive. Sample

items include: “I can’t stop thinking about how I was wronged by this person” and “I

wish for good things to happen to the person who wronged me.” For purposes of this

study, items were combined to form a single forgiveness score. Scores on the total scale

can range from 15 to 75, with higher scores reflecting higher levels of forgiveness.

Cronbach’s alpha was adequate for the total scale (.87) and the test-retest reliability for the total scale over an average of 15.2 days was .80.

Forgiveness Likelihood Scale. The Forgiveness Likelihood Scale examines the general tendency or willingness of a person to forgive an offender (Rye et al., 2001;

Appendix D). The scale consists of ten items that describe hypothetical acts of wrongdoing. Respondents were asked to imagine that the wrongdoing occurred to them and then consider the likelihood that they would be willing to forgive the offender. The items use a Likert-type format, with response possibilities ranging from 1 (not at all

likely) to 5 (extremely likely). Sample items include, “Your significant other has just

broken up with you, leaving you hurt and confused. You learn that the reason for the

break up is that your significant other started dating a good friend of yours. What is the

likelihood that you would choose to forgive your significant other?” and “Your friend

has been talking about you behind your back. When you confront this person, he/she

denies it, even though you know that he/she is lying. What is the likelihood that you

would choose to forgive your friend?”

A factor analysis demonstrated that a one-solution factor was the most appropriate

(Rye et al., 2001). Cronbach’s alpha was .85 and the test-retest reliability over an

average of 15.2 days was .81. This scale was significantly correlated with the Enright

Forgiveness Inventory (r=.25, p<.001), and with a global forgiveness item from the

Enright Forgiveness Inventory (r=.23, p<.001). The Forgiveness Likelihood Scale was also significantly correlated with other constructs including trait anger (r= -.31, p<.001), religious well-being (r=.23, p<.001), and social desirability (r=.17, p<.01). Scores on this scale range from 10 to 50 with higher scores reflecting an increased willingness to forgive.

Depression

CES-D. Depression was measured with the Center for Epidemiologic Studies

Depression scale (Radloff, 1977; Appendix E). This scale consists of 20 self-rated

Likert-type items, with response possibilities ranging from 1 (Rarely or none of the time) to 4 (Most or all of the time). The scale contains items pertaining to depressed mood, feelings of guilt and worthlessness, feelings of helplessness and hopelessness, psychomotor retardation, loss of appetite, and sleep disturbance. To assess current state, participants were asked how often they experienced symptoms in the past week. The possible range of scores is 20 to 80, with higher scores indicating more symptoms of depression. Sample items include “I felt lonely,” and “I felt that people dislike me.”

Internal consistency was demonstrated by an average coefficient alpha of .86 and an average Spearman-Brown of .88 across four different field tests. Test-retest correlations were in the moderate range, with all but one falling between .45 and .70.

These correlations were, on average, larger for the shorter time intervals. In all the samples, the pattern of correlations of the CES-D with other scales gives reasonable evidence of discriminant validity. The highest r’s were with scales designed to measure symptoms of depression: Lubin, r=.37 to .70, Bradburn Negative Affect, r=.55 to .63, and

Bradburn Balance, r=.61 to .72. Correlations of the CES-D with scales designed to

measure different variables were as follows: Medications, r=.20 to .24, Disability Days,

r=.28 to .32, and Social Functioning, r=.13 to .24.

Anger Rumination

Anger Rumination Scale. Anger rumination was measured using the Anger

Rumination Scale (ARS) (Sukhodolsky et al., 2001; Appendix F). This scale measures

the tendency to think about current anger-provoking situations and to recall anger episodes from the past. It consists of 19 Likert-type items, with response possibilities ranging from 1 (Almost Never) to 4 (Almost Always). Factor analyses revealed the presence of four factors: angry afterthoughts (cognitive rehearsal of recent anger episodes), thoughts of revenge (thoughts and ideas of retribution), angry memories

(thoughts about anger episodes from the past), and understanding of causes (thinking about the causes of an anger event). However, for purposes of this study, a single index score was computed. Sample items include: “Whenever I experience anger, I keep thinking about it for awhile” and “I have difficulty forgiving people who have hurt me.”

Cronbach’s alpha for the total ARS scale was .93. A test-retest reliability coefficient of .77 was obtained (N=179) over a one month period. Correlation coefficients were computed between ARS and other measures. The correlations of the total ARS score with the subscales of the STAXI, a 44-item questionnaire extensively used in research on anger, were moderate and in the predicted direction: ARS and trait- anger (r=.57), ARS and anger-in (r=.52), ARS and anger-out (r=.43), ARS and anger control (r=-.35). The ARS was also significantly related to the Measure of Negative

Affectivity (r=.54). Moderate and significant correlations were also found between the

ARS and the TMMS mood repair subscale (r=-.44) and TMMS mood clarity subscale

(r=-.25). Negative relationships were demonstrated between the ARS and the measures of life satisfaction and social desirability. Scores on the Anger Rumination Scale can range from 19 to 76, with higher scores indicating greater levels of anger rumination.

Forgiveness Perspectives

Items for the Forgiveness Perspectives scale were originally developed by Matros

(2009; Appendix G). A factor analysis revealed the presence of six subscales (Matros,

2009): Offender Unworthy (4 items), Contingent Forgiveness (4 items), Emotional

Benefits (4 items), Offense Impact (6 items), Internal Attributions (7 items), and External

Attributions (3 items). Participants were asked to rate each item on a 5-point Likert

scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

The first subscale, Offender Unworthy, measures to what extent the victim

believes that the offender does not deserve forgiveness. A sample item is: “Individuals

who have wronged me are not worthy of forgiveness.” Overall scores on this subscale

ranged from 4 to 20, with higher scores indicating that the offender is viewed by the

victim as less worthy of forgiveness. The second subscale, Contingent Forgiveness,

measures the extent to which the victim only forgives if the offender meets certain

conditions. A sample item is: “I would consider forgiveness only if the offender

apologized.” Overall scores on this subscale ranged from 4 to 20, with higher scores

indicating that forgiveness should be conditional (i.e. based on the offender making

amends, apologizing, taking responsibility, etc.). The third subscale, Emotional Benefits,

measures an individual’s belief that forgiveness has positive benefits. A sample item is:

“Forgiving someone makes me feel better about myself.” Overall scores on this subscale

ranged from 4 to 20, with higher scores indicating a greater likelihood of feeling better

after choosing to forgive. The fourth subscale, Offense Impact, measures how

individuals believe the offense will impact their lives. A sample item is: “The

wrongdoing I experienced is going to cause me significant hardship.” Overall scores on

this subscale ranged from 6 to 30, with higher scores on these items indicating a greater

perception of a negative, versus a neutral or positive, impact from the offense. The fifth

subscale, Internal Attributions, measures to what extent the victim views the offender’s

behavior as being caused by personality defects. A sample item is: “This person wronged

me because they are a fundamentally bad person.” Overall scores on this subscale ranged

from 7 to 35, with higher scores indicating a more negative view of the offender, versus

viewing the offender as unintentionally causing harm. The sixth subscale, External

Attributions, measures to what extent the victim views the wrongdoing as influenced by

factors external to the offender. A sample item is: “This person would never have hurt

me if they had not been going through a difficult time when the offense took place.”

Overall scores on this subscale ranged from 3 to 15, with higher scores indicating a

greater emphasis on the influence of external factors.

Adult Attachment Style

Relationship Questionnaire. Adult attachment style, based on the four-category model, was measured using the Relationship Questionnaire (Bartholomew & Horowitz,

1991; Appendix H). This measure consists of four short paragraphs describing secure, dismissing, fearful, and preoccupied attachment styles. Each participant was asked to select the paragraph that best describes how they generally feel in romantic relationships.

The secure style related to fearful (r=-.32), preoccupied (r=-.40), and dismissing

(r=-.18). The fearful style related to dismissing (r=-.17) and the preoccupied style related to dismissing (r=-.21). Both components of the measure, the Likert scale and the four paragraphs, were found to have moderate test-retest reliability and good construct validity. The correspondence of categorical attachment style classifications over a one to two month period was 76%. Additionally, in factor analyses and structural equation analyses, the self-report measure, Bartholomew’s interview measure, and peer descriptions of individuals’ attachment patterns all converge (Levy, Blatt, & Shaver,

1998).

Procedure

Undergraduate students were recruited from Introduction to Psychology classes at a medium sized Midwestern Catholic university and a small Northeastern liberal arts college. Both males and females who have been involved in a romantic relationship and were at least 18 years of age were included in the sample. The researcher administered the questionnaires to groups of approximately 25 to 30 students. Questionnaires were counter-balanced with two different orders and participants were randomly assigned to complete one of the two versions. The researcher explained the instructions and confidentiality prior to distributing the questionnaires, and was available throughout the session to answer any questions. An informed consent form (Appendix L) was provided to explain that participation in the study was voluntary and could be withdrawn at any time. Participants signed their names on the informed consent form indicating their willingness to participate. Confidentiality was maintained by requesting that participants refrain from putting their names on the questionnaires. After completion of the session,

31 the participants were given a debriefing form (Appendix M) and received one experimental credit for their participation.

CHAPTER III

RESULTS

Preliminary Analyses

Means, standard deviations, and Cronbach alphas were computed for all major study variables (see Table 2). Cronbach alphas were acceptable (range=.71-.93) for forgiveness, mental health, and forgiveness perspectives measures.

Participants completed several questions pertaining to the context of a wrongdoing they experienced (see Table 3). The nature of the offenses reported were as follows: being lied to (81.7%), not following through on obligations (75.0%), being cheated on (49.4%), being called names or unkind words (44.5%), gossiped about

(30.5%), physically harmed (12.2%), being stolen from (6.7%), and other offenses

(34.1%). The percentages for these offenses add up to more than 100% because participants often reported more than one type of offense.

For continuous variables pertaining to wrongdoing, frequencies were computed

(see Table 3). The amount of time since the offense occurred ranged from .04 to 7 years

(M=1.48, SD=1.29). On a scale from 1 to 4, with 1 indicating a very low level and 4 indicating a very high level, participants provided ratings for offense severity (M=2.89,

SD=.86), degree of hurt now (M=2.13, SD=.90), and degree of anger now (M=1.94,

SD=.80).

Correlations were computed between Forgiveness measures (i.e. Forgiveness

Scale and Forgiveness Likelihood Scale) and Forgiveness Perspectives (i.e. Offender

Unworthy, Contingent Forgiveness, Emotional Benefits, Offense Impact, Internal

Attributions, and External Attributions) (see Table 4). Forgiveness was positively

correlated with Forgiveness Likelihood (r= .31, p<.01) and Emotional Benefits (r= .32, p<.01). Forgiveness was negatively correlated with the following: Offender Unworthy

(r= -.35, p<.01), Contingent Forgiveness (r= -.21, p<.01), Offense Impact (r= -.47, p<.01), and Internal Attributions (r= -.44, p<.01). Forgiveness Likelihood was positively correlated with Emotional Benefits (r= .34, p<.01) and External Attributions (r= .17,

p<.05) and negatively correlated with Offender Unworthy (r= -.44, p<.01), Contingent

Forgiveness (r= -.47, p<.01), and Internal Attributions (r= -.17, p<.05). Offender

Unworthiness was positively correlated with Contingent Forgiveness (r= .40, p<.01) and

negatively correlated with Emotional Benefits (r= -.45, p<.01). Contingent Forgiveness

was also negatively correlated with Emotional Benefits (r= -.31, p<.01). Offense Impact

was positively correlated with Internal Attributions (r= .32, p<.01). Finally, Internal

Attributions was negatively correlated with External Attributions (r= -.22, p<.01).

Analyses of Major Study Questions

According to Baron and Kenny (1986), the first step in testing mediation is to

examine the relationship between the independent variable (i.e., Forgiveness), and the

dependent variable (i.e., Depression) (see Table 5). Depression was negatively correlated

with Forgiveness (r= -.26, p<.01). Additional correlations were computed to examine

whether forgiveness measures (i.e. Forgiveness Scale, Forgiveness Likelihood Scale,

Offender Unworthy, Contingent Forgiveness, Emotional Benefits, Offense Impact,

Internal Attributions, and External Attributions) were related to the mental health

variables (i.e. Depression and Anger Rumination). Along with being negatively

correlated with Forgiveness, Depression was positively correlated with Offense Impact

(r= .36, p<.01) and negatively correlated with Emotional Benefits (r= -.23, p<.01).

Anger Rumination was positively correlated with Offender Unworthy (r=.35, p<.01),

Contingent Forgiveness (r= .27, p<.01), and Offense Impact (r= .36, p<.01). Anger

Rumination was also negatively correlated with the following: Forgiveness (r= -.39,

p<.01), Forgiveness Likelihood (r= -.28, p<.01), and Emotional Benefits (r= -.27, p<.01).

Following the model presented by Baron and Kenny (1986), the second step in

testing for mediation is to examine the relationship between the independent variable

(i.e., Forgiveness) and the mediator (i.e., Anger Rumination) (see Table 5). Anger

Rumination was negatively correlated with Forgiveness (r= -.39, p<.01).

According to Baron and Kenny (1986), the third step is to evaluate the relationship between the potential mediator (i.e., Anger Rumination) and the dependent variable (i.e., Depression) while controlling for the independent variable (i.e.,

Forgiveness). Thus, a multiple regression was computed in which Depression was entered as the independent variable and Anger Rumination and Forgiveness were entered as predictors. Anger Rumination significantly predicted Depression (B = .45, p < .001)

with Forgiveness in the equation.

According to Baron and Kenny (1986), if each of the previous three conditions

are met, the final step is to determine whether the IV (i.e., Forgiveness) predicts the DV

(i.e., Depression) after controlling for the mediator (i.e., Anger Rumination). Several

hierarchical multiple regressions were computed to examine the role of Anger

Rumination as a mediator. These multiple regressions utilized the Forgiveness measures

that were significantly correlated with both Anger Rumination and Depression (i.e.,

Forgiveness, Emotional Benefits, and Offense Impact). In all multiple regression

equations, Depression was entered as the dependent variable.

In the first set of equations, Anger Rumination was entered in step 1, and

Forgiveness was entered in step 2 (see Table 6). After controlling for the Anger

Rumination variable, the relationship between Forgiveness and Depression was not

significant (R2∆=.01, p>.05). A Sobel test was computed for this equation (Sobel= -4.03, p<.0001). The reverse was also computed, with Forgiveness entered in step 1 and Anger

Rumination entered in step 2 (see Table 7). After controlling for the Forgiveness variable, the relationship between Anger Rumination and Depression was significant

(R2∆=.17, p<.001).

In the second set of equations, Anger Rumination was entered in step 1, and

Emotional Benefits was entered in step 2 (see Table 8). After controlling for the Anger

Rumination variable, the relationship between Emotional Benefits and Depression was

not significant (R2∆=.01, p>.05). A Sobel test was computed for this equation (Sobel= -

3.14, p<.01). The reverse was also computed, with Emotional Benefits entered in step 1

and Anger Rumination entered in step 2 (see Table 9). After controlling for the

Emotional Benefits variable, the relationship between Anger Rumination and Depression

was significant (R2∆=.19, p<.001).

Finally, in the third equation, Anger Rumination was entered in step 1, and

Offense Impact was entered in step 2 (see Table 10). After controlling for the Anger

Rumination variable, the relationship between Offense Impact and Depression remained significant (R2∆=.04, p<.01).

One-way analyses of variance (ANOVA) were computed to test differences in forgiveness measures (i.e. Forgiveness Scale, Forgiveness Likelihood Scale, Offender

Unworthy, Contingent Forgiveness, Emotional Benefits, Offense Impact, Internal

Attributions, and External Attributions) across attachment styles (i.e. Relationship

Questionnaire) (see Table 11). Forgiveness of a specific offender differed significantly across attachment styles, F(3, 160) = 4.73, p<.01. Bonferroni post hoc analyses revealed that participants with secure attachment (M=56.75, SD=8.60) scored higher than participants with fearful attachment (M=50.14, SD=11.63). Offense Impact also differed significantly across attachment styles, F(3, 160) = 6.42, p<.001. In this case, participants with secure attachment (M=13.87, SD=4.05) scored lower than participants with fearful attachment (M=16.86, SD=4.61) and preoccupied attachment (M=16.85, SD=4.09).

Internal Attributions differed significantly across attachment styles, F(3, 160) = 3.54, p<.05. In this case, participants with secure attachment (M=17.15, SD=5.79) scored lower than participants with fearful attachment (M=20.71, SD=5.76). Contrary to hypotheses, no significant differences were found between attachment styles for:

Forgiveness Likelihood, Offender Unworthy, Contingent Forgiveness, Emotional

Benefits, and External Attributions.

CHAPTER IV

DISCUSSION

Major Study Questions

Does anger rumination serve as a mediator between forgiveness and depression?

As expected, this study found that forgiveness was inversely related to depression.

A number of studies have similarly shown that forgiveness is related to decreased levels of depression and failure to forgive is positively related to depression (Brown, 2003;

Maltby et al., 2001; Touissant et al., 2008). A number of intervention outcome studies have found that forgiveness can lead to decreases in depression (Ingersoll-Dayton,

Campbell, & Ha, 2009; Reed & Enright, 2006; Rye et al., 2005).

As expected, this study also found that forgiveness is inversely related to anger rumination. Previous studies also indicate that forgiveness and anger rumination are related. For instance, Barber et al. (2005) found that anger rumination was significantly negatively correlated with forgiveness of others. Therefore, greater anger rumination is expected to predict un-forgiveness of others.

In this study, three forgiveness measures were significantly correlated with both anger rumination and depression (a prerequisite for a mediation relationship):

Forgiveness, Emotional Benefits, and Offense Impact. Given that the preconditions described by Baron and Kenny (1986) were met, hierarchical multiple regressions were computed to test meditational hypotheses. Results showed that anger rumination mediates the relationship between forgiveness and depression. Specifically, when anger

rumination was controlled for, the relationship between Forgiveness and Depression

disappeared. Similarly, when anger rumination was controlled for, the relationship

between Emotional Benefits and Depression disappeared. However, the relationship

between Offense Impact and Depression remained significant, even after controlling for

the effect of Anger Rumination. Thus, anger rumination does not appear to explain why

individuals who perceive an offense as having a major impact on their lives are also more

depressed.

Previous studies have noted the negative correlation between anger rumination

and forgiveness of others (Barber et al., 2005), as well as the positive correlation between anger rumination and depression (Gilbert et al., 2005). Several studies have considered anger rumination as a mediator between forgiveness and depression. For instance, a study by Rye et al. (2008) found that anger rumination mediates an indirect pathway from forgiveness to depression. Specifically, increased forgiveness predicts lower anger rumination which in turn predicts lower depression. A study by Stoia-Caraballo et al.

(2008) found that anger rumination mediates the relationship between forgiveness and sleep quality. In a similar vein, Berry, Worthington, O’Connor, Parrott, and Wade (2005) found that depressive rumination (cognitive or emotional components related to depression) mediates the relationship between forgiveness and depression. Further,

Ysseldyk, Matheson, and Anisman (2007) found that avoiding negative ruminative brooding (i.e. moody pondering and criticism of self and others) may be an important link in the connection between forgiveness and depression. In other words, ruminative brooding partially mediated the relationship between forgiveness and depression. A complicating factor in this area of research is the identification of multiple types of

rumination: anger rumination (considered in the present study), ruminative brooding, depressive rumination, and ruminative reflection (i.e. contemplating the basis for one’s feelings) (Miller et al., 1995; Ysseldyk et al., 2007).

Research has identified a number of negative consequences resulting from rumination. For instance, relative to distraction, rumination maintains and exacerbates depressed mood (Morrow & Nolen-Hoeksema, 1990; Nolen-Hoeksema & Morrow,

1993). Rumination has also been shown to increase global negative attributions, reduce the effectiveness of problem solving (Lyubomirsky & Nolen-Hoeksema, 1995) and increase the accessibility of negative memories (Lyubomirsky, Caldwell & Nolen-

Hoeksema, 1998), relative to distraction. Despite the negative consequences, rumination might be maintained because individuals believe it benefits them in some way. A study by Watkins and Baracaia (2001) found that high levels of self-reported rumination are correlated with high levels of endorsements of positive reasons for rumination.

Ruminators appear to have ambivalent views about rumination, with practically all participants reporting serious disadvantages to rumination, but at the same time 80% of responders reporting at least one perceived benefit for rumination, such as increased understanding and improved problem-solving.

Does attachment style relate to forgiveness and forgiveness perspectives?

This study found evidence that forgiveness relates to attachment style. Consistent with hypotheses, forgiveness of a specific offender significantly differed between secure and fearful attachment styles, with securely attached individuals being more likely to forgive. This is consistent with a number of other studies which have found a relationship between attachment style and willingness to forgive a romantic partner.

Previous research by Webb et al. (2006) noted that participants reporting a secure attachment style had higher scores on self, situation, other, and total forgiveness measures relative to all insecure attachment styles. Additionally, Lawler-Row et al.

(2006) found that securely attached young adults reported more trait and greater state forgiveness compared to insecurely attached young adults. Similarly, Burnette et al.

(2007) found that securely attached individuals displayed greater trait forgiveness than did insecurely attached individuals. Unlike other studies, the present study did not find a relationship between trait forgiveness and attachment style.

A unique contribution of this study was the finding that forgiveness perspectives relate to attachment style. In particular, secure attachment significantly differed from both fearful attachment and preoccupied attachment on how they perceive the offense.

Securely attached individuals were prone to judging the impact of the offense as less severe than both fearful and preoccupied individuals. Finally, fearful and secure attachment style differed significantly with respect to Internal Attributions. Specifically, securely attached individuals were less likely to view the offender as being a fundamentally malicious person.

Attachment style is related to individuals’ beliefs, attitudes, and expectations about self and others. Individuals with a negative model of others expect people to be uncaring and rejecting. Individuals with a negative model of self question their own value and worth and feel intense anxiety about rejection (Burnette et al., 2007; Watt,

McWilliams, & Campbell, 2005). Model of self and model of other may affect perspectives toward forgiveness. Individuals with a negative model of others are mistrustful, avoid others, and are uncomfortable with interpersonal closeness (Watt et al.,

2005). It may be that model of other influences how an individual evaluates the offender and the offender’s behavior. Individuals with a negative model of self and high anxiety about rejection exaggerate the negative consequences of conflict and respond with excessive anger and hurt (Burnette, Davis, Green, Worthington, & Bradfield, 2009). It may be that model of self influences how an individual judges the impact of an offense and/or the benefits of forgiveness.

Previous research suggests that insecurely attached individuals may judge the impact of an offense to be more severe than those that are securely attached. Insecurely attached individuals, compared to those who are securely attached, have difficulty with emotion regulation in times of threat (Mikulincer, 1998). Further, Sroufe (1984) recognized that securely attached individuals are able to tolerate increasing levels of tension while maintaining organized behaviors. Therefore, when evaluating the impact of an offense, securely attached individuals may be less likely to perceive the offense as disrupting their emotional state and behavior. A study by Mikulincer (1998) lends support to the idea that insecurely attached individuals may be more likely to view an offender as malicious. Participants were exposed to hypothetical scenarios differing in the intentions of a romantic partner (hostile, ambiguous, non-hostile). Individuals with insecure attachment styles were found to be more prone to attribute hostile intent across all types of hypothetical scenarios, while securely attached individuals attributed hostile intent only when there were clear contextual cues (in the hostile scenario). This research demonstrates the tendency of insecurely attached individuals to view a romantic partner differently than securely attached individuals.

Interestingly, in this study, secure attachment and dismissing attachment styles

did not significantly differ across any of the forgiveness measures. Both the fearful and

dismissing styles are associated with difficulties becoming close to and relying on others,

but they differ on internalized sense of self-worth. Only the fearful style is consistently

related to social insecurity and lack of assertiveness. Individuals with a dismissing

attachment style, on the other hand, often downplay the importance of others who wrong

them, and thereby are able to maintain a high self esteem (Bartholomew & Horowitz,

1991). It may be that by downplaying the importance of particular relationships,

individuals with a dismissing attachment style are able to resemble, at least superficially,

securely attached individuals in some ways.

Study Limitations

Several limitations should be considered when interpreting the results of this study. First, caution should be taken in generalizing the results to the general population because of the demographic characteristics of this population. All participants in this study were college students. Therefore, the majority of participants were relatively young (18 to 22 years old), and Caucasian (84.7 %). A strength of this study is that students were included from two different geographical regions--the Midwest and the

Northeast. Further research is needed to examine the relationship between forgiveness, attachment style, depression, and anger rumination among individuals with varying demographic characteristics.

Another limitation to this study is that a cross-sectional design was employed.

The results provide a “snapshot” of how variables relate, but cannot inform how the relationship between attachment and forgiveness changes over time. Indeed, some

43 research has indicated that attachment representations may be contextually activated and partner-specific, rather than global (Mikulciner, Shaver, Gillath, & Nitzberg, 2005). In this way, an individual’s attachment style may be expected to change from relationship to relationship. Yet, several studies have indicated that attachment is stable across time, and is unlikely to change from relationship to relationship (Bohlin, Hagehull, & Rydell, 2000;

Kirkpatrick & Davis, 1994). Additionally, this study was limited by the exclusive use of self-report questionnaires for data collection. Self-report data have been frequently used when studying attachment style and forgiveness, but it has been pointed out by other researchers that observer report and behavioral measures should be considered to enhance our understanding of the forgiveness process (McCullough, Hoyt, & Rachal, 2000).

A further limitation to this study involves the manner in which attachment was assessed. This study utilized a working model four category-approach (Griffin &

Bartholomew, 1994). Yet, other research has favored a behavioral and emotional regulation perspective focusing on the avoidant and anxiety dimensions (3 category approach) (Fraley & Shaver, 1998). In spite of these criticisms, some researchers have agreed that a four category-approach is a valid way to assess attachment (Griffin &

Bartholomew, 1994). Additionally, the present study relied on a forced choice, categorical, measure to assess adult attachment style. Brennan, Clark, and Shaver (1998) note that, compared to categorical measures, dimensional measures of attachment have greater power and precision. It may be that a primary justification for using categorical measures of attachment is convenience. Further, Fraley and Waller (1998) found no evidence for a categorical nature of attachment styles. However, Bartholomew and

Horowitz (1991) reported that conclusions from their study were very similar regardless

of whether a correlational analysis was used (involving four continuous ratings) or

whether a between-groups comparison of four groups was used. Thus, there is some

evidence that a continuous attachment rating is not advantageous to a categorical rating.

It is noteworthy that this study added interesting information regarding the influence of

model of self versus model of other on Forgiveness and Forgiveness Perspectives. As

previously mentioned, secure attachment and dismissing attachment did not significantly

differ across any Forgiveness or Forgiveness Perspectives measures. Both secure and

dismissing attachment styles involve a positive model of self, while both preoccupied and

fearful attachment styles involve a negative model of self. Therefore, it seems that model

of self might be the most influential factor in determining Forgiveness and Forgiveness

Perspectives.

It should be noted that the percentage of participants who were securely attached

was somewhat lower than other studies. In this study, approximately 41% of participants

endorsed a secure attachment style, leaving the remaining 59% of participants endorsing

an insecure attachment style (fearful, preoccupied, or dismissing). In a study using the

same measure of attachment style (Relationship Questionnaire), approximately 54% of

participants endorsed secure attachment, with the remaining 46% endorsing a form of

insecure attachment (Lawler-Row et al., 2006). A study using the three-category model found 36% securely attached, with the remaining 64% categorized as insecurely attached

(Burnette et al., 2007). Reviews based on a three-category model (secure, anxious,

avoidant) have tended to find that the distribution for adult attachment styles is similar to

those found for infants, with approximately 55% of individuals classified as secure and

45% classified as insecure (Shaver & Clark, 1994; Shaver & Hazan, 1993). Mickelson,

Kessler, and Shaver (1997) point out attachment style percentages are based largely on

samples of college students. These researchers conducted a study on adult attachment

style using a nationally representative sample and found approximately 59% securely

attached, and 36% endorsing an insecure attachment style (approximately 5% could not

be classified). Notably, most prior studies using college populations found

approximately twice as many anxiously attached individuals. The researchers note that

this discrepancy could be due to cohort effects and societal changes. It is also possible

that anxious attachment style is more common among college students and may change

to secure attachment or avoidant attachment with age.

A final limitation of this study is that it is retrospective in nature. In other words,

participants were required to think of a wrongdoing that occurred in the past and rate

feelings of forgiveness and forgiveness perspectives. Knutson, Enright, and Garbers

(2008) warn that a participant’s understanding of the (forgiveness) process he or she went

through may actually be altered by the very act of forgiving someone. Therefore,

retrospective data regarding forgiveness should be interpreted with caution.

Clinical and Research Implications

This study demonstrates that anger rumination may mediate the relationship between forgiveness and depression. Therefore, in the clinical setting, psychologists may be able to help clients decrease rumination and thereby decrease depression. Specifically, mindfulness training, or present-focused and intentional awareness, may be helpful in reducing rumination (Borders, Earleywine, & Jajodia, 2010). Clients are encouraged to redirect their attention from past, negative memories and future worries to present sensations. Further, clients are instructed to adopt an attitude of nonjudgmental

acceptance. Unpleasant thoughts and feelings are not labeled as “bad,” but rather are

accepted as just one of many sensations to experience (Borders et al., 2010). Cognitive

behavioral therapy can also be used to combat depression and rumination. Helpful

techniques include: positive self-talk, assertiveness training, recognizing and altering automatic thoughts, evaluating cognitive distortions, practicing new behaviors (i.e. attending a therapy group) and thought stopping (i.e. snapping a rubber band against your wrist or yelling “stop” when experiencing negative, ruminative thoughts).

There is also evidence from this study that attachment style relates to forgiveness

and forgiveness perspectives. Insecurely attached individuals may be more likely to view

an offense as having a significant, negative impact on their lives and the offender as

fundamentally malicious. It is likely that by perceiving an offense as severe and the

offender as inherently bad, the forgiveness process will be negatively impacted. It may

be beneficial for clinicians to work with insecurely attached clients on examining their

beliefs about the offense and offender. Further, assisting clients in developing more

complex views of the offender may be helpful. Working with insecurely attached

individuals to examine contextual cues to indicate when another person is being hostile

could be beneficial in reducing the tendency to attribute malicious intent to non-hostile acts. Finally, clinicians may strive to help insecurely attached clients learn to better regulate their emotions and cope with anger and distress in times of conflict. Clinicians can assist clients by using cognitive reshaping or reframing to reduce unpleasant emotions. Another possibility is for clients to utilize mindfulness training and practice nonjudgmental acceptance of emotions (Chambers, Gullone, & Allen, 2009).

Future studies would be helpful in clarifying the relationship between depression, anger rumination and forgiveness. It may be that depression is better predicted by sadness rumination as opposed to anger rumination. Sadness rumination is conceptualized as repetitive thinking that focuses on one’s sadness, and attempts at understanding one’s affect (Peled & Moretti, 2010). It would also be important for future studies to utilize a more diverse sample. In regard to attachment style, it would be helpful to find if the relationship between attachment style, forgiveness, and forgiveness perspectives was replicated using a dimensional measure or 3-category approach.

APPENDIX A Demographics/Background Information

Please complete the following questionnaire. All of your responses will remain confidential. Please do not place your name on this questionnaire.

1. Age: ______

2. Gender: Male____ Female____ (1) (2)

3. Year in school: 1 2 3 4 Other (please specify):_____ (5)

4. Race: _____American Indian _____Asian or Pacific Islander (1) (2)

_____African American _____Latino _____Caucasian (3) (4) (5)

_____Other (Specify) (6)

APPENDIX B Relationship History

1. Are you currently involved in a romantic relationship?

_____Yes _____No (1) (2)

2. Have you previously been involved in a romantic relationship?

_____Yes _____No (1) (2)

3. Have you ever been wronged by a romantic partner?

_____Yes ______No (1) (2)

Think about a romantic partner who has wronged or mistreated you (may be a current partner or previous partner). If you have been wronged or mistreated by more than one romantic partner, select a partner whose actions were very hurtful. If you have never been wronged by a romantic partner consider another person who has wronged or mistreated you and whose actions were especially hurtful. Indicate the nature of the other person’s hurtful actions by checking yes or no for ALL of the questions below.

Yes No (0) (1)

______4. This person lied to me. ______5. This person spread gossip about me. ______6. This person cheated on me. ______7. This person failed to follow through on his/her obligation(s) to me. ______8. This person called me names or used unkind words. ______9. This person physically harmed me. ______10. This person stole from me. ______11. Other (not listed above).

Please write the initials of the person who wronged/mistreated you here. Do NOT write the person’s full name. ______

12. What is/was your relationship to the person who wronged/mistreated you (circle one)?

(1) Romantic partner

(2) Friend

(3) Family member or relative

(4) Acquaintance

(5) Stranger

(6) Co-worker

13. How long ago did the mistreatment by this person occur? Fill in a number for each space. For example, if it occurred 1 year and 2 months ago, you would write “1” year and “2” months.

Number of years_____ Number of months_____

14. How hurt do you feel NOW as a result of this person’s wrongful actions?

Not Hurt At All A Little Hurt Moderately Hurt Extremely Hurt (1) (2) (3) (4)

15. In your opinion, how severe was the wrongdoing that was committed against you?

Not Severe At All A Little Severe Moderately Severe Very Severe (1) (2) (3) (4)

16. How angry do you feel NOW as a result of the person’s wrongful actions?

Not Angry At All A Little Angry Moderately Angry Extremely Angry (1) (2) (3) (4)

APPENDIX C The Forgiveness Scale

Think of how you responded to the romantic partner who wronged or mistreated you. If you have never been wronged by a romantic partner, consider another person who has wronged you. The person you are thinking of should be the person whose initials you wrote down. Indicate the degree to which you agree or disagree with the following statements.

Strongly Disagree Neutral Agree Strongly Disagree Agree

1. I can’t stop thinking about 1 2 3 4 5 how I was wronged by this person.

2. I wish for good things to 1 2 3 4 5 happen to the person who wronged me.

3. I spend time thinking about 1 2 3 4 5 ways to get back at the person who wronged me.

4. I feel resentful toward the 1 2 3 4 5 person who wronged me.

5. I avoid certain people and/or 1 2 3 4 5 places because they remind me of the person who wronged me.

6. I pray for the person who 1 2 3 4 5 wronged me.

7. If I encountered the person 1 2 3 4 5 who wronged me I would feel at peace.

8. This person’s wrongful actions 1 2 3 4 5 have kept me from enjoying life.

Strongly Disagree Neutral Agree Strongly Disagree Agree

9. I have been able to let go of my 1 2 3 4 5 anger toward the person who wronged me.

10. I become depressed when I 1 2 3 4 5 think of how I was mistreated by this person.

11. I think that many of the 1 2 3 4 5 emotional wounds related to this person’s wrongful actions have healed.

12. I feel hatred whenever I 1 2 3 4 5 think about the person who wronged me.

13. I have compassion for the 1 2 3 4 5 person who wronged me.

14. I think my life is ruined 1 2 3 4 5 because of the person’s wrongful actions.

15. I hope the person who 1 2 3 4 5 wronged me is treated fairly by others in the future.

APPENDIX D Forgiveness Likelihood Scale

Imagine the scenarios below happened to you. Based on the information provided, consider the likelihood that you would choose to forgive the person. Then, circle the response that is most true for you.

1. You share something embarrassing about yourself with a friend who promises to keep the information confidential. However, the friend breaks his/her promise and proceeds to tell several people. What is the likelihood that you would choose to forgive your friend?