Knowing the Abuser Inside and Out: The Development and Psychometric Evaluation of the Identification With the Aggressor Scale

Abstract

Identifying with the aggressor is a process wherein victims of abuse, particularly during childhood, take on their perpetrator’s experience. The victim defers to the perpetrator and adopts the perpetrator’s experience, learns the perpetrator’s desires and needs, and gratifies them. Although the clinical and theoretical literature suggests that identification with the aggressor occurs in the aftermath of abuse and has negative long-term implications, to date this concept has not been empirically investigated. To facilitate an exploration of this subject, the current study evaluated the psychometric properties of a new measure: the Identification With the Aggressor Scale (IAS). The study was conducted among convenience samples of students using online surveys. In Study 1, the IAS was administered to 318 students. In Study 2, the IAS, and a battery of questionnaires assessing features of abuse, dissociation, posttraumatic stress disorder (PTSD) symptoms, and posttraumatic guilt were administered to a convenience sample of 368 students. Four reliable IAS factors emerged from Study 1: Adopting the perpetrator’s experience concerning the abuse, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator. In Study 2, a confirmatory factor analysis confirmed the scale’s underlying factor structure. A history of childhood abuse, recurrence and severity of abuse, and the perpetrator being a parental figure were all associated with higher IAS scores. In addition, IAS scores were correlated with dissociation, PTSD symptoms, and posttraumatic guilt. The present findings indicate that the IAS has good psychometric properties, making it useful as an assessment tool in future research.

Keywords

trauma childhood abuse identifying with the aggressor PTSD

Being subjected to abuse can lead to the formation of unique bonds between victims and their perpetrators. Victims may develop an affiliative attachment, protectiveness, and affection toward their abusers, alongside a minimization and rationalization of the abuse (Jülich, 2005). These patterns of relations suggest that although victims of abuse may survive these concrete instances of abuse, they often do so at a cost: that is, they may learn their perpetrators’ desires and needs by adopting their perpetrators’ intrapsychic experiences. The concept of identification with the aggressor, which denotes abuse victims’ tendency to take on their perpetrators’ psychological experience, has received some theoretical and clinical attention. Nevertheless, to the best of our knowledge, this subject has not to date been empirically investigated. As such, the goal of this study was to fill this gap by evaluating the psychometric properties of a new measure: the Identification With the Aggressor Scale (IAS).

Identification with the aggressor, a concept which was originally introduced in the psychoanalytic literature by Ferenczi (1932, 1933), describes a process wherein survivors of abuse, particularly during childhood, fuse with, take on, and introject their abusers’ experience. Identifying with the aggressor is a wide, multifaceted process which goes beyond behavioral mimicking of the attacker’s aggression, as described by Anna Freud (1936), and denotes a pervasive change in the perceptual world of the abused child, who may “become one” with the attacker (Frankel, 2002). Ferenczi’s theory reveals four main aspects of identification with the aggressor: losing one’s agency and replacing it with that of the perpetrator, becoming hypersensitive to the perpetrator, adopting the perpetrator’s experience concerning the abuse, and identifying with the perpetrator’s aggression.

According to this theory, identification with the aggressor is an automatic reaction that occurs in the face of extreme threat, the aim of which is to protect the victim’s self from the traumatic reality of intolerable abuse (Ferenczi, 1932; Frankel, 2002). This phenomenon may occur during an ongoing and overwhelming attack that the child cannot escape, avoid, or prevent, especially when he or she is attached to and dependent on the perpetrator. In such situations, the powerless child may subordinate himself/herself mentally to the perpetrator, replacing his or her own experience and agency with that of the perpetrator (e.g., Frankel, 2002). The child gives up his or her own sense of self and dissociates large chunks of his or her subjective experience (Ferenczi, 1933). The child’s self is essentially demolished by terror, leaving him or her without feelings and emotions of his or her own (Coates & Moore, 1997).

To anticipate the abuse and minimize its hazards, the abused child may mold his or her own experience into that of the perpetrator’s (Ferenczi, 1933; Frankel, 2002). The child may become hypersensitive to the perpetrator’s emotions, get into the perpetrator’s mind, and learn “from the inside” the perpetrator’s wants and needs. The child may feel what the perpetrator feels, experience the perpetrator’s pathological passions as if they were his or her own, and thus deny, minimize, or rationalize the abuse (Ferenczi, 1932). The child also learns who the perpetrator wants him or her to be and gauges precisely how to appease, seduce, flatter, placate, or otherwise disarm the perpetrator (Frankel, 2002). Getting into the perpetrator’s mind, the child may also take on the perpetrator’s aggression. Thus, the child might inflict aggression both inward and outward (Davies & Frawley, 1994; Frankel, 2002).

The theory of identification with the aggressor implies that this phenomenon might be associated with the features of childhood abuse, as well as with dissociation, posttraumatic stress disorder (PTSD) symptoms, and posttraumatic guilt. As was mentioned before, identification with the aggressor is claimed to have a “protective” quality, as it aims to defend the abused child from danger and insanity (Frankel, 2002). Thus, it is reasonable to expect that identification with the aggressor would be related to the severity of the abuse, such that harsher and recurrent abuse might be linked with heightened levels of identification with the aggressor. It has been suggested that identification with the aggressor serves an important interpersonal role as well, as it enables the child to preserve a positive relationship with the abuser (Ferenczi, 1933). This function becomes particularly relevant in cases when the perpetrator is the child’s attachment figure (Davies & Frawley, 1994; Howell, 2014). In such situations, the child has dissociative parts of the self that contain different elements of the relationship with the abuser. The self-state that identifies with the aggressor contains the positive aspects of the relations between child and abuser and thus maintains an idealized attachment relationship with the abuser, a needed figure in the child’s life (Davies & Frawley, 1994). Based on this argument, it is reasonable to expect that identification with the aggressor would be more prominent among survivors of abuse inflicted by a parental figure than survivors of abuse inflicted by a nonparental figure.

Identification with the aggressor may be related with dissociation, which is defined as a disturbance in the normally integrative functions of identity, memory, consciousness, and perception of the environment (Carlson & Putnam, 1993). According to theoretical literature, dissociation enables the abuse victim to empty his or her mind and split off his or her emotions, feelings, and sensations; the vacuum thus created is then filled by the experience of the perpetrator (Ferenczi, 1932, 1933). At the same time, it is argued that identification with the aggressor may further dissociation. By closely identifying with and knowing the perpetrator, victims can detect the specific parts of the experience which most endanger them and need to be nullified by dissociation (Frankel, 2002). This enduring experience of dissociation may become embedded in the survivors’ mental processes long after the abuse has ended (Classen, Koopman, & Spiegel, 1993).

Although identification with the aggressor protects the victim and enables his or her survival during the time of the abuse, it may be associated with long-term elevated levels of posttraumatic distress. Given that identifying with the aggressor requires a splitting-off of parts of one’s experience, it may hinder the processing of the trauma. Furthermore, some argue that introjecting the aggressor’s experience, which is an essential reaction during the abuse itself, could lead to a constant reliving of the traumatic past (Frankel, 2002; Van der Kolk, 1989). These intrusive experiences may be manifested in heightened posttraumatic stress symptoms (i.e., PTSD symptoms), which comprise intrusion, avoidance, hyperarousal, and negative alterations in cognitions and mood (American Psychiatric Association, 2013).

Finally, identification with the aggressor may also be associated with intensified guilt. It has been argued that survivors of childhood abuse who identify with their aggressors take on the perpetrators’ “badness” or the perpetrators’ perception that the children themselves are bad (Ferenczi, 1932; Frankel, 2002). Thus, elevated posttraumatic guilt, which denotes feelings of self-blame or a distressing sense of responsibility concerning events and behaviors around the trauma (Kubany et al., 1996), may be fueled by identification with the aggressor.

Research in the trauma field has revealed some trends that are in line with the theory of identification with the aggressor. For example, previous studies have shown that victims of abuse develop strong bonds with their perpetrators and minimize the abuse (Jülich, 2005) and often hold negative feelings about the self as bad and blameworthy (e.g., Feiring, Taska, & Chen, 2002). Furthermore, research has provided evidence regarding the propensity for perpetration of violent acts (e.g., Papalia, Ogloff, Cutajar, & Mullen, 2018) as well as for additional victimization by others (Walker, Freud, Ellis, Fraine, & Wilson, 2017) among victims of abuse. Nevertheless, these studies have not explored identification with the aggressor—which is argued to underlie these various outcomes—directly. Hence, the validity of the concept of identification with the aggressor is still far from being clear. In fact, to the best of our knowledge, the attention that the concept of identification with the aggressor has received has been limited to clinical case studies and theoretical writing within the trauma literature (e.g., Frankel, 2002). Accordingly, the model of identification with the aggressor has been mainly applied in psychoanalytical oriented psychotherapies and has not been implemented in evidence-based treatments aimed for victims of abuse. Thus, there is a definite need for validation of this concept.

The current article aims to facilitate the empirical exploration of the concept of identification with the aggressor via the development of a new measure: the IAS. This article describes the scale and the results of two studies assessing its psychometric properties. The first study presents the self-report questionnaire and its structure. The second study aims to confirm the scale’s structure and evaluate its reliability and validity.

Study 1

This study aimed to develop a scale to measure identification with the aggressor—creating the scale’s items and examining its factor structure.

Method

Participants and procedure

We conducted an online survey of a convenience sample of students. Students were recruited via Israeli social media groups dedicated to students of colleges and universities in Israel. The survey was accessible through Qualtrics, a secure web-based survey data collection system. The survey took 10 min to complete, on average, and was open from July 1, 2017, to October 1, 2017. The survey was anonymous, and no data were collected that linked participants to recruitment sources. The Tel Aviv University Institutional Review Board (IRB) approved all procedures and instruments.

Clicking on the link to the survey guided potential respondents to a page that provided information about the purpose of the study, the nature of the questions, and a consent form (i.e., the survey was voluntary; respondents could skip any questions or quit at any time; responses would be anonymous). The first page also offered researcher contact information. Each participant was given the opportunity to take part in a lottery that included five gift vouchers of 30$ each.

A total of 370 participants entered the survey site. Only 318 (85.95%) who completed all of the questionnaires utilized in the current study were included in the analysis. The average age of the participants was 26.80 years (SD = 5.24). The vast majority of the sample, 84.9% (n = 270), were women. Their average number of years of education was 15.04 (SD = 2.39; range: 12-22). The majority of the sample, 69.8% (n = 222), were undergraduate students; 18.9% (n = 60) were graduate students; 3.1% (n = 10) were PhD students; and 6.9% (n = 22) were students studying in various recognized educational/vocational programs. About half of the sample, 54.4% (n = 173) were single, and 42.5% (n = 135) reported that they were currently in an intimate relationship. The majority of the sample, 86.8% (n = 276), reported that they were in a heterosexual relationship.

Measures

Background variables

Participants completed a brief demographic questionnaire that assessed gender, age, education, and relational status.

The IAS

The IAS was developed to measure identification with the aggressor. Based on the literature, the following expressions of identifying with the aggressor were conveyed: subordination to the perpetrator, hypersensitivity to the perpetrator’s inner experience, compliance with the perpetrator’s needs, replacement of one’s own experience and agency with that of the perpetrator, and identification with the perpetrator’s aggression. In accordance with these expressions, an initial pool of 40 items was generated. Four clinical experts, who specialize in trauma and childhood abuse, were presented with the lists of manifestations and items and were asked to indicate to what extent they thought each reflected an expression of identification with the aggressor (not at all; to a certain extent; very much). They were also encouraged to offer modifications in the wording of the items and to make note of any items they felt were missing, or of any redundancies.

To assess the clarity of the questionnaire to laypeople, the scale was also presented to three individuals (university students with whom the principal investigators were acquainted via their social networks) who were not familiar with trauma research in general or with this research study in particular. These individuals were asked to read the items and to indicate any item whose meaning was unclear. As a result of these procedures, minor modifications were made in the wording of the items, and 10 items were omitted due to redundancy.

The final version of the questionnaire included a total of 30 items. Items were phrased in the third person (i.e., “some people”) to avoid the implications of any social undesirability of the experiences (see, for example, Carlson & Putnam, 1993). First, respondents were asked to describe the severest abuse or offense (in the event that they had not been abused) they had ever experienced. Next, respondents were presented the items as reflecting “possible reactions that people may experience as a result of abuse or offense.” Participants were asked to rate on an 11-point Likert-type scale, ranging from 0% (never) to 100% (all the time), the frequency with which they, themselves, experienced each manifestation of identification with the aggressor in regard to their perpetrator (or offender, in the event that they had not been abused). Two hundred five (64.47%) participants reported a history of at least one type of childhood abuse (i.e., physical, sexual, or emotional childhood abuse) compared with 113 participants (35.53%) who did not report childhood abuse and who referred to a nonabusive offense in the past when filling out the scale.

Data analysis

Overall, 0% to 6.3% of data was missing across variables. To determine whether the missing data were random, analyses of differences were conducted using Little’s Missing Completely at Random (MCAR) test (Collins, Schafer, & Kam, 2001). The analysis indicated that the data were missing completely at random, χ²(53121) = 41,321.81, p = 1.00. We used SPSS 25, employing a maximum-likelihood (ML) estimation procedure, which is considered to be an optimal method for handling missing data (Collins et al., 2001).

Descriptive analyses of the items were examined. The IAS’s structure was assessed by exploratory factor analysis (EFA) with Promax rotation. The number of factors was determined by two criteria. The first was based on Cattell’s “scree test” and referred to the drop in the magnitude of the eigenvalues. The second was based on parallel analysis according to the guidelines provided by Hayton, Allen, and Scarpello (2004). Subsequent to the results of the EFA, mean scores were calculated for the total score and for the score of each factor. Cronbach’s alpha assessed the IAS’s internal consistency. Pearson correlations assessed the direction and magnitude of the relationships between the factors. The associations between the IAS total score and factors and demographic variables were examined via Pearson correlations and t tests.

Results

Factor analyses

Table 1 presents means and standard deviations of the scale’s items. Findings of both scree plot and parallel analysis indicated that the scale had a four-factor structure. These four factors were included in the EFA with a Promax rotation, accounting for 53.85% of the identification with the aggressor experience. In regard to item selection for the four factors, factor loadings of .40 and higher were considered meaningful. Items with loadings below .40 on all factors were excluded from further analyses. Items that appeared to have double loadings on two factors (i.e., loadings of greater than .40) were excluded.

Table 1.

Means and SDs of IAS Items and Results of the EFA With Promax Rotation (Study 1).

	M	SD	Factor 1	Factor 2	Factor 3	Factor 4
4. Sometimes the needs and desires of people are similar to those of their perpetrator	32.08	30.25	.65
5. Some people feel that the point of view of their perpetrator is the right one	29.25	30.49	.71
16. Some people care about their perpetrator	50.09	34.41	.65
17. Some people understand their perpetrator	38.84	32.52	.80
18. Some people feel that their perpetrator was right	28.65	30.13	.73
19. Some people feel a certain pleasure from the offense	20.28	26.76	.63
20. Some people feel emotions that are similar to those of their perpetrator	24.62	26.18	.80
21. Some people understand their perpetrator’s desire to hurt them	24.94	28.16	.66
29. Some people identify with their perpetrator	24.69	28.29	.78
25. Some people are afraid that they are as aggressive as their perpetrator	37.23	33.03		.86
26. Some people are afraid that they resemble their perpetrator	38.18	34.63		.89
27. Some people believe that they too can be as aggressive as their perpetrator	35.53	33.83		.91
28. Some people feel that they and their perpetrator are in fact not so different from one another	30.75	31.08		.87
30. Some people feel that they behave like their perpetrator	25.44	29.21		.87
3. Some people do not know what they want in the presence of their perpetrator	51.67	31.66			.65
8. Some people feel “wiped out” in the presence of their perpetrator	63.65	31.65			.70
12. Some people feel that they have no desires or needs	38.81	32.45			.63
13. Some people act just like their perpetrator would want them to act	53.46	31.06			.73
15. Some people feel that the way their perpetrator would want them to feel	51.76	32.04			.59
7. Some people “read the thoughts” of their perpetrator	33.90	29.25				.68
9. Some people sense what their perpetrator’s needs or feels	40.66	30.64				.79
10. Some people “get” their perpetrator accurately	40.63	30.58				.84
11. Some people can identify when their perpetrator poses a danger	50.85	30.77				.57
Percent of variance			35.90	10.04	7.03	6.27
Eigenvalues			8.26	2.31	1.62	1.42

Note. Factor 1: Adopting the perpetrator’s experience; Factor 2: Identifying with the perpetrator’s aggression; Factor 3: Replacing one’s agency with that of the perpetrator; Factor 4: Becoming hypersensitive to the perpetrator.

Seven items did not load on any factor and therefore were deleted following the first factor analysis. After excluding these seven items, a second factor analysis was conducted with the remaining 23 items to ensure that all had factor loadings of .40 or higher. The second EFA yielded the same four factors, explaining 59.14% of the variance of identification with the aggressor (see Table 1 for the final factor loadings). Factor 1, “adopting the perpetrator’s experience concerning the abuse” (eigenvalue = 8.26), comprised nine items representing the feeling that one’s needs, desires, and emotions were similar to the perpetrator’s; perceiving the perpetrator’s point of view as correct; understanding the perpetrator; caring about the perpetrator; feeling that the perpetrator was right; and experiencing a certain pleasure from the abuse. Factor 2, “identifying with the perpetrator’s aggression” (eigenvalue = 2.31), comprised five items representing the fear of being as aggressive as the perpetrator; the fear of being like the perpetrator; the belief that one had the potential to be as aggressive as the perpetrator; the feeling that one was not so different from the perpetrator; and the feeling that one could behave like the perpetrator. Factor 3, “replacing one’s agency with that of the perpetrator” (eigenvalue = 1.62), comprised five items representing not knowing what one wanted when facing the perpetrator; feeling “abolished” in the presence of the perpetrator; feeling as if one had no desires or needs; behaving exactly as the perpetrator would have wanted one to behave; feeling the way the perpetrator would have wanted one to feel. Factor 4, “becoming hypersensitive to the perpetrator” (eigenvalue = 1.42), comprised four items representing “reading the thoughts” of the perpetrator; identifying what the perpetrator needed or felt; “getting” the perpetrator; and identifying when the perpetrator posed a danger.

Reliability

To assess the IAS’s reliability, its internal consistency was evaluated. The internal consistency of the factors ranged from fair to high (.88 for Factor 1; .93 for Factor 2; .69 for Factor 3; and .72 for Factor 4), indicating acceptable reliability. The internal consistency of the total score was high (.91) indicating good reliability.

In accordance with these findings, four scores were computed as the average score of each factor, so that higher scores of Factor 1, Factor 2, Factor 3, and Factor 4 reflected higher levels of adopting the perpetrator’s experience; identifying with the perpetrator’s aggression; replacing one’s own agency with that of the perpetrator; and becoming hypersensitive to the perpetrator, respectively.

The correlations between the four factors were positive and significant, Factors 1 and 2 r =.65, p < .000; Factors 1 and 3 r =.40, p < .000; Factors 1 and 4 r =.36, p < .000; Factors 2 and 3 r =.35, p < .000; Factors 2 and 4 r =.31, p < .000; Factors 3 and 4 r =.41, p < .000. Analyses revealed that the IAS’s total score and four factors were not significantly associated with age, r = −.08, p =.18; r = −.04, p = .45; r = −.07, p =.22; r = −.04, p =.46; r = −.09, p =.11, respectively, or education, r = −.05, p =.43; r = −.03, p =.63; r = −.02, p =.77, r = −.04, p =.53; r = −.08, p =.17, respectively. Analyses indicated nonsignificant differences between women and men in levels of IAS’s total score and four factors, t(315) = −.24, p = .81; t(315) = .01, p = .99; t(315) = .29, p = .77; t(315) = −1.42, p = .15; t(315) = .11, p = .91, respectively.

As demonstrated by the results, the structure of the IAS is in accordance with the qualities of identification with the aggressor phenomenon and with its presentation in the literature (Frankel, 2002). Thus, the current findings support the IAS’s factorial validity.

Study 2

Study 2 had three main objectives: (a) confirming the IAS’s underlying factor structure; (b) assessing the IAS’s criterion validity by exploring its associations with dissociation, PTSD symptoms, and posttraumatic guilt; (c) assessing the IAS’s construct validity by examining the associations between the IAS factors and the history of childhood abuse, the severity of the childhood abuse, the recurrence of the childhood abuse, and the identity of the perpetrator. In line with the aforementioned review, four hypotheses were formed:

Hypothesis 1: Identification with the aggressor would be associated with elevated dissociation, PTSD symptoms, and posttraumatic guilt.

Hypothesis 2: Individuals who reported a history of childhood abuse would have higher scores of identification with the aggressor compared with nonabused individuals.

Hypothesis 3: Severity of the childhood abuse would be associated with higher levels of identification with the aggressor.

Hypothesis 4: The recurrence of the childhood abuse and the identity of the perpetrator would be linked with identification with the aggressor so that individuals who reported recurrent childhood abuse or abuse inflicted by a parental figure would report higher levels of identification with the aggressor compared with individuals who were survivors of single-episode abuse or abuse inflicted by a nonparental figure, respectively.

Method

Participants and procedure

We conducted an online survey of a convenience sample of students. Students were recruited via Israeli social media groups dedicated to students of colleges and universities in Israel. The survey was accessible through Qualtrics, a secure web-based survey data collection system. The survey took 15 min to complete, on average, and was open from August 20, 2017, to October 1, 2017. The survey was anonymous, and no data were collected that linked participants to recruitment sources. The Tel Aviv University Institutional Review Board (IRB) approved all procedures and instruments.

A total of 425 participants entered the survey site. Only 375 (88.23 %) completed the survey, and of those, only the 368 respondents who completed all of the questionnaires related to the current study were included in this study. The average age of the participants was 27.16 years (SD = 5.75). The majority of the sample, 85.6% (n = 315), were women. The average number of years of education was 14.82 (SD = 2.03). The majority of the sample, 71.5% (n = 263), were undergraduate students; 16.6% (n = 61) were graduate students; 2.7% (n = 10) were PhD students; and 7.3% (n = 27) were students studying in recognized vocational programs. Half of the sample, 53.0% (n = 195), were single, and 42.4% (n = 156) reported that they were currently in an intimate relationship. The majority of the sample, 88.3% (n = 325), reported that they were in a heterosexual relationship.

Measures

Background variables

Participants completed a brief demographic questionnaire that assessed gender, age, education, and relational status.

Childhood trauma questionnaire (CTQ)

This questionnaire includes 28 self-reported items that indicate childhood maltreatment, including physical abuse (e.g., hit hard enough to see a doctor), sexual abuse (e.g., was molested), emotional abuse (e.g., felt that parents wished he or she was never born), physical neglect (e.g., did not have enough to eat), and emotional neglect (e.g., felt loved—reversed item) (Bernstein et al., 2003). The items were rated on a 5-point, Likert-type scale with response options ranging from 1 (never true) to 5 (very often true). Participants were classified as having a history of abuse if they had scores which were higher than one of the cutoff scores suggested by Tietjen et al. (2010): physical abuse ≥ 8; sexual abuse ≥ 6; and emotional abuse ≥ 9. Two hundred sixty-four (71.74%) of the participants were classified as having a history of childhood abuse. In addition, a total score which reflects the severity of the abuse was calculated by summing all 28 items, after reversing the specified items. Cronbach’s alphas for this sample indicated a satisfactory to high internal consistency (.94), indicating acceptable reliability.

Features of abuse

Participants who had a history of childhood abuse were asked to specify the nature of the relationship between them and the perpetrator. Specifically, participants were asked to specify whether the perpetrator was a parental figure—and if so, a biological parent or a step-parent—or a nonparental figure (which could be any other family member or someone outside of the family). In addition, to evaluate whether the abuse was a recurrent event, they were asked whether the abuse had been a one-time event or whether it had occurred more than once.

IAS

This scale was previously described in Study 1.

Trauma-Related Guilt Inventory (TRGI)

Posttraumatic guilt was assessed via the TRGI (Kubany et al., 1996), a 32-item questionnaire made up of three scales: distress, global guilt, and guilt cognitions that specifically refer to guilt regarding the participants’ severest traumatic event. Cronbach’s alphas for this sample indicated a satisfactory to high internal consistency (.87 for distress; .91 for global guilt; .89 for guilt cognitions), indicating acceptable reliability.

The Dissociation Experiences Scale–II (DES-II)

Dissociation was assessed via DES-II, a self-report measure consisting of 28 items that assess the frequency of various daily life experiences of dissociative phenomena (Bernstein & Putnam, 1986). Respondents were asked to estimate the percentage of time that the various experiences occurred in their daily lives on an 11-point scale, ranging from 0% to 100%. The DES-II has been shown to have high validity and reliability (Frueh, Johnson, Smith, & Williams, 1996). Three subscales assessing absorption, depersonalization–derealization, amnesia, and a total score were computed. In this study, Cronbach’s alphas indicated a high internal consistency (.83 for absorption; .86 for depersonalization–derealization; .84 for amnesia; and .93 for total score).

PTSD Inventory (PTSD-I)

An updated version of the PTSD-I (Solomon et al., 1993) was used to assess PTSD symptoms according to the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013). Participants indicated the frequency with which they experienced items reflecting PTSD symptoms—consisting of reexperiencing, avoidance, negative alterations in cognition, and hyperarousal—on a 4-point Likert-type scale, ranging from a frequency of (1) least to (4) greatest. The PTSD-I has strong reliability and convergent validity (Solomon et al., 1993). In this study, a total score of PTSD symptoms was computed. Cronbach’s alpha indicated a high internal consistency of .95.

Data analysis

Overall, 0% to 28% of data was missing across variables. The MCAR test indicated that the data were missing completely at random, χ²(4470) = 43,195.68, p = 1.00. The ML estimation procedure was used for handling missing data (Collins et al., 2001). First, descriptive analyses of the items were examined. Next, confirmatory factor analysis (CFA) was conducted to confirm the underlying factor structure, obtained in Study 1, with the sample from Study 2, as recommended (Fabrigar, Wegener, MacCallum, & Strahan, 1999). Data were analyzed using structural equation modeling (SEM) techniques via AMOS software package (version 24). Multiple indicators (the scale’s items) for each latent variable (Factors 1-4) were used in the tested model. A bootstrap procedure was employed.

Several fit indices were used to examine the fit of the hypothesized model to the data: comparative fit index (CFI), normed fit index (NFI), Tucker–Lewis index (TLI), and root mean square error of approximation (RMSEA). For CFI, NFI, and TLI, values greater than .90 indicate a good fit between the model and the data (e.g., Arbuckle, 2007). For RMSEA, values of less than .05 represent a good fit (Browne, Cudeck, Bollen, & Long, 1993). Finally, chi-square was computed; however, because it is sensitive to sample size, we used the ratio of chi-square to degrees of freedom. Values below 3 indicate a good fit.

The IAS’s criterion validity was evaluated by a series of Pearson correlations between IAS’s factors and total score and dissociation, PTSD symptoms, and posttraumatic guilt. The IAS’s construct validity was evaluated by a series of t tests and Pearson correlations. First, we compared participants with a history of childhood abuse with participants without a history of childhood abuse in IAS scores. Next, we conducted Pearson correlations between severity of abuse and the IAS’s scores. Finally, we conducted a series of t tests, with recurrence of abuse (single experience vs. nonsingle experience) and the identity of the perpetrator (parental figure vs. nonparental figure) as the independent variables, and the four factors of the IAS as the dependent variables. These analyses were conducted among participants with a history of childhood abuse (sexual, physical, or emotional abuse).

Results

IAS’s structure

Table 2 presents means, standard deviations, skewness, and kurtosis of the IAS’s items. The hypothesized four-factor model is presented in Figure 1. All the predicted regression coefficients were significant (all ps < .001). The fit indices of the model indicated a good fit between the model and the data, χ²/df = 2.31, p < .001, NFI = .90, TLI = .92, CFI = .94, and RMSEA = .06, confirming the structure found in Study 1. Pearson correlations between the factors were significant, Factors 1 and 2 r = .59, p < .001; Factors 1 and 3 r = .50, p < .001; Factors 1 and 4 r = .47, p < .001; Factors 2 and 3 r = .38, p < .000; Factors 2 and 4 r =.35, p < .001; Factors 3 and 4 r = .35, p < .001.

Table 2.

Descriptive Statistics of IAS Items (Study 2).

	M	SD	Skewness		Kurtosis
	M	SD	Sk	SE	K	SE
4. Sometimes the needs and desires of people are similar to those of their perpetrator	31.17	29.53	.61	.13	−.83	.25
5. Some people feel that the point of view of their perpetrator is the right one	29.13	30.31	.75	.13	−.69	.25
16. Some people care about their perpetrator	47.42	33.18	−.06	.13	−1.30	.25
17. Some people understand their perpetrator	39.51	30.57	.18	.13	−1.11	.25
18. Some people feel that their perpetrator was right	29.84	29.29	.64	.13	−.77	.25
19. Some people feel a certain pleasure from the offense	20.43	28.16	1.38	.13	.94	.25
20. Some people feel emotions that are similar to those of their perpetrator	25.46	26.64	.87	.13	−.22	.25
21. Some people understand their perpetrator’s desire to hurt them	26.44	28.29	.79	.13	−.57	.25
29. Some people identify with their perpetrator	28.80	28.41	.67	.13	−.64	.25
25. Some people are afraid that they are as aggressive as their perpetrator	40.84	33.15	.25	.13	−1.26	.25
26. Some people are afraid that they resemble their perpetrator	40.27	34.28	.30	.13	−1.27	.25
27. Some people believe that they too can be as aggressive as their perpetrator	37.61	32.33	.38	.13	−1.17	.25
28. Some people feel that they and their perpetrator are in fact not so different from one another	30.54	29.92	.63	.13	−.83	.25
30. Some people feel that they behave like their perpetrator	29.08	29.08	.67	.13	−.76	.25
3. Some people do not know what they want in the presence of their perpetrator	50.14	30.99	−.19	.13	−1.07	.25
8. Some people feel “wiped out” in the presence of their perpetrator	62.26	32.59	−.67	.13	−.79	.25
12. Some people feel that they have no desires or needs	42.91	33.04	.18	.13	−1.22	.25
13. Some people act just like their perpetrator would want them to act	52.36	31.81	−.27	.13	−1.09	.25
15. Some people feel that the way their perpetrator would want them to feel	51.28	32.77	−.23	.13	−1.20	.25
7. Some people “read the thoughts” of their perpetrator	36.20	30.87	.42	.13	−1.01	.25
9. Some people sense what their perpetrator’s needs or feels	40.95	29.85	.24	.13	−.91	.25
10. Some people “get” their perpetrator accurately	39.97	29.46	.35	.13	−.80	.25
11. Some people can identify when their perpetrator poses a danger	53.78	29.56	−.19	.13	−.93	.25

Figure 1.

The results of the confirmatory factor analysis (Study 2).

Reliability

The internal consistency of the IAS’s factors ranged from fair to high (.88 for Factor 1; .93 for Factor 2; .77 for Factor 3; and .77 for Factor 4), indicating acceptable reliability. The internal consistency of the total score was high (.91), indicating good reliability.

Criterion validity

As hypothesized, Pearson correlations indicated that the total score and four factors of the IAS were significantly correlated with dissociation and PTSD symptoms (Table 3). The higher the levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator, the higher the levels of dissociation and total PTSD scores.

Table 3.

Pearson Correlations Between IAS Factors and Severity of Childhood Abuse, PTSD, Dissociation, and Guilt (Study 2).

	Total score	Factor 1	Factor 2	Factor 3	Factor 4
Severity of childhood abuse
Total score	.27***	.16**	.30***	.26***	.12*
Dissociation
Total score	.33***	.26***	.24***	.38***	.15**
Absorption	.26***	.18**	.16**	.36***	.13*
Depersonalization–derealization	.33***	.29***	.26***	.31***	.12*
Amnesia	.30***	.25***	.23***	.28***	.14**
PTSD symptoms
Total score	.24***	.16**	.20***	.26***	.12*
Posttraumatic guilt
Distress	.27***	.20***	.16**	.34***	.13*
Global guilt	.20***	.21***	.11*	.24***	.03
Guilt cognitions	.21***	.21***	.14**	.22***	.04

p < .05. **p < .01. ***p < .001.

Factors 1, 2, and 3 of the IAS were also significantly correlated with posttraumatic guilt (Table 3). The higher the levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, and replacing one’s agency with that of the perpetrator, the higher the levels of posttraumatic guilt. Factor 4 of the IAS was significantly correlated with the distress subscale of guilt; the higher the levels of becoming hypersensitive to the perpetrator, the higher the levels of guilt-distress.

Construct validity

As hypothesized, participants with a history of childhood abuse had higher IAS scores than nonabused participants. The levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator were higher among participants with a history of childhood abuse compared with nonabused participants, t(366) = −3.18, p < .01; t(366) = −3.26, p < .01; t(366) = −3.86, p < .01; t(366) = −2.71, p < .01, respectively.

As hypothesized, Pearson correlations indicated that severity of childhood abuse was significantly correlated with IAS scores (Table 3). The higher the severity of childhood abuse, the higher the levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator. A series of t tests revealed that participants who suffered from more than one episode of abuse had higher levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator than those who experienced a single episode of abuse, t(261) = 3.30, p < .01; t(252.19) = 5.75, p < .001; t(261) = 3.81, p < .001; t(261) = 3.33, p < .01, respectively.

Similarly, levels of adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, and becoming hypersensitive to the perpetrator were higher among participants who were abused by a parental figure compared with participants who were abused by a nonparental figure, t(261.73) = −2.63, p < .01; t(211) = −5.96, p < .001; t(211) = −3.03, p < .01, respectively. No significant differences were found between the groups in levels of replacing one’s own experience and agency with that of the perpetrator, t(211) = −.71, p = .48.

Discussion

Despite the theoretical and clinical writing on identification with the aggressor within the trauma field (e.g., Frankel, 2002), research on this concept is extremely scarce. This gap might be partly attributed to the lack of a measurement for assessing this phenomenon. The importance of the construction and validation of the IAS is, thus, clear.

The results of the present study support the psychometric properties of the IAS scale and indicate that the IAS may serve as a useful tool in exploring identification with the aggressor among adult survivors of childhood abuse. The current findings yielded four factors of the IAS, including adopting the perpetrator’s experience, identifying with the perpetrator’s aggression, replacing one’s agency with that of the perpetrator, and becoming hypersensitive to the perpetrator, which reflect the varied facets of the identification with the aggressor phenomenon (Ferenczi, 1933; Frankel, 2002).

The measure showed good internal consistency in its four factors. The present results also demonstrated that the IAS has good construct and criterion validity. In line with the theoretical literature, levels of identification with the aggressor, as measured by the IAS, were higher among participants who had a history of childhood abuse compared with nonabused participants; among participants who experienced more than a single episode of abuse compared with survivors of single-episode abuse; and among survivors of childhood abuse inflicted by a parental figure compared with survivors of childhood abuse inflicted by a nonparental figure. In addition, as hypothesized, levels of identification with the aggressor, as measured by the IAS, were significantly correlated with the severity of the childhood abuse, as well as with standard measures of dissociation, PTSD symptoms, and posttraumatic guilt.

The current findings indicated higher levels of identification with the aggressor among survivors who experienced more than a single episode of abuse or abuse inflicted by a parental figure, compared with survivors of a single episode of abuse or abuse inflicted by a nonparental figure. In addition, the findings indicated a significant relationship between the levels of identification with the aggressor and the severity of the childhood abuse, such that the higher the levels of abuse, the higher the levels of identification with the aggressor.

The present findings may reflect the protective function of identification with the aggressor and reveal how it becomes, over time, embedded in the survivors’ experience. Brutal abuse inflicted by the caregiver on the child poses an unbearable and menacing threat to the child. In the face of such a threat, emptying the self of its own thoughts and feelings, taking on the perpetrator’s experience and aggression as well as replacing one’s own agency with the perpetrator’s, are all reactions that might help the child preserve his or her self (Frankel, 2002; Howell, 2014) as well as his or her relational attachments, which are essential for survival (Davies & Frawley, 1994). Over time, this reaction of identification with the aggressor becomes entrenched in the child’s developing self (Frankel, 2002; Van der Kolk, 1989). Thus, even decades after the trauma, survivors of childhood abuse inflicted by their caretakers show this complex presentation of strong bonding with the perpetrator, and internalization of the perpetrator’s experience, alongside a subordination and nullification of the self.

Our findings revealed significant associations between identification with the aggressor, dissociation, PTSD symptoms, and posttraumatic guilt. Higher levels of identification with the aggressor were significantly correlated with elevated dissociation and PTSD symptoms, as well as with posttraumatic guilt. The present findings also demonstrated some of the potentially negative consequences of identification with the aggressor. To take on the perpetrator’s experience, the child splits off and dissociates large parts of his or her self (Howell, 2014). Although doing so enables an anticipation of the aggressor’s behavior and a prevention of abuse escalation, it also jeopardizes healing, and fosters a reliving of the past trauma (Frankel, 2002), as manifested in PTSD symptoms.

Adopting the aggressor’s experience may also bring forth feelings of self-blame and guilt over the abuse. As part of taking on the perpetrator’s beliefs, feelings, and needs, the child may introject the perpetrator’s “badness” or the perpetrator’s distorted views which regard the child as being the cause of his or her own abuse. In this way, the abused child may not only suffer from confusion and invalidation but may also experience guilt and self-blame over the abuse. Elevated levels of posttraumatic guilt have been documented among survivors of childhood abuse and have been found to be related to negative outcomes, such as increased psychopathology, self-harm, and revictimization (Filipas & Ullman, 2006; Peters & Range, 1997). A variety of mechanisms have been offered to explain why guilt is so often found among survivors of childhood abuse (Whiffen & MacIntosh, 2005). The present findings imply that identifying with the aggressor may be involved in the formation of guilt among childhood abuse survivors and that exploring identification with the aggressor in regard to guilt could expand our understanding of this phenomenon.

The results of the present study should be considered in light of its limitations. First, this study was based on self-report measures which may have been subject to response biases and shared method variances. Second, the present study did not include data regarding the onset of abuse, and recurrence of the abuse was measured by a single item that was created for the purpose of this study. Given that early onset of chronic abuse might negatively affect the development of the self, future studies should explore the role of the onset of childhood abuse as well as its chronicity in shaping one’s identification with the aggressor, while utilizing widely used methods. Third, despite the relatively large sample size in both studies, possible implications of the use of convenience sampling and of student participants only should also be considered. Furthermore, there was an overrepresentation of females in the present sample, and the data regarding participants’ ethnicity were lacking. These factors limit the generalizability of the study and point to the need to explore the phenomenon of identification with the aggressor among a variety of populations, and specifically among clinical samples of male and female survivors of childhood abuse, with varied cultural backgrounds. Finally, due to the study’s cross-sectional design, readers should be cautious in assuming causal relationships between the study variables. Future longitudinal studies should explore the IAS’s predictive validity, its stability over time, and its sensitivity to changes that may result from specific clinical interventions that are geared toward working through identification with the aggressor among childhood abuse survivors.

Notwithstanding the limitations above, the present results indicate that the IAS has good psychometric characteristics, supporting the feasibility of its use in measuring identification with the aggressor. Prior to the development of the IAS, there had been no reliable or valid means of quantifying identification with the aggressor. Victims of abuse may be deeply connected to their abusers and appear to take on their perpetrator’s experience. This very real phenomenon could pose a significant obstacle to their recovery and may lead to a perpetuation of abusive interactions with others and/or reenactments of the traumatic past (Van der Kolk, 1989). Although Ferenczi’s theory mainly focuses on childhood abuse (Frankel, 2002), it may also explain the affiliative attachment between victims of other types of abuse (e.g., domestic abuse or captivity) and their perpetrators. Such distorted relations are often labeled traumatic bonding (Dutton & Painter, 1993) or Stockholm syndrome (Graham, 1994). Hence, exploring the phenomenon of identification with the aggressor might make an important contribution to our understanding and treatment of survivors of childhood abuse as well as survivors of different types of abuse during adulthood.

Footnotes

Acknowledgements

The authors thank Prof. Eli Somer, Dr. Zivya Seligman, Dr. Daniela Shebar-Shapira, and Avigail Frenkel (MSW) for their help in examining the scale’s content validity and Dr. Talia Meital Schwartz-Tayri for her help in participants’ recruitment.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Yael Lahav

Anat Talmon

Author Biographies

Yael Lahav, PhD, Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, is a practicing clinical psychologist. Her research interests include traumatic stress and the psychological sequelae of childhood sexual abuse and war captivity.

Anat Talmon, PhD, is a postdoctoral fellow at the Psychology department at Stanford University. Her research focuses on the long-term implications of childhood maltreatment on both adult survivors’ body and mind.

Karni Ginzburg, PhD, is professor at the Bob Shapell School of Social Work, Tel Aviv University, the head of the MSW program for Social Work in Stress and Trauma Situations, and the head of the School’s Doctoral Program. Her research is focused on the reciprocal relations between exposure to traumatic stress, and emotional and somatic manifestations of distress.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental health disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Arbuckle

J. L.

(2007). Amos 16.0 user’s guide. Spring House, PA: Amos Development Corporation.

Bernstein

E. M.

Putnam

F. W.

(1986). Development, reliability, and validity of a dissociation scale. Journal of nervous and mental disease, 174, 727-735.

Bernstein

D. P.

Stein

J. A.

Newcomb

M. D.

Walker

Pogge

Ahluvalia

. . . Zule

(2003). Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse & Neglect, 27, 169-190.

Browne

M. W.

Cudeck

Bollen

K. A.

Long

J. S.

(1993). Alternative ways of assessing model fit. Sage Focus Editions, 154, 136-136.

Carlson

E. B.

Putnam

F. W.

(1993). An update on the dissociative experiences scale. Dissociation: Progress in the Dissociative Disorders, 6(1), 16-27.

Classen

Koopman

Spiegel

(1993). Trauma and dissociation. Bulletin of the Menninger Clinic, 57, 178-194.

Coates

S. W.

Moore

M. S.

(1997). The complexity of early trauma: Representation and transformation. Psychoanalytic Inquiry, 17, 286-311.

Collins

L. M.

Schafer

J. L.

Kam

C.-M.

(2001). A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological Methods, 6, 330-351.

10.

Davies

J. M.

Frawley

M. G.

(1994). Eight transference countertransference positions. In Davies

J. M.

Frawley

M. G.

(Eds.), Treating the adult survivor of childhood sexual abuse (pp. 167-187). New York, NY: Basic Books.

11.

Dutton

D. G.

Painter

S. L.

(1993). Emotional attachments in abusive relationships: A test of traumatic bonding theory. Violence and Victims, 8(2), 105-120.

12.

Fabrigar

L. R.

Wegener

D. T.

MacCallum

R. C.

Strahan

E. J.

(1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4(3), 272-299.

13.

Feiring

Taska

Chen

(2002). Trying to understand why horrible things happen: Attribution, shame, and symptom development following sexual abuse. Child Maltreatment, 7(1), 25-39.

14.

Ferenczi

(1932). The clinical diary of Sandor Ferenczi ( Dupont

, Ed., Balint

Jackson

N. Z.

, Trans.). Cambridge, MA: Harvard University Press.

15.

Ferenczi

(1933). Confusion of tongues between adults and the child ( Mosbacher

, Trans.). In Balint

(Ed.), Final contributions to the problems and methods of psycho-analysis (pp. 156-167). London, England: Karnac Books.

16.

Filipas

H. H.

Ullman

S. E.

(2006). Child sexual abuse, coping responses, self-blame, posttraumatic stress disorder, and adult sexual revictimization. Journal of Interpersonal Violence, 21, 652-672.

17.

Frankel

(2002). Exploring Ferenczi’s concept of identification with the aggressor: Its role in trauma, everyday life, and the therapeutic relationship. Psychoanalytic Dialogues, 12, 101-139.

18.

Freud

(1936). The ego and the mechanisms of defense. New York, NY: International Universities Press.

19.

Frueh

B. C.

Johnson

D. E.

Smith

D. W.

Williams

M. A.

(1996). A potential problem with the response format of the Dissociative Experiences Scale: A significant correlation with intelligence among combat veterans with PTSD. Journal of Traumatic Stress, 9, 651-656.

20.

Graham

L. R.

(with Rawlings

E. I.

Rigsby

R.K.

). (1994). Loving to survive: Sexual terror, men’s violence and women’s lives. New York: New York University Press.

21.

Hayton

J. C.

Allen

D. G.

Scarpello

(2004). Factor retention decisions in exploratory factor analysis: A tutorial on parallel analysis. Organizational Research Methods, 7, 191-205.

22.

Howell

E. F.

(2014). Ferenczi’s concept of identification with the aggressor: Understanding dissociative structure with interacting victim and abuser self-states. The American Journal of Psychoanalysis, 74(1), 48-59.

23.

Jülich

(2005). Stockholm syndrome and child sexual abuse. Journal of Child Sexual Abuse, 14, 107-129.

24.

Kubany

E. S.

Haynes

S. N.

Abueg

F. R.

Manke

F. P.

Brennan

J. M.

Stahura

(1996). Development and validation of the Trauma-Related Guilt Inventory (TRGI). Psychological Assessment, 8, 428-444.

25.

Papalia

Ogloff

J. R.

Cutajar

Mullen

P. E.

(2018). Child sexual abuse and criminal offending: Gender-specific effects and the role of abuse characteristics and other adverse outcomes. Child Maltreatment, 23, 399-416.

26.

Peters

D. K.

Range

L. M.

(1997). Self-blame and self-destruction in women sexually abused as children. Journal of Child Sexual Abuse, 5(4), 19-33.

27.

Solomon

Benbenishty

Neria

Abramowitz

Ginzburg

Ohry

(1993). Assessment of PTSD: Validation of the revised PTSD Inventory. The Israel Journal of Psychiatry and Related Sciences, 30, 110-115.

28.

Tietjen

G. E.

Brandes

J. L.

Peterlin

B. L.

Eloff

Dafer

R. M.

Stein

M. R.

. . . Aurora

S. K.

(2010). Childhood maltreatment and migraine (part I). Prevalence and adult revictimization: A multicenter headache clinic survey. Headache: The Journal of Head and Face Pain, 50(1), 20-31.

29.

Van der Kolk

B. A

. (1989). The compulsion to repeat the trauma. Psychiatric Clinics of North America, 12, 389-411.

30.

Walker

H. E.

Freud

J. S.

Ellis

R. A.

Fraine

S. M.

Wilson

L. C.

(2017). The prevalence of sexual revictimization: A meta-analytic review. Trauma, Violence, & Abuse, 20, 67-80. doi:10.1177/1524838017692364

31.

Whiffen

V. E.

MacIntosh

H. B.

(2005). Mediators of the link between childhood sexual abuse and emotional distress: A critical review. Trauma, Violence, & Abuse, 6(1), 24-39.