The construct dimensionality of victim blame: The situations of physical child abuse and rape

Penn.

Pergamon 0191~8869(95)00017-S

End!4

fhjf Vol. 19. No. I. pp. 21-31. 1995

Copyright P 1995 Elsevw Scmce Ltd Pnnred in Great Britain. All rights reserved 0191.8869/95-59.30 tO.00

THE CONSTRUCT DIMENSIONALITY OF VICTIM BLAME: THE SITUATIONS OF PHYSICAL CHILD ABUSE AND RAPE* Robert T. Muher,‘? ‘Psychology

Department,

Robert A Caldwell’

University of Massachusetts, University, East Lansing, (Received

and John E. Hunter?

Boston, MA 021253393 MI 48824, U.S.A.

I2 December

and ‘Michigan

State

1994)

Summary-The dimensionaliry of the victim blame construct was investigated on 897 college undergraduates. Participants were administered questionnaires to assess the extent to which they blame victims of physical child abuse and rape. Several other personality measures were also given. We hypothesized that the multidimensionality of victim blame would be corroborated. Specifcally. we asserted that behavioral, characterological, and abuser justification components of victim blame would emerge as separate constructs. We also predicted that blame directed toward child abuse victims and blame directed toward victims of rape would emerge as distinct factors. A first-order and second-order confirmatory factor analysis was carried out. The findings did not support the hypotheses. Instead, a global victim blame construct emerged with child blame and rape blame as specific factors. Behavioral, characterological, and abuser justification components of victim blame also appeared to stem from a more global victim blame construct.

INTRODUCTION

The tendency to blame victims of personal misfortune has been addressed by several social psychologists, such as Lemer (1980). In his view, blaming victims of tragedy is one way of insuring that the conception of a ‘just world’ is not threatened. Other theorists, such as Shaver (1970) put forth the defensive-attribution hypothesis. Shaver conceived of two important motivating factors that influence people when they evaluate victims of misfortune. Individuals have a need to defend against the possibility that random misfortune may happen to themselves (harm-avoidance). In addition, persons are motivated to defend against the possibility that they will be held responsible if they were to suffer a similar fate (blame-avoidance). Related to the phenomenon of victim blame is the tendency to blame one’s self, following one’s own misfortune. This phenomenon has been found in rape victims (e.g. Ledray, 1986; Libow & Doty, 1979). The most common explanation for self-blame is that victims feel uncomfortable with the belief that they are helpless or vulnerable individuals. It is comforting to feel that one could have somehow prevented the disaster. Blaming one’s self restores a sense of control of outcome (Symonds, 1975).

The behavioral/characterological

blame distinction

Research on victim blame was influenced dramatically by a conceptualization suggested by Janoff-Bulman (1979). She was one of the first to suggest that the construct of blame is multidimensional. She distinguished between two kinds of self-blame; behavioral and characterological. Behavioral self-blame refers to attributions of fault with respect to one’s behavior. In other words, individuals may blame themselves for having engaged in a certain activity. In blaming one’s behavior, the individual is concerned with the future avoidability of the negative outcome. Behavioral self-blame is control motivated. Characterological self-blame refers to attributions of fault concerning one’s character. An individual may blame him/herself for the kind of person s/he is. Characterological self-blame is esteem related. Janoff-Bulman (1979) concluded that behavioral self-blame is typical of rape victims and characterological self-blame is typical of depressed persons. The Janoff-Bulman (1979) distinction has influenced the development of many recent research *A

modified version of this paper was presented at the Ninety-ninth Association, San Francisco, CA, I8 August 1991. +To whom all correspondence should be addressed.

Annual

Convention

of rhe Americcrn

fsyhologictrl

Robert T. Muller et al

22

programs (e.g. Hill & Zautra, 1989: Howard, 1984; Jensen & Gutek, 1982; Meyer & Taylor, 1986; Schoeneman, Hollis, Stevens & Fischer, 1988; Thornton, 1984). as well as theoretical analyses (e.g. Shaver & Drown, 1986). Applying the Janoff-Bulman self-blame distinction to victim blame, Thornton (1984) suggested that when evaluating ‘personally similar’ victims (i.e. victims similar to themselves), observers should invoke behavioral blame, determining that something the victim did (or failed to do) brought about the consequence. Due to the perception that behavior is controllable, observers may make such attributions to a similar victim with the belief that they themselves can behave in a different manner and avoid negative consequence (harm-avoidance). Moreover, evaluating personally similar victims, observers should be less inclined to ascribe characterological blame as that would depict one’s own stable character in a negative light (blame-avoidance). A third dimension to victim blame may be referred to as ‘abuser justification’, the notion that some individuals, under certain circumstances, may tend to justify the actions of violent offenders (justifying the perpetrator). Krulewitz and Nash (1979) su,,oDested that rapists who ‘complete’ an attack are commonly seen as less responsible, since the victim is perceived as not having resisted the attacker adequately. Herzberger and Tennen (1985) found that in reacting to scenarios of abusive parental discipline, Ss rated fathers’ identical behaviors as more appropriate than those of mothers. Cohen (1984) suggested that abused children will justify their parents’ behaviors as an act of familial ‘loyalty’. Although many studies have assumed that victim blame is a multidimensional construct consisting of (at least) behavioral and characterological components, validity studies have not been conducted to determine directly the dimensionality of the victim blame construct. Shaver and Drown (1986) pointed out the insufficient attention that has been paid to construct validity in the measurement of blameworthiness, particularly with respect to the behavioral/characterological distinction. Yet this issue has remained largely ignored. One of the purposes of this study was to determine the extent to which Janoff-Bulman’s multidimensional conceptualization can be meaningfully distinguished. We suggested that behavioral, characterological, and abuser justification components of victim blame would emerge as distinct factors. Situational

specificity

of victim blame

Prior research has assumed that victim blame in the case of rape is separate and distinct from victim blame in other areas, such as child abuse. Several scales have been developed which assume specificity of these constructs. For example, Burt’s (1980) Rape Myth Acceptance Scale, Feild’s (1978) Attitudes Toward Rape Questionnaire. and Ward’s (1988) Attitudes Toward Rape Victims Scale (ARVS) were designed to measure unfavorable or negative attitudes toward rape victims. Similarly, Deitz and colleagues developed an empathy scale specific to the rape situation (Deitz, Blackwell, Daley & Bentley, 1982; Deitz, Littman & Bentley, 1984). Many studies have used these measures of victim blame that are specific to the rape situation. For example, using Burt’s Rape Myth Acceptance Scale, Jenkins and Dambrot (1987) and Krahe (1988) looked at abuser justification specifically in the rape situation. Using the Bamett and Feild (1977) version of the Attitudes Toward Rape Questionnaire to measure perceived victim responsibility, Thornton, Ryckman and Robbins (1982) looked at the relationship between several personality factors (such as dogmatism, locus of control, etc.) and victim blame. Using the Deitz et al. (1982) scale to measure victim empathy, Coller and Resick ( 1987) suggested that “victim empathy is not a component of victim blame” (p. 123). The second purpose of this study was to examine the situational specificity of victim blame. We suggested that a meaningful distinction would emerge between persons in the tendency to blame victims of different types of misfortune. Hypotheses I. Behavioral and characterological

components of victim blame, as well as abuser justification would all emerge as distinct factors. 2. Blame directed toward rape victims and blame directed toward victims of child abuse would also emerge as separate and distinguishable constructs.

Construct dimensionality

of victim blame

23

METHOD Subjects Ss consisted of 897 undergraduate students at Michigan State University recruited via the human Spool (introductory psychology students). Participants were 68% female. Most (87.6%) of the Ss were Caucasian. Blacks were represented by 7.0% of the Ss. The mean age of the students was 18.9 yr. On the Duncan Socio-economic Index (Hauser & Featherman, 1977). the mean and standard deviation of socio-economic status (SES) scores were 57.47 and 19.24, respectively.

Materials Principal

measures

The Child Abuse Scenarios Questionnaire (CASQ) is a multidimensional measure of blame developed by Muller (199 I ). In accordance with Janoff-Bulman’s (1979) multidimensional conceptualization of blame, the construct of victim blame is measured in terms of behavioral and characterological components. An additional dimension, abuser justification, was also reflected in item construction. The questionnaire consists of eight scenarios of child abusive situations, parts of which were derived from Stollak, Scholom, Kallman and Saturansky (1973). In each situation, the parent’s behavior is seen as abusive in accordance with Gil’s (1970) definition of child abuse (the use of physical force in training and punishing children). The definition of child abuse used for these scenarios is also consistent with Heifer’s (1987) definition (any interaction or lack of interaction between family members which results in some non-accidental harm to the individual’s physical and/or developmental states). Responses for each of the items are made on a seven-point Likert scale. Following each of the eight scenarios, Ss indicate the extent to which they agree/disagree with four statements designed to measure behavioral child blame. For example, “this child was responsible for what happened to him/her because of what s/he did or said.” Next, Ss indicate the extent to which they agree/disagree with four statements designed to measure characterological child blame. For example, “this child got what s/he deserved because of the kind of person s/he is.” Following this, Ss state the extent to which they agree/disagree with four statements measuring abuser justification. For example, “this parent’s behavioral response toward the child was justified.” Finally, Ss indicate the extent of their agreement or disagreement with four statements measuring their sense of personal similarity (identification) with the victim. For example, “I identify with this child.” The Attitudes Toward Rape Victims Scale (ARVS) (Ward, 1988) is a 25item questionnaire which is designed to tap supportive/favorable and unsupportive/unfavorable predispositions toward rape victims with special emphasis on those attitudes which reflect disbelief, blame, or denigration, and/or trivialization of the seriousness of rape and its effects on victims. Ss indicate on a five-point intensity scale the extent of their agreement with specific attitudinal statements with respect to rape. The ARVS has been found to have a Cronbach’s “1reliability of 0.83-0.86, and test-retest reliability of 0.80 (Ward, 1988). The ARVS’s construct validity was demonstrated through correlations of the ARVS with Burt’s (1980) Adversarial Sexual Beliefs scale (r = 0.41, PcO.05) as well as Burt’s Acceptance of Interpersonal Violence scale (r = 0.26, P < 0.05). For the purposes of the current study, on the basis of content, four subscales were derived from the Attitudes Toward Rape Victims Scale. These were items which reflected the beliefs that rape victims are: (a) lying, (b) overly provocative, (c) sexually desiring rape, and (d) sexually experienced. For example, an item from the ‘lies’ scale read: “Women often claim rape to protect their reputations.” An item from the ‘provocativeness’ scale read: “Women do not provoke rape by their appearance or behavior” (reverse scored). An item from the ‘sexual desire’ scale read: “Most women secretly desire to be raped.” An item from the ‘sexual experience’ scale read: ‘Sexually experienced women are not really damaged by rape.” Additional

measures

In order to test for parallelism of constructs (a test of construct dimensionality). several questionnaires were included which were expected to differentially correlate with each of the relevant constructs under investigation. These were the following:

21

Robert T. Muller er al. ( 1) The Duncan Socio-economic Index (Duncan SEI) (Hauser & Featherman, 1977) was used to measure socio-economic status. It is the most commonly used single SES measure (Campbell & Parker, 1983). Following questions pertaining to SES, Ss responded to several basic demographic questions, including gender, age, and race. (2) An adapted version of the Conjkt Tucks Scale (CTS) (Straus, 1989; Straus & Gelles, 1989). This scale was used in order to provide an indication of the respondent’s childhood experience with physically abusive, psychologically aggressive, or neglectful parenting. The measure used here listed 25 possible ways to deal with conflict ranging from discussing the issue calmly to using a knife or gun. Ss indicated how frequently each conflict tactic was used during their childhood (under 17 yr of age). Conflict tactics were reported separately for each parent. (3) The Interpersonal Reactivio Index (IRI) (Davis, 1980, 1983) is a 28-item, multidimensional self-report measure consisting of four seven-item subscales, each tapping some aspect of the global concept of empathy. Ss respond to self-descriptive statements on a five-point intensity scale ranging from “does not describe me well” to “describes me very well”. The four subscales consist of: (a) perspective-taking (PT), (b) fantasy (FS), (c) empathic concern (EC), and (d) personal distress (PD). (4) The JLN World Scale, adapted from Rubin and Peplau (1973) is a 16-item questionnaire designed to measure the extent of an individual’s belief that the world is fair. Ss indicate the extent of their agreement with each statement by means of a seven-point intensity scale ranging from strongly disagree to strongly agree. (5) The I. P. and C Scales (Levenson, 1981), is a multidimensional measure of locus of control. The questionnaire measures the extent to which the S believes: (a) s/he has control over his/her own life (I Scale), (b) powerful others have control over his/her life (P Scale), and (c) chance forces have control over his/her life (C Scale). Ss indicate on a six-point intensity scale the extent to which they agree with each of the 24 statements (eight statements for each of the three scales) regarding the locus of control in their lives. (6) The Se(f-Derogation Index (SDI) is a 12-item measure of self-derogation and blame, developed by Muller (1991). Responses are made on a five-point intensity scale ranging from ‘not at all’ to ‘considerably’. Six of the items assess the extent to which, during their childhoods, Ss experienced self-derogation and self-blame for parental punishment. Six additional items assess the extent to which Ss experience current derogation of their childhood selves and current self-blame for former parental punishment.

Procedure Ss were administered the aforementioned questionnaires in the following order. A letter of consent was given first, followed by the Child Abuse Scenarios Questionnaire, the I, P, and C Scales, the Just World Scale, the Conflict Tactics Scales (father then mother), the Interpersonal Reactivity Index, the Self-Derogation Index, the Duncan Socio-economic Index, and the Attitudes Toward Rape Victims Scale. One month later, Ss completed the Child Abuse Scenarios Questionnaire and the Self-Derogation Index (for the purpose of establishing test-retest reliability). Debriefing followed.

RESULTS

Reliabilities For the Child Abuse Scenarios Questionnaire, reliability estimates were calculated across the eight scenarios for each of the behavioral, characterological, and abuser justification subscales. For those three scales respectively, parallel-forms [Cronbach r (Cronbach, 195 I)] reliabilities were rs = 0.97, 0.97, and 0.95; test-retest reliabilities were rs = 0.78, 0.78, and 0.79. For the Attitudes Toward Rape Victims Scale, reliability estimates were calculated for the lies, provocativeness, sexual desire, and sexual experience subscales. For those four scales respectively, parallel-forms reliabilities were rs = 0.75, 0.57, 0.62, and 0.57. For all reliabilities in this study, Ps < 0.001. These reliabilities are considered satisfactory for the purposes of this investigation (Kraemer, 1981).

Construct dimensionality

of victim blame

3

Confirmatory factor analyses Several earlier researchers (Hunter & Gerbing, 1982; Hunter, Gerbing & Boster, 1982) have suggested that in order to conclude that a scale is structurally unidimensional, the correlations among the items should conform to the covariance structure (Joreskog, 1978) of a unidimensional measurement model as evaluated by a confirmatory factor analysis. In such a case, the correlation matrix should be consistent with two product rules (Hunter & Gerbing, 1982); internal consistency and external consistency (parallelism). If the correlations among the items within a cluster form a Spear-man Rank 1 matrix, they are said to be internally consistent (Green, Lissitz & Mulaik, 1977). Items are externally consistent if their correlations with all variables outside the cluster are directly proportional to one another within the bounds of sampling error (Hunter et al., 1982). In order to analyze the measurement model of the current investigation, one estimation procedure was used. A multiple-groups centroid factor analysis (confirmatory factor analysis) was calculated using PACKAGE (Hunter & Cohen, 1969; Hunter, Gerbin,,0 Cohen & Nicol, 1980). Communalities were computed by iterations within each cluster. The cluster solution was sought by successively repartitioning the items until the criteria of unidimensionality was achieved for each cluster.

Child blame (1) For each of the eight scenarios, four items measured behavioral child blame. Each of the eight behavioral child blame subscales demonstrated both internal and external consistency. Moreover, the eight subscales were parallel to one another. Consequently, they were combined into a global behavioral child blame scale. The factor loadings for all items of the behavioral child blame scale onto the behavioral child blame dimension ranged from 0.59 to 0.82 (Ps < 0.001). (2) For each of the eight scenarios, four items measured characterological child blame. Each of the eight characterological child blame subscales demonstrated both internal and esternal consistency. Moreover, the eight subscales were parallel to one another. Consequently. they were combined into a global characterological child blame scale. The factor loadings for all items of the characterological child blame scale onto the characterological child blame dimension ranged from 0.66 to 0.83 (f’s < 0.001). (3) For each of the eight scenarios, four items measured abuser justification. From each of the eight abuser justification subscales, one item was rejected on the basis of both internal and external consistency. The rejected item read: “Most people would have responded as this parent did.” Once this item was eliminated from each of the eight subscales, all the final subscales demonstrated both internal and external consistency. In addition, since these eight subscales were parallel to one another, they were combined into a global abuser justification scale. The factor loadings for all items of the abuser justification scale onto the abuser justification dimension ranged from 0.53 to 0.76 (Ps < 0.001).

Test of Hypothesis 1. Hypothesis 1 asserted that behavioral and characterological components of victim blame, as well as abuser justification would all emerge as distinct factors. In order to test this hypothesis, a second-order confirmatory factor analysis was conducted. In this procedure. the three child blame scales were treated as three items. The results indicated that the scales met the second-order criteria of both internal consistency as well as parallelism. This suggests that the three components of child blame are specific factors of a higher order child blame factor. Table 1 demonstrates parallelism in the correlations of the three relevant scales with outside constructs. The three child blame scales loaded onto the global child blame factor as follows: 0.97, 0.88, and 0.79 for behavioral, characterological, and abuser justification respectively (see Fig. I). It is instructive to note that mean scores on behavioral child blame were significantly higher than were mean scores on characterological child blame. The difference between the means corresponded to an effect size of q = 0.68, (P < 0.001).

Robert T. Muller et al.

26

Table I. Correlartons of the three dimenstons of child blame uith outside constructs Chtld blame dimewon’

I

Outside construct Subject gender Subject age Socio-economic status Physical abuse (father) Physical abuse (mother) Psychological aggression (father) Psychological aggression (mother) Neglect (father) Neglect (mother) Perspective taking empathy Fantasy empathy Empathic concern Personal distress empathy Just world Internal locus of control Powerful others locus of control Chance locus of control Personal similarity Childhood self-derogation Childhood self-blame Current derogarion of childhood self Current blame of childhood self Rape victim blame

2

-0.21 - 0.07 - 0.01 0.1 I 0.1s 0.05 0.08 0.06 0.1 I - 0.22 - 0.23 - 0.23 0.07 - 0.02 -0.1-t 0.25 0.26 0.16 0.09 - 0.01 0.13 0. I3 0.13

*I. Behavioral child blame; 2. characterological

- 0.23 - 0.05 0.04 0.06 0. I3 O.&t 0.04 0.0-I

0.01 -0.16 -0.21 - 0.22 0.08 0.0 I - 0.09 0.21 0.23 0. I7 0.05 0.02 0.1 I 0.13 0-U

3 - 0.22 - 0.03 - 0.06 0.1 I 0.2’ 0.02 0.10 0.07 0.13 - 0.1s 0.23 - 0.29 0. I2 - 0.05 -0.1-t 0.25 0.21 0.16 0.10 - 0.05 0.18 0.10 0.40

child blame; 3. abuser justification.

Rape blame The factor loadings for all items of the lies scale onto the lies dimension ranged from 0.54 to 0.74; except for one item for which the factor loadin g was 0.24 (all Ps < 0.001). The factor loadings for all items of the provocativeness scale onto the provocativeness dimension ranged from 0.42 to 0.64 (Ps < 0.001). The factor loadings for all items of the sexual desire scale onto the sexual desire dimension ranged from 0.52 to 0.70 (Ps < 0.00 1). The factor loadings for all items of the sexual experience scale onto the sexual experience dimension ranged from 0.33 to 0.70 (Ps < 0.001). A further test of the dimensionality of the Attitudes Toward Rape Victims Scale was carried out. That is, in order to test the possibility that the four rape victim blame scales (lies, provocativeness, sexual desire, and sexual experience) are specific factors of a global rape victim blame dimension,

Victim Blame

Rape Blame

Child Blame

Behavioral

Fig.

Characterological

I.

Higher

Abuser Justification

order and lower

order victim

ties

Provocativeness

Sexual ksire

blame

factors

and their factor

Sexual Experience loadings

Construct Table 2. Correlations

dimensionality

of the four dimensions

of victim blame

of rape victim

blame wiih outside const~cts

Rape victim Outside

I

conswct

?

Subject gender Subject age Socio-economic status Physical abuse (father) Physical abuse (mother) Psychological aggression (father) Psychological aggression (mother) Neglect (father) Neglect (mother) Perspective taking empathy Fantasy empathy Empathic concern Personal distress empathy Just world Internal locus of control Powerful others locus of control Chance locus of control Personal similarity Childhood self-derogation Childhood self-blame Current derogation of childhood self Current blame of childhood self Child blame

-

*I.

-l. sexual experience.

Lies; 2, provocativeness:

0.24 -0.15 0.02 0.15 0.1 I 0.07 0.06 0.01 0.14 - 0.20 - 0.20 - 0.22 O,l3 -0.18 - 0.24 0.25 0.33 0.10 0.08 0.0 I 0.14 0.06 0.37

3. sexual de,lrs;

blame dimension* 3 - 0.33 - 0.04 0.02 0.26 0.12 0.13 0.09 0.13 0.08 -0.19 -0.13 - 0.25 0.16 - 0.06 - 0.20 0.23 0.25 0.26 0.21 - 0.03 0.26 - 0.02 0.33

- 0.27 -0.12 - 0.03 0.09 0.06 0.00 - 0.02 - 0.04 0.07 - 0.06 - 0.26 -0.15 0.02 0.00 -0.15 0.18 0.1 I 0.01 0.02 0.02 0.03 0.08 0.44

1 - 0.24 - 0.06 0.05 0. I3 0. I2 0.05 0.06 0.08 0.09 - 0.05 -0.13 -0.17 0. I3 -0.16 -0.1-l 0.10 0.10 0.20 0.19 - 0.0-i 0.22 - 0.05 0.33

a second-order confirmatory factor analysis uas conducted. The four rape victim blame scales were treated as four items. The findings indicated that the scales met the second-order criteria of both internal consistency as well as parallelism. These results indicate a higher order rape victim blame factor. Table 2 demonstrates parallelism in the correlations of the four relevant scales with outside constructs. The four rape victim blame scales loaded onto the global rape victim blame factor as follows: 0.80, 0.60, 0.97, and 0.81 for lies. provocativeness, sexual desire, and sexual experience respectively (see Fig. 1). A test for a global victim blame construct

(test of Hypothesis

2)

Hypothesis 2 asserted that blame directed toward rape victims and blame directed toward victims of child abuse would emerge as separate and distinguishable constructs. This hypothesis was tested as follows. The child blame factor and the rape victim blame factor were tested for parallelism to one another. Table 3 demonstrates the correlations of the two relevant factors with outside constructs. These results demonstrate parallelism between the two dimensions of victim blame indicating a still higher order victim blame factor. The correlation between the two dimensions of victim blame was r = 0.48 (P < 0.001). Thus, child blame and rape victim blame are estimated to load onto the global victim blame factor at r = 0.70 (see Fig. 1).

DISCUSSION

The behavioraVcharacterologica1

blame distinction

The current study was oriented toward elucidating the nature of victim blame as a personality factor. We asserted that behavioral, characterological, and abuser justification components of child blame would all emerge as distinct factors. The results did not support this proposition. Instead, the findings demonstrated that across a multitude of personality variables, behavioral child blame is parallel to characterological, and to abuser justification (to within samplin, 0 error). This implies that all that differentiates the three scales is shared specific error [a kind of measurement error (see Hunter & Gerbing, 19821. The three facets of child blame appear to be reflections of a single higher order child blame factor. Thus, there may be differences between individuals in the extent to which they blame

Robert T. Muller ef al.

28

Table 3. Correlations of the two dimensions of victim blame with outside COflSWWX5

Victim blame dlmenslon” Ourside consmxf Subject gender Subject age Socio-economic status Physical abuse (father) Physical abuse (mother) Psychological aggression (father) Psychological aggression (mother) Seglect (farher) Seglecr (mother) Perspective raking empathy Fantasy empathy Empathic concern Personal distress empathy Just world Internal locus of control Powerful orhers locus of conrrol Chance locus of control Personal similanry Childhood self-derogation Childhood self-blame Current derogation of childhood self Current blame of childhood self

I -0.x - 0.06 - 0.01 0.1 I 0.20 0.05 0.09 0.07 0.09 - 0.21 - 0.26 - 0.28 0.10 - 0.02 -0.1-I 0.27 0.27 0.19 0.09 -0.01 0.16 0.1-l

z - 0.3A -0 12 0.02

0.20 0. I3 0.08 0 06 0.06 0.12 - 0.16 - 0.23 - 0.25 0.14 -0.13 - 0.23 0.21 0.25 0.18 0 I7 - 0.01 0.21 0.02

*I, Child blame; 2. rape victim blame.

abused children in general. However, speaking of a person’s tendency toward behavioral vs characterological blame adds little additional information about that person. It should be noted that mean scores for behavioral child blame were higher than those for characterological. This finding suggests that as a group, Ss ascribed more behavioral than characterological blame. However, mean differences do not imply individual differences. SS retained very similar positions relative to one another on child blame whether they were asked about another’s behavior or character. The differences between rhe two rank orders were small and trivial (i.e. uncorrelated with a host of other personality factors). Thus, these two kinds of blame do not act differently in distinguishing between persons. The current results contradict those of prior research. Janoff-Bulman (1979) distinguished between behavioral and characterological self-blame, and considered that distinction important. She concluded that characterological self-blame was typical of depressed persons, while behavioral self-blame was typical of rape victims. This one study has influenced the development of many research programs (e.g. Hill & Zautra, 1989; Howard, 1984; Jensen & Gutek, 1982; Meyer&Taylor, 1986; Schoeneman er al., 1988; Thornton, 1984), as well as theoretical analyses (e.g. Shaver & Drown, 1986). However Janoff-Bulman’s conclusions seem to have been inaccurate. Let us look at her depression data first. Janoff-Bulman found that depressed persons characterologically self-blamed more than nondepressed persons did. The significant F-test she reported corresponds to a correlation coefficient of r = 0.19. She also found that depressed and nondepressed groups did not significantly differ on behavioral self-blame. The nonsignificant F-test she reported here corresponds to a correlation study indicates that any coefficient of r = 0.14. A samplin, 0 error analysis of Janoff-Bulman’s Thus, the difference in statistical correlation below r = 0.17 would not have been ‘significant’. significance between her two F-values can be explained as a function of her sample size alone. Furthermore, is there really any meaningful difference (in terms of magnitude of effect) between a correlation of 0.19 and 0.14? Consequently, her conclusion that characterological, but not behavioral self-blame distinguishes depressed groups from nondepressed appears to be unwarranted. Next, Janoff-Bulman’s (1979) rape data suggested that amon, u rape victims who self-blamed, behavioral self-blame was more common than characterological self-blame. The major drawback to this finding is that she never determined whether there was a difference between behavioral and characterological blame for the depressed sample, nor for any other group. It may simply be the case that anyone who self-blames will be more likely to blame his/her own behaviors, rather than concluding that there is something wrong with his/her character. This would be consistent with the

Construct

dimensionality of victim blame

29

‘blame-avoidance motives’ posited by defensive attribution theorists (e.g. Shaver. 1970). Thus, once again Janoff-Bulman’s (1979) study does not allow for the conclusion that the behavioral/characterological distinction is one that is important in evaluatin, 0 individual differences in personality. The findings of the current study seem to contradict the results of a second investigation as well. Thornton (1984) found that in evaluating personally similar victims, behavioral blame was ascribed more than characterological blame while the reverse was true in evaluating personally dissimilar victims. At first glance, these results seem to demonstrate the importance of distinguishing between behavioral and characterological components of victim blame. However, the question Thornton should have asked was whether there were differences on each of the blame factors between Ss who are personally similar vs personally dissimilar to the victim. In other words, in order to corroborate his model he should have found that as personal similarity increases, behavioral blame increases; and as personal similarity increases, characterological blame decreases. However, Thornton did not make these comparisons. In the current study it was demonstrated that the correlation between personal similarity and behavioral child blame was virtually identical to the correlation between personal similarity and characterological child blame. Once again, this finding is consistent with the position that behavioral and characterological components of victim blame do not distinguish Ss differently.

Situntiotd specificit?: of victim blame

We posited that blame directed toward victims of child abuse and blame toward rape victims would also emerge as separate and distinguishable constructs. The results did not corroborate this assertion. For example, we found that the correlation between childhood experience of various forms of abuse and child blame was extremely close to the correlation between childhood abuse and rape blame (size differences entirely attributable to the standard error of the TS).Thus, even differences in the extent to which Ss had been abused as children, did not yield differences in S rank order when they were asked about the extent to which they blame child abuse victims versus rape victims. Furthermore, even extent of S personal involvement (‘personal similarity’) with the child victim did not yield differences in S rank order on the variables of child blame vs rape blame (size differences were l/3 standard error of the TS!). Thus, it appears that child blame and rape blame are merely facets of a more global victim blame personality factor. The current findings are not consistent with the suppositions made in prior research. Earlier investigators have assumed that victim blame in the case of rape is separate and distinct from victim blame in other areas. The research of Burt ( 1980), Feild ( 1978), Ward ( 1988). and Deitz et al. (1982, 1984) assumes that victim blame toward rape victims is a meaningful personality factor, and that such a construct as ‘rape empathy’ exists. It seems more likely that these measures of rape victim blame are merely reflecting individual differences in global victim blame. It is possible that attribution of responsibility to victims may be a personality trait that differs in magnitude from person to person (cf. Suedfeld, Hakstian, Rank &i Ballard, 1985). Recent arguments have been put forth (e.g. Funder, 1991) su,,Doesting that global personality traits have far greater explanatory power than do narrow ones. Global traits relate specific behavioral observations to a complex and general pattern of behavior (Funder, 199 1). Furthermore , general personality traits can be very stable across the lifespan (McCrae & Costa, 1984). It is important to note that the situations of child abuse and rape are both situations of interpersonal violence in which one individual is physically victimized by a powerful other. This limits the conclusions that may be drawn. Future research may better address the question of whether victim blame is a global personality factor by utilizing very different victimization situations. For example, it may be interesting to look at whether blaming rape victims is an individual difference factor that is clearly distinct from, say, blaming victims of poverty.

Acknowledgemenn-This research was funded by fellowships granted to the first author. and Humanities Research Council of Canada for their support (Grants No. 152-89-015 acknowledge the helpful comments of Gary Stollak throughout this research.

We wish to thank the Social Sciences No. 452-90-0226). The authors

I and

30

Roben T. Muller er al. REFERENCES

Bamett,

N. J. & Feild. H. S. (1977).

Personnel.

Sex differences

in university

students’

attitudes

toward

rape. Journal

of College Srudenr

18. 93-96.

Burt, M. (1980). Cultural myths and suppon ior rape. Journal of’ Person&r? and Social Ps.~cholog_v,38. 2 17-230. Campbell, R. T. & Parker, R. N. (1983). Subsranlive and statistical considerations in the interpretation of multiple measures of SES. Social Forces, 62, 450466. Cohen, P. M. (1984). Violence in the family: An act of loyalty? Psychorherapy 21, z-19-253. Caller, S. A. & Resick, P. A. (1987). Women’s attributions of responsibility for data rape: The influence of empathy and sex-role stereotyping. Violence and Vicfims. 2, 115-125. Cronbach, L. J. (I 95 I ). Coefficient alpha and the internal structure of tests. Psychomerrika. 16, 297-334. Davis, M. H. (1980). A multidimensional approach to individual differences in empathy. JSASCaralogofSelecred-Documenrs

in Psychology. IO. 55. Davis,

M. H. (1983).

Persona&s

Measuring

differences in empathy: Evidence for a multidimensional approach. Journal of I 13-l 26. P. C. & Bentley, B. J. (I 982). Measurement of empathy toward rape victims and rapists.

individual

and Social Ps.vchology. 44.

Deitz, S. R., Blackwell.

K. T.. Daley.

Journal of Personality and Social Psyhology.

43, 372-383.

Deitz, S. R.. Littman, M. & Bentley. B. J. (1981). Attribution of responsibility for rape: The influence of observer empathy, victim resistance, and victim attractiveness. Se,r Roles, 10, 261-279. Feild, H. S. (I 978). Attitudes toward rape: A comparative analysis of police, rapists, crisis counselors, and citizens. Journal of Personality and Social Psvchologj, 36. 156-l 79. Funder, D. C. (1991). Global t&s: A neo-Allponian approach 10 personality. Psychologrcal Science. 2. 31-39. Gil, D. (1970). Violence against children: Physical child abuse in rhe United Srares. Cambridge, MA: Harvard University Press. Green, S. B., Lissltz. R. W. & Mulaik. S A. ( 1977). Limitations of coefficient alpha as an index of test unidlmensionality.

Educalional and Pqchological

Measurement. 37. 827-838.

Hauser, R. M. & Featherman, D. L. (1977). Tile process ofsrrarrfication: Trends and analyses. New York: Academic Press. Helfer, R. E. (I 987). The developmental basis of child abuse and neglect: An epidemiological approach. In Heifer, R. E. & Kempe, R. S. (Eds.). The barrered child (pp. 60-80). Chicago, IL: University of Chicago Press. Herzberger, S. D. & Tennen, H. (1985). “Snips and snails and puppy dog tails”: Gender of agent. Se.r Roles. 12.853-865. Hill, J. L. & Zaurra. A. 1. (1989). Self-blame attributions and unique vulnerability as predictors of post-rape demoraiization.

Journal of Social and Clinical Ps?_chologx. 8, 368-375. Howard,

J. A. (1984).

The ‘*normal”

victim:

The effects

of gender stereotypes

on reactions

to victims.

Social Ps.vcholog>

Quarrerly, 47, 270-28 I. Hunter,

J. E. & Cohen.

S. H. (1969).

P.ACK.-\GE:

Educational and Psychological Measurement.

A system

of computer

routines

for the analysis

of correlational

data.

29. 697-700

Hunter, J. E. & Gerbing. D. W. (1952). Unidimensional measurement, second order factor analysis, and causal models. In Straw, B. M. & Cummings, L. L. (Eds.). Research in organixrional behavior (Vol. 1). Greenwich, CT: JAI Press. beliefs and personality: Construct invalidity of the Hunter, J. E., Gerbing. D. W. & Bosrer. F. J. i1981-). Machiavellian Machiavelliamsm dimension. Journal of Personalif? and Social Psychology. 43, 1293-l 305. Hunter, J. E.. Gerbing, D. W., Cohen, S. H. Br Nicol, T. S. (I 980). PACKAGE 1980: A s.vstem ofFORTRAN rourinesfor rhe analysis of correlational dam. (Available from Academic Computing Services, Baylor University, Waco, TX.) Janoff-Bulman, R. (1979). Characterological versus behavioral self-blame: Inquiries Into depression and rape. Journal of Personaliry and Social Psychology. 37. 1798-l 809. Jenkins, M. J. & Dambrot, F. IM. (1987). The attribution of date rape: Observers’ attitudes and sexual experiences and the dating situation. Journal of Applied Social Psyzhology. /7. 375-895. Jensen, I. W. & Gutek. B. A. (1982). Attributions and assignment of responsibility in sexual harassment. Journal Social Issues. 38.121-136. Joreskog, K. G. (1978). Structural analysis of covariance and correlation matrices. PsJchomerrika. 43. 4154117. Kraemer, H. C. (I981 ). Coping strategies in psychiatric clinical research. Journal of Consulting and Clinical Ps.vchoiog?: 49. 309-3 19. Krahe, B. (1988). Victim and obsener characteristics as determinants of responsibility attributions to victims of rape. Journal

of

of Applied Social Psychology. 18. 50-58. Krulewitz, J. E. & Nash, J. E. (1979). Effects of rape victim resistance, assault outcome, and sex of observer on attributions about rape. Journal of Persona/i& J7. 557-574. Ledray, L. E. (1986). Recovering from rape. Kew York: Holr. Lemer, M.J. ( 1980). The belief in a jusr icorld. New York: Plenum. Levenson, H. (I98 I). Differentiating among internality, powerful others, and chance. In Lefcourt, H. M. (Ed.). Research tvirh the locsrs of conrrol construct: Assessmenr merhods. New York: Academic Press. Libow, J. A. & Doty, D. W. (1979). An exploratory approach to self-blame and self-derogation by rape victims. American

Journal of Orrhopsychiatry. 49. 670-679. McCrae. R. R. & Costa, P. C., Jr (I 981). Onergillg lives. er7during dispositions. Boston, IMA: Little. Brown. Meyer, C. & Taylor. S. E. (1986). Adjustment to rape. Journcli of Persona1it.v and Social Psychology. 50. I12&-1234. Muller, R. T. (199 I ). Victim blame and child abuse. Unpublished master’s thesis, Michigan State University. East Lansing. MI. Rubin, Z. & Peplau. A. (1973). Belief in a just world and reactions to another’s lot: A study of participant\ in the national draft lottery. Journal of Social Issues. 29. 73-93. Schoeneman, T. J.. Hollis, J. F., Stevens. V. J. & Fischer. K. ( 1988). Recovering stride versus letting it slide: Attributions for “slips” following smoking cessation treatment. Pqcholog~ and Health. 2. 335-3-17. Shaver, K. G. (1970). Defensive attribution: Effects of severity and relevance on the responsibility assigned for an accident. Journal of Personalif? and Social Psxcho1og.v. 14. IO I-I I 3. Shaver, K. G. & Drown, D. (I 986). On causality. responsibility, and self-blame: A theoretical note. Journal of Pervonali!\

and Social Ps?;chologx. 50, 697-702.

Construct Stollak. G. E.. Scholom, children in problem

dimensionality

of victim

blame

A., Kallman, J. & Saturansky, C. (1973). Insensitivity to children: situations. Journal of Abnormal Child Psychology. 2. 169-l 80.

31 Responses

of undergraduates

to

Straus, M. A. (1989). Manual for rhe Conflict Tucrics Scales. Durham: Family Research Laboratory. Straus, M. A. & Gelles, R. J. ( 1989). Physical violence in American families: Riskfacrors nndaduprcrtions ro violence in 8. I45 families. New Brunswick, NJ: Transaction Books. Suedfeld. P., Hakstian, A. R., Rank, D. S. & Ballard, E. J. (1985) Ascription of responsibility as a personality variable. Journal of Applied Social Psychology, IS. 285-3 I I. Symonds, M. (I 975). Victims of violence: Psychological effects and after effects, The American Journnl of Psychoanalysis, 35, 19-26. Thornton. B. (1984). Defensive attribution of responsibility: Evidence for an arousal-based motivational bias. Journal of Personality and Social Psychology, 46. 72 I-734. Thornton. B.. Ryckman, R. M. & Robbins, M. A. (1982). The relationships of observer characteristics to beliefs in the causal responsibility of victims of sexual assault. Humon Relarions. 35. 321-330. Ward, C. (1988). The attitudes toward rape victims scale: Construction, validation and cross-cultural applicability. Psvcholo~~ of Women Quarterly. 12, 127-146.