Corrections and few write up of a master thesis

Mulleretal.2008mediation.pdf

Home >Business & Finance homework help >Management homework help >Corrections and few write up of a master thesis

Childhood Physical Abuse, Attachment, and Adult Social Support: Test of a Mediational Model

Robert T. Muller, Kristin Gragtmans, and Robin Baker York University

This study examined the mediating role of attachment on the relationship between childhood physical abuse and perceived social support in adulthood. The 2 underlying dimensions of attachment, view of self and view of other, were both hypothesized to be potential mediators. Young adults, with and without a history of childhood physical abuse, completed a series of questionnaires inquiring about past abuse experiences and current levels of attachment and social support. Results indicated a robust mediational effect. Namely, both attachment variables were significant mediators in the relationship between childhood physical abuse and social support. In addition, the mediation occurred across all sources of social support, that is, social support from family/close friends, peers, and authority figures. Theoretical and clinical implications are discussed.

Keywords: trauma, maltreatment, attachment, social support

Considerable research has demonstrated that a history of child maltreatment can have long-lasting, negative effects on individu- als. The adverse effects of abuse continue into adulthood and include a wide variety of problems such as difficulty with inter- personal adjustment and relationships, poor coping strategies, sub- stance abuse, and an increased prevalence of depression, person- ality disorders, and psychiatric symptoms (Higgins, 2003; Leitenberg, Gibson, & Novy, 2004; Muller, Kraftcheck, & McLewin, 2004; Muller, Sicoli, & Lemieux, 2000; Trickett & McBride-Chang, 1995).

Abuse and Social Support

Previous work examining the relationship between childhood maltreatment and social support has suggested that social support can be a protective factor against the adverse effects of abuse (Muller, Goebel-Fabbri, Diamond, & Dinklage, 2000). Research focusing on abused children, as well as on adult survivors of childhood abuse, has demonstrated that social support is linked to higher levels of self-esteem and competency (Runtz & Schallow, 1997) and to lower levels of psychopathology, such as depression, externalizing of problems, and symptoms of posttraumatic stress disorder (Muller, Goebel-Fabbri, et al., 2000; Runtz & Schallow, 1997; Winefield, Winefield, & Tiggemann, 1992).

Unfortunately, a number of studies have shown that individuals with a history of maltreatment report significantly less social support in terms of perception of, quality of, and use of such

support. Adults reporting higher levels of childhood abuse tend to report lower perceived support from family (Pepin & Banyard, 2006; Weber & Cummings, 2003), spouses (Varia, Abidin, & Dass, 1996), and friends (Pepin & Banyard, 2006). The quality of social support experienced by such individuals also appears to be compromised, with some studies (e.g., McCarthy & Taylor, 1999) finding significantly lower quality of adult romantic relationships. Finally, adult survivors of child abuse also appear to make poor use of the social supports available to them. For example, such individuals demonstrate greater difficulty with interpersonal rela- tionships (Mullen, Martin, Anderson, Romans, & Herbison, 1996), as well as a tendency to confide in inappropriate and unreliable sources of support during times of crisis (Andrews & Brown, 1988). In addition, they have higher levels of interpersonal sensi- tivity (Malinosky-Rummell & Hansen, 1993), are less sociable (Varia et al., 1996), and have less depth in relationships with parents (Grist Litty, Kowalski, & Minor, 1996).

Despite an accumulating body of research pointing to childhood abuse as a risk factor for the development of problematic social supports, to date very little research has examined potential mech- anisms mediating this relationship. A more complete understand- ing of the developmental psychopathology of an area includes awareness not only of the contributing risk and protective factors but also of the mechanisms by which they operate (Rutter & Sroufe, 2000), that is, an understanding of the direct and indirect processes by which risk factors exercise their influence. The clin- ical value of understanding such risk mechanisms lies in explana- tions of how effects are carried forward and the way in which they influence a person later in life (Rutter & Sroufe, 2000). Trauma theorists and researchers (e.g., Muller, 1997) have begun to look at the role that attachment patterns have in potentiating varying pathways of psychological functioning among survivors of abuse. The current study examines the potential mediating role of attach- ment in the relationship between childhood abuse and current social support.

Robert T. Muller, Kristin Gragtmans, and Robin Baker, Department of Psychology, York University, Toronto, Ontario, Canada.

The authors acknowledge the support of the Social Sciences and Hu- manities Research Council of Canada Grant 410-98-1022.

Correspondence concerning this article should be addressed to Robert T. Muller, Behavioural Sciences Building, 4700 Keele Street, Toronto, ON M3J 1P3. E-mail: [email protected]

Attachment

John Bowlby’s attachment theory assumes that affectional bonds in relation to primary caregivers are adaptive to survival and reduce the risk that an individual will come to harm (Bowlby, 1980a, 1980b, 1982). Early experiences in infancy, childhood, and adolescence are among the important determinants of which de- velopmental pathway an individual will follow (Sroufe, Egeland, & Kreutzer, 1990). From these experiences, working models of self and of other are used to form expectations of patterns of interactions between self and other in relationships. An individu- al’s attachment organization—the expectation of self and other within the context of the attachment relationships— determines the pattern of affectional bonds made throughout the life cycle.

Following Bowlby’s tradition, Bartholomew and Horowitz (1991) proposed a model of adult attachment based on individuals’ models of self and other. Mental representations of self reflect the individual’s perception of being worthy of love and having a sense of competence in relationships, whereas representations of other reflect the individual’s trust of others and expectations as to their dependability and kindness. The model of self may be either positive (self as worthy of love) or negative. The model of other can also be either positive (other as responsive and loving) or negative. Individuals who have developed positive models of both self and other are considered to be secure in their attachment strategies. Secure attachment is widely thought of as a resilience factor (Griffin & Bartholomew, 1994a).

Abuse, Attachment, and Social Support

There is a growing body of literature demonstrating a relation- ship between childhood abuse and patterns of attachment. In particular, studies of abused children (Toth & Cicchetti, 1996), as well as formerly abused adults (Muller, Sicoli, et al., 2000), have documented low levels of attachment security. Muller, Sicoli, et al. demonstrated that 76% of a sample of formerly abused adults were rated insecure in their attachment styles. Similarly, among a sam- ple of high-risk adults, one such risk being a history of maltreat- ment, Weinfield, Sroufe, and Egeland (2000) reported that approx- imately 68% endorsed an insecure attachment style. In contrast, large-scale meta-analyses of nonclinical adult samples indicate that the majority (about two thirds) of such individuals tend to have a secure attachment pattern (van IJzendoorn & Bakermans- Kranenburg, 1996).

Attachment theorists have suggested that it is the perceived unavailability of attachment figures that results in attachment insecurity (Mikulincer & Shaver, 2005). Children develop expec- tations of their own (self) and their caregiver’s (other) behavior in the context of their relationships. Such expectations are built on experience and serve the purpose of allowing the child to organize behavior around what is expected from the caregiver. As maltreat- ing parents move back and forth between affection for and hostility toward their children, the parent is experienced as unpredictable and volatile (Muller & Lemieux, 2000) and the child develops a model of self as ineffective and worthless in the context of rela- tionships. Such children have little confidence in their ability to elicit caring and support and are not likely to seek assistance or reassurance when anxious (Crittenden, 1992).

A secure attachment experience in infancy and childhood has been considered to lay the foundation for an individual’s ability to

establish, tolerate, and maintain intimacy in relationships, leading to increased perception and receipt of support (Sarason, Pierce, & Sarason, 1990). Research is starting to indicate that social support extends beyond the quantity and quality of one’s relationship transactions; social support is also influenced by such factors as the support recipient’s personality, expectations, and needs and preferences (Sarason, Sarason, & Shearin, 1986). Sarason et al. (1990) suggested that one such personality variable is a sense of social support. With age, patterns of attachment become more characteristic of the individual’s personality, and working models of self and other become solidified. Positive models of self and other create a sense of self-efficacy, interpersonal skills, low levels of relationship anxiety, and positive expectations in the context of relationships. Each of these facilitates the individual’s ability to enter into relationships and to positively interpret the events that occur within them. As demonstrated in the literature, both attach- ment security and a sense of social support are relatively stable constructs (Sarason et al., 1990). A handful of studies have sys- tematically examined the relationship between attachment style and social support and have generally found insecure attachment to be a significant risk factor in the prediction of poor social support (Asendorph & Wilpers, 2000; Florian, Mikulincer, & Bucholtz, 1995; Ognibene & Collins, 1998).

Gaps in the Literature

Despite past work on the relationships between abuse, attach- ment, and social support, there remain several gaps in the litera- ture. First, based on the previous review of theory and research, there appears to be good reason to propose a mediational model of the relationships among abuse, attachment, and social support, such that attachment would mediate the relationship between his- tory of abuse and current social support (see Figure 1). Prior research has examined such mediating effects in a very limited way. In particular, one study by McCarthy and Taylor (1999) did find support for such a model; however, that study utilized a very small sample of 39. The current investigation makes use of a much larger sample to reduce sampling error and to provide a more robust test of the mediational model. Furthermore, the participants in McCarthy and Taylor’s study consisted exclusively of women who had experienced poor parenting during childhood. The par- ticipants in the current study represent a much broader segment of the population, including men and women and individuals both with and without histories of abuse. Finally, McCarthy and Taylor limited their outcome variable to adult love relationships, which may differ conceptually from social support from other sources.

Figure 1. Mediational model of the relationships among childhood phys- ical abuse, attachment, and social support. The direct effect of physical abuse on social support is represented by �’.

81ABUSE, ATTACHMENT, AND SOCIAL SUPPORT

The current study looked at a wider range of social support providers.

In addition to methodological differences, the current study departs from McCarthy and Taylor (1999) on theoretical grounds. McCarthy and Taylor chose to examine adult attach- ment categories as measured by Hazan and Shaver’s (1987) adult attachment questionnaire, whereas the authors of the current study chose to examine the underlying continuous di- mensions of adult attachment, view of self and view of other. Previous research has found that continuous attachment dimen- sions (e.g., view of self and view of other) can be measured reliably by self-report measures, whereas categorical attach- ment patterns such as Hazan and Shaver’s secure, avoidant, and ambivalent styles are less likely to demonstrate reliability and validity by self-report (Griffin & Bartholomew, 1994a). Attach- ment theorists postulate that individuals may not be consciously aware of their attachment patterns and may therefore be inca- pable of the introspection required by a self-report question- naire measuring attachment styles.

Griffin and Bartholomew (1994a) have found evidence suggest- ing that all adult attachment measures share fundamental dimen- sions first identified by Bowlby, the view of self and the view of other. As yet, research has not looked at the mediational model of abuse, attachment, and social support by examining the two un- derlying attachment dimensions separately. Looking at these con- structs separately is both theoretically and clinically important to the extent that an understanding of such mediational variables can speak to issues of treatment. Clinical intervention would be fo- cused differently if one or the other attachment variable was found to be more or less relevant. For example, interventions focused on building self-concept would look different than those oriented toward building a sense trust in others. From a methodological perspective, prior studies have shown that view of self and view of other are relatively statistically independent of each other (e.g., Muller, Lemieux, & Sicoli, 2001), so they need to be treated as such. A test is needed of a mediational model that examines the extent to which each attachment variable uniquely demonstrates mediating effects.

It is important to note that a number of researchers (e.g., Muller, 1997) have argued that the study of those at risk only makes sense if it can alter the odds for such persons. The current study helps shed light on one of the paths by which individuals move along maladaptive developmental trajectories. A greater understanding of the processes by which risk and resilience occur can be beneficial in terms of increasing the knowledge base of developmental psychopathology; equally important, it can provide a stronger theoretical and empirical context upon which to structure intervention programs aimed at altering the odds for vulnerable individuals.

Current Study

The purpose of the current study is to investigate attachment as a potential mediator in the relationship between history of abuse and social support. We hypothesized that a history of physical abuse would predict both attachment and perceived social support and that attachment would predict perceived social support: (a) physical abuse would be negatively related to each of the dimensions of attachment (positive view of self and positive

view of other), (b) physical abuse would be negatively related to all types of social support, (c) positive view of self and positive view of other would be positively related to all types of social support, and (d) positive view of self and positive view of other would each have a mediating effect in the relationship between childhood physical abuse and perceived social support (see Figure 1).

Method

Participants

Originally, 876 York University students agreed to participate in the study. Many students were enrolled in an introductory psychology course (59.3%) and were recruited through the undergraduate research participant pool (URPP), receiving a 1% bonus mark toward their final grade as incentive. The remainder of the students (40.7%) were from second- and third-year psychology courses, receiving either partial course credit or the opportunity to participate in a drawing for prizes (i.e., movie certificates) as determined by the course instructor. A demographic questionnaire provided information regarding age, gender, year at York University, culture, and religion. In addition, socioeconomic status was derived from questions regarding parental income, occupation, and education (Holling- shead, 1975; Mueller & Parcel, 1981).

The participant pool was made up of 82.4% women and 17.6% men. The majority of the students were between the ages of 16 and 24 years (91.3%) and were in their first year of University at York (54.5%). The predominant ethnic group reported was Euro Canadian (58.1%), followed by Asian Canadian (14%). The most commonly identified religious backgrounds were Roman Catholic (38%) and other (31.1%). Participants most com- monly reported that their fathers (38.9%) and mothers (43.9%) had at- tended university or college. Family income levels were rated as follows: 0 –$25,000 (8.6%), $25,000 –$45,000 (20.1%), $45,000 –$65,000 (27.1%), $65,000 –$90,000 (23.9%), and $90,000� (20.4%).

Procedure

Approval for this project was granted by the York University human subjects review committee. Participants were asked to complete a series of questionnaires for a study concerning “important relationships in an indi- vidual’s environment.” The students were informed that their participation in the study was voluntary, all responses were anonymous, and they could withdraw at any time and still benefit from a bonus mark or an opportunity in the drawing. Groups of 15 to 20 participants completed each of the questionnaires, as outlined in the Measures section. There were six order- ings of the measures (Versions 1– 6), each beginning with demographic information and ending with the Record of Maltreatment Experiences (ROME). No effects were found for the order in which the questionnaires were completed.

Measures

Relationship Scales Questionnaire. The Relationship Scales Questionnaire (RSQ; Griffin & Bartholomew, 1994b) is a 30-item questionnaire designed to measure adult attachment. Individuals rate each item on a 5-point Likert scale (1 � not at all like me to 5 � very much like me) according to how well each item describes their method of relating to those with whom they have close relationships. The results can be coded to represent four patterns of attachment or to represent two dimensions of attachment. Attach- ment patterns include (a) secure attachment (e.g., “I find it easy to get emotionally close to others”), (b) fearful attachment (e.g., “I worry that I will be hurt if I allow myself to become too close to

82 MULLER, GRAGTMANS, AND BAKER

others”), (c) preoccupied attachment (e.g., “I worry that others don’t value me as much as I value them”), and (d) dismissing attachment (e.g., “I prefer not to have other people depend on me”). The mean rating for each of the four subscales is computed, creating four continuous variables corresponding to attachment styles.

Relationship Questionnaire. The Relationship Questionnaire (RQ; Bartholomew & Horowitz, 1991) is an adaptation of Hazen and Shaver’s (1987) three-category adult attachment measure, designed to test Bartholomew’s (1990) four-category model of attachment. It consists of four short paragraphs describing four attachment styles: secure, fearful, preoccupied, and dismissing. In the first part, individuals choose the paragraph that best describes them in their closest relationships. In the second part, individuals rate each description according to personal applicability on a 7-point scale, ranging from 1 (not at all like me) to 7 (very much like me). The RQ may be combined with the RSQ by calculating z scores for each of the attachment styles and then averaging parallel scores. Examinations of the RQ have demonstrated both discriminant and convergent validity (Bartholomew & Horowitz, 1991; Griffin & Bartholomew, 1994b). In the current study, the correlations among the subscales of the RQ and the RSQ were 0.56 ( p � .01) for attachment security, 0.59 ( p � .01) for fearful attachment, 0.53 ( p � .01) for preoccupied attachment, and 0.42 ( p � .01) for dismissing attachment.

Two dimensions of attachment characterize an underlying model of self and other (Schafer & Bartholomew, 1994), also scored continuously. By combining the four attachment patterns (secure, fearful, preoccupied, and dismissing) of the composite measure of attachment, model of self and other were calculated. Model of self was calculated by subtracting attachment patterns typified by negative view of self (fearful and preoccupied) from the sum of attachment patterns distinguished by positive view of self (secure and dismissing). Similarly, model of other was calcu- lated by subtracting attachment patterns characterized by negative view of others (fearful and dismissing) from the sum of attachment patterns distinguished by positive view of others (secure and preoccupied). Construct validity of the self and other dimensions has been demonstrated (Bartholomew, 1990), as has convergent validity (Brennan, Shaver, & Tobey, 1991). In addition, conver- gent and discriminant validity of these two dimensions have been demonstrated using multitrait-multimethod matrices and confirma- tory factor analyses (Griffin & Bartholomew, 1994a). Moderate to high test–retest reliability and stability were demonstrated over an 8-month period, ranging from r � 0.81 to 0.84 for view of self and from 0.72 to 0.85 for view of other (Schafer & Bartholomew, 1994).

Multi-Dimensional Support Scale. The Multi-Dimensional Support Scale (MDSS; Neuling & Winefield, 1988; Winefield et al., 1992) is a self-report measure of social support. Individuals rate the amount and adequacy of supportive behaviors received from three sources: family/close friends, peers, and authority fig- ures. Nineteen items per support source describe a number of supportive actions. Individuals rate the frequency with which they received each type of supportive behavior from individual support sources within the past month on a 4-point scale ranging from 1 (never) to 4 (usually/always).

Research using the MDSS has demonstrated high internal reli- ability when administered to a sample of young adults (Winefield

et al., 1992) and modest reliability with cancer patients (Neuling & Winefield, 1988), with Cronbach’s alphas ranging from .81 to .90 in the former study and from .50 to .81 in the latter study. The scale has also shown predictive validity for adjustment and measures of psychological well-being, including self-esteem, depression, anx- iety, and health (Neuling & Winefield, 1988; Winefield et al., 1992). Winefield et al. recommended that items concerning dis- traction as a supportive behavior be eliminated from the MDSS in future research, as it was not rated as helpful by their sample. Thus, the current study excluded the three distraction items from analy- ses. Furthermore, the authors of the current study conducted a confirmatory factor analysis to evaluate the unidimensionality of the purported support subscales of family/close friends, peers, and authority figures. Results of the confirmatory factory analysis on the entire sample revealed that the model was a good fit: goodness of fit index (GFI) � .97, comparative fit index (CFI) � .98, and root-mean-square error of approximation (RMSEA) � .04. Cron- bach’s alpha coefficients indicated that the support source sub- scales were reliable: Family/Close Friends, � � .81; Peers, � � .85; and Authority Figures, � � .86.

ROME, self-report. The ROME (Wolfe & McGee, 1994) is a self-report, retrospective measure of victimization from birth to 17 years. The perpetrators of abuse measured by the ROME are mothers and fathers. Five subscales make up the ROME: positive parenting practices, psychological maltreatment, physical mal- treatment, exposure to family violence, and sexual abuse (Wolfe & McGee, 1994). Only the items relating to physical abuse were of interest for this study. Each item is rated for the frequency with which it occurred on a 5-point Likert scale: 0 � never; 1 � very infrequently (once or twice during childhood); 2 � occasionally (1–3 times per year); 3 � frequently (1–2 times per month); and 4 � very frequently (once a week or more). Cronbach’s alpha reliabilities for each of the subscales are 0.82, 0.89, 0.81, 0.81, and 0.91, respectively (R. A. McGee, personal communication, July 1995). Similar Cronbach’s alpha reliabilities were found for the current sample, with subscale values of 0.90, 0.88, 0.72, 0.80, and 0.90, respectively. Test–retest reliability coefficients ranged from 0.70 for psychological maltreatment to 0.93 for sexual abuse (Wolfe & McGee, 1994).

In the current study, level of physical abuse was measured continuously across all participants. It is important, however, to note whether participants in our sample meet criteria for a history of child abuse. Consistent with previous abuse categorization criteria (Muller, 1996; Straus, 1989), physical maltreatment items were divided into minor and major physical abuse items. Four minor items, considered to be less harmful, included (a) threw or smashed or hit or kicked something; (b) threw something at you; (c) pushed, grabbed, or shoved you; and (d) slapped you. Five major items are considered to be more destructive and potentially lethal: (a) kicked, bit, or hit you with a fist; (b) hit or tried to hit you with something; (c) beat you up; (d) threatened you with a knife or gun; and (e) used a knife or gun. The minimum maltreat- ment criteria consisted of the endorsement of one of the following experiences: at least one minor physical abuse experience with frequencies ranging from 1–2 times a month to once a week or more; at least two major physical abuse experiences with a fre- quency of once or twice during childhood plus at least one minor physical abuse experience of any frequency; at least one major physical abuse experience with a frequency of 1–3 times a year

83ABUSE, ATTACHMENT, AND SOCIAL SUPPORT

plus at least one minor physical abuse experience of any fre- quency.

Correlations were calculated to determine the relationships among physical abuse, attachment, and social support (see Table 1). Relationships among abuse, attachment, and social support were found. Modest to moderate correlations in the positive direc- tion were found between model of self and model of other, and reported levels of support from each of family/close friends, peers, and authority figures. This is indicative that positive scores on view of self and other were associated with higher levels of perceived support from a variety of sources. In addition, higher levels of physical abuse were associated with lower scores on view of self and other. All correlations were modest to moderate in magnitude.

There were 285 participants (33%) who met criteria for having experienced some form of physical maltreatment. Among those with an abuse history, 22% were considered to be have been most severely maltreated. These individuals reported having experi- enced at least one major physical abuse item with frequencies ranging from 1–2 times per month to once a week or more and were deemed the most severely maltreated. Moderately abused individuals were those reporting one of the following experiences: (a) at least two major physical abuse experiences with a frequency of 1–3 times a year, and (b) at least one of the three most severe major physical abuse experiences (beat you up, threatened you with a knife or gun, or used a knife or gun) once or twice during childhood. Forty percent of the abuse group fell into the moderate abuse category. Finally, 38% of the abuse group fell into the minimal abuse category. Thus, univariate analyses of the ROME indicate that there was considerable variability in the sample with respect to history of physical abuse.

Results

Preliminary Analyses and Evaluation of Statistical Assumptions

All variables were examined for outliers and missing data. Forty-seven cases were found to be missing data on at least one scale (n � 12 for abused, and n � 35 for nonabused) and were eliminated from further analyses; all remaining cases were deemed valid. The resulting sample size was 829 (n � 285 for abused, and

n � 544 for nonabused). Correlations were examined between predictor variables to confirm the absence of multicollinearity. All distributions appeared normal, and residuals appeared to have a straight-line relationship with the predicted dependent variable scores. Homoscedasticity was also confirmed. To determine whether demographic variables should be included in analyses, one-way analyses of variance (ANOVAs) were conducted for each demographic variable, including age, gender, year at York Uni- versity, culture, religion, and socioeconomic status. Abuse history, attachment (view of self and view of other), and social support (family/friend, peer, and authority) all served as the dependent measures. Marginal relationships (which failed to reach statistical significance) were found between both culture and religion for each of physical abuse history, view of self, and view of other. Similarly, socioeconomic status was marginally related to view of self. Results indicated that none of the demographic variables had a significant relationship with the dependent variable (social sup- port), and thus they were excluded from further analyses. Means and standard deviations for all measures are displayed in Table 1.

Correlational Analyses

Correlations were calculated in order to determine the relation- ships among physical abuse, attachment, and social support (Table 1). Relationships among abuse, attachment, and social support were found. Modest to moderate correlations in the positive direc- tion were found between model of self and model of other, and reported levels of support from each of family/close friends, peers, and authority figures. This is indicative that positive scores on view of self and other were associated with higher levels of perceived support from a variety of sources. In addition, higher levels of physical abuse were associated with lower scores on view of self and other. All correlations were modest to moderate in magnitude.

The mediating effects of attachment on the relationship between childhood physical abuse and perceived social support were esti- mated using the product of coefficient method described by Mac- Kinnon, Lockwood, Hoffman, West, and Sheets (2002) and MacKinnon, Warsi, and Dwyer (1995). AMOS 7.0 statistical pack- age (Arbuckle, 1999) was used to calculate maximum likelihood estimates simultaneously, providing the necessary path coeffi- cients (see Table 2 and Figure 2).

Table 1 Correlation Matrix for Abuse, Attachment, and Social Support Variables (N � 829)

Variables Mean Standard deviation

Physical abuse

Attachment dimension Social support providers

Self Other Family/friends Peers Authority

Physical abuse 1.7298 .26 Attachment dimension

Self .0048 2.15 �.12**

Other �.0006 2.03 �.10** .15**

Social support Family/friends 3.1023 .67 �.14** .31** .23**

Peers 2.6598 .74 �.08* .23** .18** .43**

Authority 2.3121 .73 �.08* .13** .16** .28** .33**

*p � .05. **p � .01.

84 MULLER, GRAGTMANS, AND BAKER

An effect was said to be mediated if (a) physical abuse had a statistically significant effect on the hypothesized mediator (�), (b) the hypothesized mediator had a statistically significant effect on the outcome variable (�), and (c) the mediated effect (��) was statistically significant, as determined by the Sobel test. First introduced by Sobel in 1982, the test divides the mediated effect (��) by its standard error, which is approximately equal to the square root of �2s�2 � �2s�2 (Sobel, 1982, 1986). The resulting test statistic is treated as a z test; values larger in absolute value than 1.96 are significant to the .05 level (Mackinnon et al., 2002).

As expected, physical abuse predicted lower levels of both mediators (positive view of self and other). In addition, each of the mediators independently predicted higher levels of the outcome variables (perceived social support). Finally, the mediated effect (��) of view of self and view of other was statistically significant,

as determined by the Sobel test, for each type of social support (see Table 3).

If the relationship between the independent variable and the dependent variable is significant when the mediator is absent from the model, yet insignificant when the mediator is added to the model, the relationship between the independent variable and the dependant variable is said to be completely mediated by the intervening variable (Mackinnon et al., 2002). In our sample, physical abuse negatively predicted social support from all pro- viders prior to adding the attachment dimensions of self and other to the model. When view of self and other were included in the model, the relationships between physical abuse and support from peers and from authority figures were no longer significant (see Figure 2). As such, view of self and other can be said to completely mediate the relationship between physical abuse and support from

Table 2 Summary of Simultaneous Regression Analyses for Variables Predicting Attachment Dimensions and Social Support

Predictors

Attachment dimensions Social support providers

Model of self Model of other Family/friends Peers Authority

B SE � B SE � B SE � B SE � B SE �

Physical abuse �1.02*** .28 �.12 �0.80** .27 �.10 �0.22* .08 �.09 �0.12 .10 �.04 �0.14 .10 �.05 Model of self 0.08*** .01 .27 0.07*** .01 .20 0.03** .01 .10 Model of

other 0.06*** .01 .19 0.05*** .01 .15 0.05*** .01 .14

Note. In terms of statistical significance, B and � are identical. R2 � .01 for View of Self; R2 � .02 for View of Other; R2 � .03 for Family; R2 � .07 for Peers; R2 � .13 for Authority. *p � .05. **p � .01. ***p � .001.

Figure 2. Mediating role of view of self and view of other in the relationship between childhood physical abuse and perceived social support by different providers. This figure depicts standardized regression weights for attachment subscales with social support subscales. Beta coefficients calculated prior to including the mediators in the model are in parentheses. *p � .05. **p � .01. ***p � .001.

85ABUSE, ATTACHMENT, AND SOCIAL SUPPORT

peers and support from authority figures but only partially mediate the relationship between physical abuse and support for family/ close friends.

MacKinnon et al. (1995) reported that the proportion of the direct effect mediated can be calculated by dividing the mediated effect (��) by the total effect (�’ � ��). Calculation of the proportion mediated revealed that the dimensions of attachment accounted for 16% to 38% of the relationship between childhood physical abuse and perceived social support (see Table 3). A multiple group model was used to test for any possible gender differences in our sample. No differences were found between men and women ( p � .081).

Discussion

The purpose of the current study was to examine one of the paths by which individuals who have experienced physical child abuse develop a sense of social support. The results were consis- tent with the expectation that the relationships among abuse, attachment, and social support can be explained through a medi- ational model. Specifically, both dimensions of attachment, view of self and view of other, mediated the effects of childhood physical abuse on perceived social support, including support from family/close friends, peers, and authority figures. These findings are consistent with earlier evidence that childhood physical abuse predicts poor attachment relationships (Muller, Sicoli, et al., 2000; Toth & Cicchetti, 1996; Weinfield et al., 2000) and that security of attachment predicts more supportive relationships (Asendorph & Wilpers, 2000; Florian et al., 1995; Ognibene & Collins, 1998; Sarason et al., 1990).

To date, the only study investigating attachment as a mediator between abuse and social support was limited by a small sample size (N � 39; McCarthy & Taylor, 1999). The current results extended these findings with a much larger sample size (N � 829), providing a more robust test of the mediational model. In addition, the current study examined a sample that included men and women and individuals both with and without abuse histories. Also, the current study included social support from different providers (family/close friends, peers, and authority figures) as outcome variables, whereas McCarthy and Taylor studied only adult love

relationships. As it is relatively unknown which sources of social support are most important and under what circumstances (Sara- son, Pierce, & Sarason, 1990), the finding that support from all sources, including family/close friends, peers, as well as authority figures, is mediated by the same attachment dimensions suggests that successful intervention techniques may be more universal than if different mediators accounted for different sources of social support.

The finding that the two underlying attachment variables are each unique mediators contributes to an understanding of the developmental trajectories individuals may find themselves on following physical abuse. An internal model of self represents the extent to which an individual feels worthy, loveable, and able to maintain meaningful relationships. In the context of interpersonal relationships, such a self model has a reciprocal relationship with social support. Feeling lovable augments feelings of competency and self-efficacy in relationships, which in turn contribute to feelings of being cared for (Sarason, Pierce et al., 1990). Prior research has found that a negative self model can impede the ability to build and maintain meaningful relationships. People with a negative self model may not seek support, failing to encounter opportunities in which to improve their sense of self in relation to others (Ognibene & Collins, 1998).

In contrast, an internal model of other represents the extent to which an individual feels other people are generally accepting and responsive and the degree to which interpersonal relationships feel comfortable and secure (Griffin & Bartholomew, 1994a). Through a history of social exchanges, individuals build expectations about the motives and beliefs of others within the context of relation- ships. Consequently, clinical intervention aimed at building each attachment variable may focus on somewhat different goals. In- terventions focused on an individual’s model of self may target building self-concept and feelings of worthiness and developing of confidence in relationships. In contrast, interventions focused on an individual’s model of other may be oriented more on learning that others can be accepting and responsive and that one can feel secure within such relationships. As both view of self and other were found to be mediators between physical abuse and social support, interventions focused on building each of these attach- ment variables may be of clinical importance.

Bowlby (1988) proposed that relationship-specific attachment representations in early childhood generalize over time. As an individual develops relationships beyond the original parent– child dyad, these generalized representations provide a guide for attachment-related affect, ideas, and behaviors. It is in this way that early attachment experiences are brought forward to influence later relationships. Researchers have found evidence of stability of attachment from infancy to adulthood, after the transition to mar- riage, and after the transition to parenthood (Crowell & Waters, 2005) consistent with Bowlby’s theory. In addition, the Berkley Longitudinal Study found that infant security (vs. insecurity) in the Strange Situation predicted secure-autonomous classification (vs. insecure classification) on the Adult Attachment Interview at age 26 in a low-risk sample (Main, Hesse, & Kaplan, 2005). Clearly, early intervention is most favorable for those at risk, as it optimizes the probability of specific secure attachment representations in childhood, which are likely to generalize into adulthood.

Nevertheless, Bowlby’s conception of attachment was that it continues to evolve throughout the life span and that psychother-

Table 3 Mediaional Analyses and Sobel Test Results

Variables

Mediated effecta

(��)

t value (Sobel Test)

Total effect (�� )

Proportion mediated

��/(�� )

Model of self Family/friends �.09 �3.31*** �.31 .28 Peers �.07 �3.06** �.19 .38 Authority �.03 �2.23* �.17 .21

Model of other Family/friends �.05 �2.62** �.31 .16 Peers �.04 �2.46** �.16 .26 Authority �.04 �2.41* �.18 .22

aMediated effect refers to the extent to which model of self and model of other mediated the effect of childhood physical abuse on all types of social support. *p � .05. **p � .01. ***p � .001.

86 MULLER, GRAGTMANS, AND BAKER

apy affords individuals with the opportunity for productive reap- praisal of working models of self and other, even in adulthood (Bowlby, 1980a, 1980b, 1982).

Theorists have argued that it makes sense to study persons at risk to the extent that the results give way to transforming the odds for such individuals (e.g., Muller, 1997). With an understanding of the mechanisms by which vulnerable individuals are able to achieve adaptive outcomes, clinicians are better equipped to help those at risk. The results of the current study help explain one of the ways by which individuals may move along maladaptive developmental trajectories. The mediational model suggests that, in part, social support is beneficial to individuals who are open to relationships. Those who have the psychological preparedness to accept intimacy and who have the capacity to form a secure attachment should be considerably more able to accept and utilize the social supports that may be available to them. As mentioned previously, theorists have suggested that attachment styles can shift in response to psychotherapy (Bowlby, 1988; Crittenden, 1997). Crittenden suggested that an attachment relationship can form between client and therapist, allowing the client to explore frightening relationship experiences. Through this exploration, models of self and other begin to be reworked, and the therapist becomes a secure base from which the client can discover self- worth and trust in others. Times of transition and crisis activate the attachment system and tend to lead to attachment anxiety. These times may be especially crucial clinically, providing opportunities for interventions.

The current study was limited by the use of self-report, retro- spective measures of maltreatment. Such measures are subject to potential biases, such as selective reporting and distortion of events. Future longitudinal studies are suggested, following both high- and low-risk children into adulthood. Such designs provide clearer information regarding direction of effects. However, it should be added that the pencil-and-paper method lends an aspect of anonymity, potentially making it less difficult for some partic- ipants to respond truthfully to sensitive subject matter such as abuse and trauma. Further exploration should include research designs using a variety of methods of measurement, including interviews and behavioral observations.

A second limitation is related to the undergraduate population from which the sample was taken and the applicability of the results to the population a large. Barrett (2005) pointed out that, whereas samples of convenience pose questions of generalizabil- ity, university samples may be ideal for studying attachment pro- cesses because the self-esteem of university students is often under threat. Barrett also suggests that including a variety of cultures in the undergraduate research pool will enhance generalizability of results. York University is highly diverse in terms of the cultural background and socioeconomic status of its student body (Orn- stein, 2006; York Institute for Social Research, 1997) Neverthe- less, the current results should be replicated on a wider variety of adult populations for further generalizability of findings.

Overall attachment security was slightly low in the current sample, presenting a third limitation. As this finding may, to some extent, compromise the representativeness of the sample, prospec- tive studies should seek to replicate our findings.

As a whole, this study contributes to the trauma literature by providing information as to possible pathways by which childhood physical abuse can lead to lower levels of perceived social support in

adulthood. An understanding of the mediational role of attachment is of clinical importance in so far as it speaks to risk, protective factors, and protective mechanisms among vulnerable individuals.

Résumé

La présente étude examine le rôle médiateur de l’attachement dans la relation entre les mauvais traitements dans l’enfance et le soutien social perçu à l’âge adulte. Les deux dimensions sous- jacentes à l’attachement, la perception de soi et la perception des autres, ont toutes deux été établies comme médiateurs possibles dans l’hypothèse. Les jeunes adultes, avec ou sans antécédents de mauvais traitements pendant leur enfance, ont répondu à divers questionnaires portant sur les mauvais traitements subis et leur niveau actuel d’attachement et de soutien social. Les résultats ont révélé un solide effet médiateur. En effet, les deux variables relatives à l’attachement se présentent comme des médiateurs importants dans la relation entre les mauvais traitements physiques pendant l’enfance et le soutien social. De plus, la médiation se manifeste dans toutes les sources de soutien social : membres de la famille et amis proches, pairs et figures d’autorité. Les répercus- sions sur les applications théoriques et cliniques sont discutées.

References

Andrews, B., & Brown, G. W. (1988). Marital violence in the community: A biographical approach. British Journal of Psychiatry, 153, 305–312.

Arbuckle, J. L. (1999). Amos 7.0. Chicago: Smallwaters Corp. Asendorph, J. B., & Wilpers, S. (2000). Attachment security and available

support: Closely linked relationship qualities. Journal of Social and Personal Relationships, 17, 115–138.

Barrett, F. (2005, September). How random is that? APS Observer, 18(9). Retrieved October 6, 2006, from www.psychologicalscience.org/ observer/getArticle.cfm?id�1838

Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspec- tive. Journal of Social and Personal Relationships, 7, 147–178.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61, 226 –244.

Bowlby, J. (1980a). Attachment and loss: Vol. 2. Separation anxiety and anger. London: Hogarth Press.

Bowlby, J. (1980b). Attachment and loss: Vol. 3. Loss, sadness and depression. London: Hogarth Press.

Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment. London: Hogarth Press.

Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London: Routledge.

Brennan, K., Shaver, P., & Tobey, A. (1991). Attachment styles, gender, and parental problem drinking. Journal of Social and Personal Rela- tionships, 8, 451– 466.

Crittenden, P. M. (1992). Children’s strategies for coping with adverse home environments: An interpretation using attachment theory. Child Abuse & Neglect, 16, 329 –343.

Crittenden, P. M. (1997). Truth, error, omission, distortion, and deception: The application of attachment theory to the assessment and treatment of psychological disorder. In S. M. C. Dollanger & L. F. DiLalla (Eds.), Assessment and intervention: Issues across the lifespan (pp. 35–76). Mahwah, NJ: Erlbaum.

Crowell, J., & Waters, E. (2005). Attachment representations, secure-base behavior, and the evolution of adult relationships. In K. E. Grossmann, K. Grossmann, & E. Waters (Eds.), Attachment from infancy to adult- hood (pp. 245–304). New York: Guilford Press.

87ABUSE, ATTACHMENT, AND SOCIAL SUPPORT

Florian, V., Mikulincer, M., & Bucholtz, I. (1995). Effects of adult attach- ment style on the perception and search for social support. The Journal of Psychology, 129, 665– 676.

Griffin, D., & Bartholomew, K. (1994a). Models of the self and other: Fundamental dimensions underlying measures of adult attachment. Journal of Personality and Social Psychology, 67, 430 – 445.

Griffin, D., & Bartholomew, K. (1994b). The metaphysics of measure- ment: The case of adult attachment. In K. Bartholomew & D. Perlman (Eds.), Advances in personal relationships: Attachment processes in adulthood (pp. 17–52). Bristol, PA: Jessica Kingsley .

Grist Litty, C., Kowalski, R., & Minor, S. (1996). Moderating effects of physical abuse and perceived social support on the potential to abuse. Child Abuse & Neglect, 20, 305–314.

Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511–524.

Higgins, D. (2003). The relationship of childhood family characteristics and current attachment styles to depression and depressive vulnerability. Australian Journal of Psychology, 55, 9 –14.

Hollingshead, A. B. (1975). Four factors index of social status. New Haven, CT: Department of Sociology, Yale University.

Leitenberg, H., Gibson, L. E., & Novy, P. L. (2004). Individual differences among undergraduate women in methods of coping with stressful events: The impact of cumulative childhood stressors and abuse. Child Abuse & Neglect, 28, 181–192.

MacKinnon, D. P., Lockwood C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83–104.

MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30, 41– 62.

Main, M., Hesse, E., & Kaplan, N. (2005). Predictability of attachment behavior and representational processes at 1, 6, and 19 years of age. In K. E. Grossmann, K. Grossmann, & E. Waters (Eds.), Attachment from infancy to adulthood (pp. 245–304). New York: Guilford Press.

Malinosky-Rummell, R., & Hansen, D. J. (1993). Long-term consequences of childhood physical abuse. Psychological Bulletin, 114, 68 –79.

McCarthy, G., & Taylor, A. (1999). Avoidant/ambivalent attachment style as a mediator between abusive childhood experiences and adult relation- ship difficulties. Journal of Child Psychology and Psychiatry, 40, 465– 477.

Mikulincer, M., & Shaver, P. R. (2005). Attachment theory and emotions in close relationships: Exploring the attachment-related dynamics of emotional reactions to relational events. Personal Relationships, 12, 149 –168.

Mueller, C. W., & Parcel, T. L. (1981). Measures of socioeconomic status: Alternatives and recommendations. Child Development, 52, 13–30.

Mullen, P. E., Martin, J. L., Anderson, J. C., Romans, S. E., & Herbison, G. P. (1996). The long-term impact of the physical, emotional, and sexual abuse of children: A community study. Child Abuse & Neglect, 20, 7–21.

Muller, R. T. (1996). Family aggressiveness factors in the prediction of corporal punishment: Reciprocal effects and the impact of observer perspective. Journal of Family Psychology, 10, 474 – 489.

Muller, R. T. (1997). A reciprocal effects model of social support and resilience: A longitudinal study of high risk university students. Ottawa, Canada: Social Sciences & Humanities Research Council, Standard Research Grant (No. 410 –98 –1022).

Muller, R. T., Goebel-Fabbri, A. E., Diamond, T., & Dinklage, D. (2000). Social support and the relationship between family and community violence exposure and psychopathology among high risk adolescents. Child Abuse & Neglect, 24, 449 – 465.

Muller, R. T., Kraftcheck, E. R., & McLewin, L. A. (2004). Adult attach-

ment and trauma. In D. R. Catherall (Ed.), Handbook of stress, trauma, and the family (pp. 203–229). New York: Brunner-Routledge.

Muller, R. T., & Lemieux, K. E. (2000). Social support, attachment, and psychopathology in high risk formerly maltreated adults. Child Abuse & Neglect, 24, 883–900.

Muller, R. T., Lemieux, K. E., & Sicoli, L. A. (2001). Attachment and psychopathology among formerly maltreated adults. Journal of Family Violence, 16, 151–169.

Muller, R. T., Sicoli, L. A., & Lemieux, K. E. (2000). Relationship between attachment style and posttraumatic stress symptomatology among adults who report the experience of childhood abuse. Journal of Traumatic Stress, 13, 321–332.

Neuling, S. J., & Winefield, H. R. (1988). Social support and recovery after surgery for breast cancer: Frequency and correlates of supportive be- haviours by family, friends and surgeon. Social Science and Medicine, 27, 385–392.

Ognibene, T. C., & Collins, N. L. (1998). Adult attachment styles, per- ceived social support and coping strategies. Journal of Social and Personal Relationships, 15, 323–345.

Ornstein, M. (2006). Ethno-racial groups in Toronto, 1971–2001: A de- mographic and socio-economic profile. Toronto: Institute for Social Research, York University.

Pepin, E. N., & Banyard, V. L. (2006). Social support: A mediator between child maltreatment and developmental outcomes. Journal of Youth and Adolescence, 35, 617– 630.

Runtz, M. G., & Schallow, J. R. (1997). Social support and coping strategies as mediators of adult adjustment following child maltreatment. Child Abuse and Neglect, 21, 211–226.

Rutter, M., & Sroufe, A. (2000). Developmental psychopathology: Con- cepts and challenges. Development and Psychopathology, 12, 265–296.

Sarason, B. R., Pierce, G. R., & Sarason, I. G. (1990). Social support: The sense of acceptance and the role of relationships. In B. R. Sarason, I. G. Sarason, & G. R. Pierce (Eds.), Social support: An interactional view (pp. 97–128). New York: Wiley

Sarason, I. G., Sarason, B. R., & Shearin, E. N. (1986). Social support as an individual difference variable: Its stability, origins, and relational aspects. Journal of Personality and Social Psychology, 50, 845– 855.

Schafer, E., & Bartholomew, K. (1994). Reliability and stability of adult attachment patterns. Personal Relationships, 1, 23– 43.

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290 –312). Washington, DC: American Sociological Association.

Sobel, M. E. (1986). Some new results on indirect effects and their standard errors in covariance structure models. In N. Tuma (Ed.), So- ciological methodology (pp. 159 –186). Washington, DC: American Sociological Association.

Sroufe, L. A., Egeland, B., & Kreutzer, T. (1990). The fate of early experience following developmental change: Longitudinal approaches to individual adaptation in childhood. Child Development, 61, 1363–1373.

Straus, M. A. (1989). Manual for the Conflict Tactics Scales. Toth, S. L., & Cicchetti, D. (1996). Patterns of relatedness, depressive

symptomatology, and perceived competence in maltreated children. Journal of Consulting and Clinical Psychology, 64, 32– 41.

Trickett, P. K., & McBride-Chang, C. (1995). The developmental impact of different forms of child abuse and neglect. Developmental Review, 15, 311–337. Durham: Family Research Laboratory.

van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (1996). Attach- ment representations in mothers, fathers, adolescents, and clinical groups: A meta-analytic search for normative data. Journal of Consult- ing and Clinical Psychology, 64, 8 –21.

Varia, R., Abidin, R., & Dass, P. (1996). Perceptions of abuse: Effects on adult psychological and social adjustment. Child Abuse & Neglect, 20, 511–526.

Weber, L. J., & Cummings, A. L. (2003). Relationships among spirituality,

88 MULLER, GRAGTMANS, AND BAKER

social support, and childhood maltreatment in university students. Coun- seling and Values, 47, 82–95.

Weinfield, N., Sroufe, L., & Egeland, B. (2000). Attachment from infancy to early adulthood in a high-risk sample: Continuity, discontinuity, and their correlates. Child Development, 71, 695–702.

Winefield, H. R., Winefield, A. H., & Tiggemann, M. (1992). Social support and psychological well-being in young adults: The Multi- Dimensional Support Scale. Journal of Personality Assessment, 58, 198 –210.

Wolfe, D. A., & McGee, R. (1994). Dimensions of child maltreatment and their relationship to adolescent adjustment. Development and Psycho- pathology, 6, 165–181.

York Institute for Social Research. (1997). Ethnic breakdown of York Univer- sity Students. Institute for Social Research Reports. Toronto: Author.

Received March 18, 2006 Revision received April 28, 2007

Accepted August 17, 2007 �

E-mail Notification of Your Lastest CPA Issue Online!

Would you like to know when the next issue of your favorite Canadian Psychological Association journal will be available online? This service is now available. Sign up at http://notify.apa.org/ and you will be notified by e-mail when issues of interest to you become available!

Avis par courriel de la disponibilité des revues de la SCP en ligne!

Vous voulez savoir quand sera accessible en ligne le prochain numéro de votre revue de la Société canadienne de psychologie préférée? Il est désormais possible de le faire. Inscrivez-vous à http://notify.apa.org/ et vous serez avisé par courriel de la date de parution en ligne des numéros qui vous intéressent!

89ABUSE, ATTACHMENT, AND SOCIAL SUPPORT