Sexual Violence—“Victim” or “Survivor”: News Images Affect Explicit and Implicit Judgments of Blame

Victims or Survivors?

There is an ongoing debate in both science and feminist circles about how to call a woman who has experienced sexual violence. The term victim has been long-established, especially in scientific literature. It is debated, however, whether this term might have a negative impact on women who have been raped and whether the alternative term survivor should be adopted instead. Dunn (2005) has examined the victim/survivor terminology with respect to women who stay in an abusive relationship. She writes that “victims are presented as trapped, and survivors, conversely, are shown as making choices, they are constructed in ways that place them at opposite poles of an agency continuum” (p. 2). She stresses that one of the central attributes of the victim construct is innocence, and that “real” victims do not have any part in their victimization. Another important issue is that the term victim implies weakness, vulnerability, and lack of agency (Thompson, 2000). These meanings may influence the self-perception of women who have experienced sexual violence. For example, McCarthy (1986) suggests that for a woman who adopts the victim label, the experience of sexual violence “becomes the controlling and dominating event in her life and controls her sexual (or personal) self-esteem” (pp. 323-324).

The term survivor, on the contrary, is often associated with strength and recovery from the experience (Thompson, 2000). Kelly (1988) describes the emergence of “survivor” as a replacement for “victim” within feminist discourse in the 1980s. Referring to interviews she conducted with women who have been raped, Kelly writes that “the focus on coping, resistance and survival reflects the experiences of the woman interviewed” and stresses the importance of “shifting the emphasis from viewing women as passive victims of sexual violence to seeing them as active survivors” (p. 163). However, a problem may be that women who call themselves survivors, and thus express that they have overcome the trauma brought on to them by sexual violence, may encounter difficulties if they want to address or even review this traumatic experience in the future, as it might lead to invalidating their constructed self-image as a survivor (Thompson, 2000). Moreover, it might be a better strategy to frame oneself as a victim to stress innocence and vulnerability, for example, in the context of a court trial. However, to recover from the experience and create a positive self-identity, adopting the term survivor might be the better strategy (Thompson, 2000). In a qualitative study, Patterson and Campbell (2010) found that women who felt empowered, like a survivor, were more likely to follow through with the criminal justice proceedings after reporting a rape.

Ascribed Meanings of the Terms Victim/Survivor and Their Effects on Target Blaming

Recent quantitative research has addressed more directly the meanings that people ascribe to the terms victim and survivor, respectively, as well as the terms’ effects on judgments of blame. In one study, U.S. undergraduates were asked to list five characteristics of either a rape victim or a rape survivor (Hockett, McGraw, & Saucier, 2014, Study 2). The results revealed differences in participants’ responses regarding content, valence, and attributions of responsibility. Specifically, the students ascribed more stable personality attributes to a rape victim, but more process-related and outcome-related attributes to a rape survivor. Also, more positive attributes were ascribed to survivors than to victims. These findings suggest that the topic of (successful) coping, which women who have been raped emphasize when adopting a survivor identity (e.g., Thompson, 2000), is also prominent in the perceptions of a more general sample. Furthermore, Hockett and her colleagues observed a gender difference regarding implicit attributions of blame and responsibility: When thinking about rape survivors (compared with rape victims), male students listed more attributes that inherently blamed the perpetrator (e.g., “scared”), whereas no such difference emerged for female students.

In a follow-up study, Hockett et al. (2014, Study 4) examined more directly the effects of presenting the labels rape victim versus rape survivor in an otherwise identical rape vignette, on attributions of target blame (e.g., “Jenna [the target] deserves what happened to her”). They found a marginal effect of the label condition, with the “rape victim” being blamed somewhat more than the “rape survivor.” This difference between the victim and survivor conditions was not replicated, however, in follow-up studies (reported in Hockett, 2013).

Another approach to measuring the meaning of the terms victim and survivor was taken by Papendick and Bohner (2017). In two studies, conducted in parallel in English and German, participants read brief rape vignettes that contained either the term victim or the term survivor but were otherwise identical. Then they were asked to rate the meaning of the term they had read on 15 bipolar items, such as weak—strong, passive—active, and bad—good (see Osgood, Suci, & Tannenbaum, 1971). Results for both language samples showed that the term survivor was rated to be much more positive, active, and strong than the term victim, again echoing the conceptualizations from feminist discourse. In their second study, Papendick and Bohner varied the labels within assessments of victim blame regarding a fictitious rape scenario. Although participants’ judgments of target blame strongly depended on their general rape-related attitudes, the authors did not observe any consistent effects of the label conditions on perceived target blame. Across several studies, furthermore, both male and female participants considered the term victim to be more appropriate than the term survivor when referring to a woman who has been raped (Hockett et al., 2014, Study 3; Papendick & Bohner, 2017, Studies 2 and 3).

Overall, there is converging evidence for clear-cut differences in the perceived meanings of the verbal labels, victim and survivor. Although generally seen as less appropriate, the term survivor connotes strength, coping, and positive attributes, whereas the term victim connotes passivity, weakness, and more negative attributes. Effects of the labels on attributions of blame appear to be less consistent, but when such effects emerged, they were also in favor of “survivors,” who tended to be blamed less than “victims.” These latter effects may be more reliable for male participants and may emerge particularly on more indirect measures (as in Hockett et al., 2014).

Media Effects on Perceptions of Sexual Violence

Before turning to our own research, which will deal with news images of victims and survivors, we will first address research on news media effects more broadly. Several studies have examined linguistic or content properties of texts surrounding sexual violence. For example, Henley, Miller, and Beazley (1995) found that American newspaper writers used the passive form of the verb “raped” much more frequently (70%) than the active (30%), in comparison with more positive verbs (e.g., thanked) where the use of the passive was less frequent. The authors interpreted this style of writing as a tendency to exonerate perpetrators of sexual violence by shifting attention away from the presumed perpetrator and toward the target; this is especially the case for agent-less passives such as “she was raped” (see also Bohner, 2001).

Shifting the focus to text content in the news media, researchers investigated the prevalence and impact of rape myths being expressed in the content of newspaper articles about sexual violence (Franiuk, Seefelt, Cepress, & Vandello, 2008). Rape myths may be defined as beliefs about rape that serve to deny, downplay, or justify male sexual violence against women (Gerger, Kley, Bohner, & Siebler, 2007). Results showed that one third of the articles in Franiuk and her colleagues’ sample did contain rape myths. Furthermore, participants who were exposed to myth-endorsing articles (vs. articles without rape myths) were more likely to exonerate the perpetrator and blame the target of violence. Similar findings were reported by Franiuk, Seefelt, and Vandello (2008) for the effects of rape myths in newspaper headlines on readers’ attitudes and social judgments: When male participants were exposed to rape-myth containing headlines (vs. headlines without rape myths), they were more likely to endorse rape-supportive attitudes. In a similar line of research, Hollander and Rodgers (2014) investigated how the news media present information about women’s resistance to sexual assault. Their findings showed that in most news articles about sexual violence, women are described as weak and incapable of defending themselves. The authors argue that this kind of one-sided presentation of women as vulnerable victims reinforces stereotypical gender beliefs.

In the present research, we aim to extend this line of inquiry into news media representations of women who have experienced sexual violence. However, instead of examining the perceptions and effects of particular terms, texts, or types of language, we will be using images that depict victims versus survivors, taken from online news articles.

Photographs in News Media

Several studies have examined the role that photographs play in newspapers, as well as in the news media in general. Zillmann, Gibson, and Sargent (1999) found that participants’ assessment of issues presented in news-magazine reports was biased in favor of the implications given by the photographs accompanying the reports.

Sargent (2007) replicated these findings by investigating image effects on selective exposure to news stories. In her study, the inclusion of threatening images accompanying news articles resulted in significantly longer self-exposure time to following text sections, even if these sections did not have an accompanying image. The author argued that “the threatening image produced an affective reaction in readers that stimulated greater and more deliberate cognitive processing of following text that was devoid of an image” (p. 720). Journalists might deliberately employ such reactions to “sensational” images to encourage readers to engage with their articles.

These findings show that it is important to investigate how a certain kind of image affects readers’ perceptions of sexual violence reports, and to take these effects into account when composing news articles/reports on sexual violence. Schwark (2017) investigated the kind of photographs shown alongside articles about sexual violence in online news media. In a qualitative content analysis, the author found that the vast majority of images in her sample perpetuate stereotypes about women who have experienced sexual violence: These women are portrayed as weak and helpless, as opposed to strong and with agency. A typical photograph from this sample would show a woman in a dark alleyway, crouching on the floor with her arms covering her face. In the current study, we refer to this kind of photograph as a victim image, because it appears to be associated with characteristics from the victim terminology. In contrast, a survivor image portrays women who have experienced sexual violence as being more active and possessing greater self-efficacy. An example would be a photograph that shows a self-help group of sexual violence survivors or a woman in a session with a female counselor.

Explicit and Implicit Measures in Research on Sexual Violence

Investigating social judgments related to gender, and especially perceptions of sexual violence against women, requires valid research methods. Although explicit, questionnaire-based measures have been widely and successfully used, some aspects of attitudes may not be accessible to conscious introspection (Glaser et al., 2015) and thus impossible to detect with explicit self-report measures. Gender researchers, therefore, have started using implicit measures as well, with promising results (e.g., Rudman & Kilianski, 2000).

In research on gender violence, more specifically, Widman and Olson (2013) have shown that implicit rape-related attitudes (assessed with an evaluative priming procedure) explained unique variance in men’s self-reported sexually aggressive behavior even after controlling for explicit attitudes. In a similar vein, Nunes, Hermann, and Ratcliffe (2013) examined associations between men’s sexual aggression (both past behavior and rape proclivity) and both explicit and implicit attitudes toward rape, the latter being measured with an implicit association test (IAT; Greenwald, McGhee, & Schwartz, 1998). They, too, found that implicit and explicit attitudes provided independent information and jointly were more strongly associated with sexual aggression than each measure on its own.

Finally, and highly pertinent to the present research, Süssenbach, Albrecht, and Bohner (2017) have devised and implemented an IAT to assess participants’ implicit evaluations of target and perpetrator in a rape vignette. Their results showed that these implicit evaluations were moderately correlated, but not redundant, with explicit case-related judgments. Also, the implicit evaluations were affected by variations in the content of the rape vignette; specifically, they were more anti-target (and pro-perpetrator), if the vignette contained features related to common rape myths (e.g., the perpetrator being an acquaintance; the target being intoxicated and dressed “provocatively”). In sum, adding implicit measures to assess attitudes regarding sexual violence may help increase the precision of research designs and yield unique information not obtainable by explicit measurement. In our own research, therefore, we have included an IAT measure inspired by Süssenbach, Albrecht, and Bohner (2017). We will return to this in the “Method” section of our main study.

The Present Research

Based on the accounts and findings discussed in the Introduction, we set out to study the impact of exposure to pictures showing either victims or survivors on participants’ explicit and implicit judgments of a rape case. To manipulate exposure to the pictures, we devised a priming procedure. As noted by Klauer (2009), “a priming effect occurs when the processing of a target stimulus is influenced by a preceding stimulus on the basis of a particular relationship between prime and target” (p. 200). In our current research, the target stimulus is a fictitious rape scenario; the preceding stimuli are the victim and survivor pictures. We assumed that presenting the pictures would activate distinct associations between these visual stimuli and their interpretations: Specifically, we predicted that women depicted as survivors would be associated with more positive attributes than women depicted as victims.

The level of accessibility of a concept depends on how strongly and how recently the concept has been activated (Moskowitz, 2005). Accordingly, in our priming procedure, we gave participants ample time to engage with the victim versus survivor images (to increase activation strength) and asked them to make case-related judgments shortly afterward (to ascertain recency of activation).

These were our research questions for the current research:

Furthermore, as some previous results point to effects of labels being stronger for men than for women, we examined whether such effects might depend on participants’ gender.

Pilot Study

Two sets of picture stimuli were available from previous research (Schwark, 2017). The victim pictures portray women after an assault has taken place, showing them in passive, vulnerable positions such as crouching on the floor with their arms shielding their bodies. The survivor pictures also show women who have been raped, but do so in more empowering situations, such as at a counselor’s office or at a self-help group with other women.

Main Study

In our main study, we aimed to investigate how participants’ exposure to sets of victim versus survivor images would affect their judgments of victim blame regarding a rape vignette. To further expand existing research on this topic, we used an IAT as an implicit measure of victim blaming, in addition to explicit rating items.

Method

Participants

Students (n = 236) were recruited at a German university and invited to take part in the study at a computer laboratory. Their mean age was 21.55 years (SD = 5.22). Of the total number of participants, 125 were female, 109 were male, and two did not indicate a binary gender identity. The latter two cases were not included in analyses involving gender.

Design and procedure

Participants were told that they would take part in a material pretest and an unrelated study afterward. They were randomly assigned either to a victim priming condition or to a survivor priming condition, in which they were exposed to color prints of the five victim pictures or the five survivor pictures from Part 2 of the pilot study. Participants were told that pictures would be needed for a university website featuring an article about sexual violence, and that their task was to rank the five photographs according to their suitability for use in the alleged article. This procedure was designed to ensure that participants thoroughly looked at and engaged with the images, to prime them with either the victim or the survivor content.

When the picture-ranking task was completed, the experimenter asked participants to complete the “second study” at a computer. They read a vignette describing a fictitious case of acquaintance rape, followed by a questionnaire asking about how much blame they assign to the victim and the perpetrator in the vignette. This was followed by a target–perpetrator IAT. Finally, we assessed participants’ acceptance of modern myths about sexual aggression (AMMSA; Gerger et al., 2007). This was done because AMMSA had been shown to correlate with both explicit and implicit measures of victim blaming in past research (Süssenbach, Albrecht, & Bohner, 2017), and we wanted to examine this correlation once more.

Rape vignette

All participants read the same vignette, describing how several colleagues, among them the victim (Tanja K.) and the perpetrator (Georg S.), went out for after-work drinks at a bar. Tanja K. ended up being slightly intoxicated and Georg S. drove her home. He entered her apartment unbidden and started kissing her. Georg S. proceeded further to hold Tanja K. down on the sofa and started undressing her. She told him to stop but did not fight back. He did not stop and ended up having sex with her. Afterward, he got dressed and left, while Tanja K. was very distressed and did not know what to do. It took her 2 weeks to tell a friend what had happened.

Explicit victim blaming

We assessed blaming of the female with five questions, each accompanied by a response scale from 1, not at all, to 7, very (much): “How much do you think the woman is to blame for what happened?,” “How much do you think the man is to blame for what happened?” (reverse-scored), “Do you think what happened was rape?” (reverse-scored), “How severe do you think the possible psychological consequences for the woman might be?” (reverse-scored), and “To what extent do you think Tanja K. made it all up?” Cronbach’s alpha for these items was .72, so they were averaged into an index of explicit victim blaming (overall M = 2.10, SD = 0.80).

Implicit victim blaming

Our procedure was equivalent to that of Süssenbach, Albrecht, and Bohner (2017). We used an IAT (Greenwald et al., 1998) as an indicator of implicit associations regarding the male and female protagonists of the rape vignette. Participants were asked to categorize stimulus words that represent different categories by pressing the “E” and “I” keys on the computer keyboard. The position of the superordinate category on either the left or the right side of the screen determined the correct response to an item. In our IAT, there were two target categories (man and woman) and two evaluative categories (good and bad). Participants were instructed that the man and woman categories referred to the protagonists of the rape case they had read before. Three stimulus words were used for each category. For the evaluative category good, the words were innocent, right, and friendly; for the evaluative category bad, the words were false, guilty, and mean. For the target category woman, the words were Ms. K., she, and Tanja; and for the target category man, the words were Mr. S., he, and Georg. In the critical blocks of the IAT, one target category and one evaluative category shared a response key (“I” or “E”). For example, in one critical block, the woman and good words shared the same key, whereas in the other critical block the man and good words shared the same key. In total, the IAT consisted of five blocks. Before each critical block, there were practice blocks for participants to familiarize themselves with the stimuli and the procedure. For a schematic illustration of the IAT procedure, see Figure 1.

OPEN IN VIEWER

Figure 1.

Schematic illustration of the implicit association test.

To capture implicit blaming of the female target, we computed a D score for each participant (Greenwald, Nosek, & Banaji, 2003). It was derived from the difference in reaction times between the two critical blocks, and coded in such a way that higher D scores represented more target blaming (overall M = −0.45, SD = 0.52). Because of a programming error, IAT data were not available for the first 17 participants.

Acceptance of modern myths about sexual aggression

Participants completed the AMMSA scale (Gerger et al., 2007). They were asked to indicate their degree of agreement (response scales from 1, I completely disagree, to 7, I completely agree) with each of the scale’s 30 items (e.g., “If a woman invites a man to her home for a cup of coffee after a night out this means that she wants to have sex”; “Many women tend to exaggerate the problem of male violence”). Cronbach’s alpha for this scale was .90, so the mean across the 30 items was used as a participant’s AMMSA score (overall M = 2.98, SD = 0.82).

Results

Correlation analyses (Table 1) showed that participants’ explicit and implicit judgments of victim blaming were moderately related. Also, participants with higher AMMSA showed higher scores on both victim-blaming measures, and men exhibited more explicit and implicit victim blaming than did women. The latter effect of participants’ gender was stronger on the implicit than the explicit measure of victim blaming. All of these findings replicate results of Süssenbach, Albrecht, and Bohner (2017). The stronger association of gender with implicit than explicit victim blaming suggests that, in addition to victim blaming, the IAT score may also reflect a more general preference for one’s gender in-group. There were, however, no significant correlations of either victim-blaming measure with the priming condition, indicating the absence of main effects exerted by the type of images participants had been exposed to.

OPEN IN VIEWER

Table 1.

Zero-Order Correlations Among Explicit and Implicit Victim Blaming, AMMSA, Priming Condition, and Participant Gender (Main Study).

	ImplicitVB	ExplicitVB	AMMSA	Priming condition
Explicit VB	.23***
AMMSA	.14*	.55***
Priming condition a	−.01	−.01	.03
Gender b	.46***	.25***	.24***	−.02

Note. AMMSA = acceptance of modern myths about sexual aggression; VB = victim blaming. Because of occasional missing values, the number of cases for these analyses varied between 219 and 236.

a

coded 1 = victim picture, 2 = survivor picture.

b

coded 1 = female, 2 = male.

*

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

Furthermore, male participants had significantly higher AMMSA scores (M = 3.19) than female participants (M = 2.79), t(232) = 3.81, p < .001. This is in line with previous findings on the AMMSA (Gerger et al., 2007) and, more generally, on gender differences in attitudes toward sexual violence (e.g., Suarez & Gadalla, 2010).

To test for possible interaction effects of the priming manipulation and participant gender, we z transformed the implicit and explicit measures of victim blaming to make them commensurable. Then we conducted a mixed-model 2 × 2 × 2 analysis of variance with priming condition and participant gender as between-subjects factors, and the z transformed explicit and implicit victim-blaming scores as two levels of a within-subjects factor. One result of this analysis of variance was a strong and significant main effect of gender, F(1, 214) = 56.53, p < .001. As had already been apparent in the correlation analyses, women showed less victim blaming than did men on both measures and across priming conditions (for condition means, see Table 2).

OPEN IN VIEWER

Table 2.

Means and Standard Deviations of Implicit and Explicit Victim Blaming (z Scores) by Participant Gender and Priming Condition (Main Study).

Victim blaming measure	Priming condition	Gender
		Male		Female
		M	SD	M	SD
Implicit	Victim	0.65	1.07	−0.40	0.68
	Survivor	0.33	1.04	−0.47	0.76
Explicit	Victim	0.13	0.88	−0.17	0.99
	Survivor	0.39	1.12	−0.33	0.87

Furthermore, and more importantly, there was a significant three-way interaction effect of gender, priming condition, and type of measure, F(1, 214) = 3.96, p = .048. To decompose this effect, we analyzed the simple interaction of priming condition by type of measure separately for each gender. For male participants, this interaction was significant, F(1, 98) = 4.86, p = .03. Specifically, on the implicit measure, men blamed the woman more in the victim condition (M = 0.65; SD = 1.07) than in the survivor condition (M = 0.33, SD = 1.04), whereas on the explicit measure, they blamed the woman more in the survivor condition (M = 0.39, SD = 1.12) than in the victim condition (M = 0.13, SD = 0.88). For female participants, on the contrary, neither the simple main effect of condition nor the condition by type of measure interaction was significant, both p > .30. Descriptively, female participants generally showed somewhat greater victim blaming in the victim condition (explicit M = 0.17, SD = 0.99; implicit M = −0.40, SD = 0.68) than in the survivor condition (explicit M = 0.33, SD = 0.87; implicit M = −0.47, SD = 0.76).

Discussion

The aim of our study was to examine whether priming with victim versus survivor pictures, as they appear in news media, may influence attributions of blame on explicit as well as implicit measures in a rape vignette. In a pilot study, we were able to show that, generally, women shown in survivor pictures were perceived more positively than women shown in victim pictures. This replicates findings of Hockett et al. (2014) as well as Papendick and Bohner (2017), who have examined the terms victim and survivor, demonstrating more positive associations with the latter. Also replicating previous research, we observed that several of our participants found the term survivor inappropriate in the context of sexual violence; nonetheless, the term’s visual representations yielded more positive associations than those of the term victim. In one of our pilot studies, we were able to show that survivor is also linked to more positive characteristics, when compared with victim. Women shown in pictures associated with survivor were described more positively overall compared with women shown in pictures associated with victim. This finding replicates results by Hockett et al. (2014) who found that participants ascribed more positive characteristics to the label rape survivor than rape victim.

The term victim is often associated with a lack of agency and a state of vulnerability (Thompson, 2000). McCarthy (1986) suggested that if a woman who has experienced sexual violence adopts the term victim for herself, the experience becomes the controlling incident in her life and has a negative impact on her self-esteem. The term survivor, however, is often associated with recovery and strength (Thompson, 2000). It implies recovery and agency, with a focus on coping with regards to the experience (Kelly, 1988). These textual-based findings seem to be applicable to photographic representations of the terms as well. The women portrayed in our survivor images were rated to be more strong, optimistic, brave, and forward-looking, compared with the women in the victim images.

These findings have important implications for both theory and practice. In a literature review, Hockett and Saucier (2015) compared academic articles regarding the use of normative (victim) and nonnormative (e.g., survivor) framings of women who have experienced rape. They argue that the nonnormative framing tends to maintain aspects of negative outcomes while stressing the positive ones (Hockett & Saucier, 2015), whereas the normative framing exclusively focuses on negative outcomes. This has implications for future research, whether it addresses the terminology (“victim” vs. “survivor”) or the visual representation. To avoid replicating common and potentially harmful stereotypes, researchers might want to take greater care when choosing their stimulus materials, both at the verbal level and at the level of pictorial stimuli.

The descriptive and correlational findings of our main study replicate existing data regarding gender effects on attitudes toward sexual violence (e.g., Gerger et al., 2007; Suarez & Gadalla, 2010); they also point to the benefits of employing implicit measures in addition to explicit measures in research on gender violence (Nunes et al., 2013; Widman & Olson, 2013) and especially victim blaming (Süssenbach, Albrecht, & Bohner, 2017). The IAT provides a measure that taps into more spontaneous or even unconscious attitudes. Assuming that sexual violence may represent an area in which respondents are not necessarily fully aware of their attitudes, and where attitudes may be subject to social-desirability concerns, it may come as no surprise that the implicit–explicit correspondence we observed was only moderate (for discussion, see Hofmann, Gschwendner, Nosek, & Schmitt, 2005) and that implicit and explicit blame showed partly divergent result patterns.

Our three-way interaction pattern suggested that male (but not female) participants ascribed more implicit blame to the woman, and less implicit blame to the perpetrator, in the victim (vs. survivor) condition. In this we see a parallel to findings by Hockett et al. (2014, Study 2), who found that men, but not women, generated fewer free associations that blamed the perpetrator when thinking about rape victims rather than rape survivors. Although Hockett and colleagues’ method was not an implicit measure per se, it did constitute a rather subtle and indirect assessment of blame.

We argue that such higher attributions of blame to victims shown by men at an implicit level might be explained by the operation of an existing schema that triggers automatic associations between victim images and blame attribution. Because the kind of survivor pictures we used in our materials are not usually shown in the press when reporting on sexual violence (Schwark, 2017), there should be no existing associations between the pictures and attributions of blame in rape cases that could be triggered.

Conversely, the higher levels of explicit blame that men exhibited in the survivor condition suggest a more deliberate and motivated underlying process. At least when being asked directly, men who see images of survivors (i.e., strong women who contradict the common stereotype of a woman who has been raped) may experience threat with respect to their male identity or, more generally, their belief in a just world (see Bohner et al., 2009). To counter these threats, men may employ conscious construals of blame. Further research is needed to examine this interpretation more directly.