Email this article 
This is an open access article distributed under the terms of the Creative Commons
Attribution License (Introduction [TOP]
Empathy is a socio-cognitive dimension with a major role in interpersonal relationships: it allows individuals to understand distinct cognitive and affective states, predisposing to the development of positive attitudes towards others, particularly in adverse situations (Batson, 1991; Tarrant, Dazeley, & Cottom, 2009). In the context of rape, empathy is the ability to understand the perspective, emotions and responses of the victim and/or perpetrator (Deitz, Blackwell, Daley, & Bentley, 1982; Smith & Frieze, 2003). This is considered a significant predictor of attitudes towards rape, given that it influences perceptions regarding the victim and the offender (Deitz et al., 1982; Deitz, Littman, & Bentley, 1984). This concept is far more complex than it first appears, because the observer might feel empathy for one of the parties or both (victim and perpetrator), and empathy for a rape victim and perpetrator are not necessarily interdependent (Smith & Frieze, 2003).
Rape victim empathy is of particular interest, given that rape victim responsibility was shown to be correlated negatively with rape victim empathy for both males and females (Deitz et al., 1982; Deitz, Littman, & Bentley, 1984; Smith & Frieze, 2003). This psychological construct is based on factors such as perceived similarity or personal experience regarding a rape situation (Batson & Shaw, 1991; Cialdini, Brown, Lewis, Luce, & Neuberg, 1997). Considering the key role of this emotional and cognitive response in predicting attitudes towards rape victims and as no Portuguese adaptation of the Rape-Victim Empathy Scale (REMV, Smith & Frieze, 2003) has been published, our purpose is to examine the psychometric properties of the Portuguese adaptation of this scale.
Rape-Victim Empathy Scale [TOP]
Rape Empathy Scale (RES), developed by Deitz and colleagues (1982), is one of the most widely used measures to assess rape empathy. Nonetheless, this scale has been criticized due to some weaknesses: it measures rape myths rather than cognitive and emotional elements of empathy; participants are asked to empathize with either the victim or the perpetrator using a one-dimensional scale (based on the assumption that rape victim and perpetrator empathy are interdependent); it only measures empathy when a man rapes a woman (a female victim and a male perpetrator) (Olsen-Fulero & Fulero, 1997; Smith & Frieze, 2003).
In the light of these limitations, Smith and Frieze (2003) adopted a revised approach and developed two gender-neutral scales to measure empathy for the rape victim and the perpetrator: Rape-Victim Empathy Scale (REMV) and Rape-Perpetrator Empathy Scale (REMP). These scales have a parallel structure, and reflect emotional and cognitive aspects regarding the rape victim and the perpetrator, measuring empathy during (the during subscale) and after rape (the post subscale). The authors confirmed the validity and reliability of these scales, and further assessed the victim and perpetrator empathy separately (based on the assumption that REMV and REMP are two independent constructs).
Rape Victim Empathy and Other Variables [TOP]
In support of their theoretical assertion, Smith and Frieze (2003) have confirmed that REMV and REMP are independent constructs (feeling empathy for one party does not imply feeling lower empathy for the other), given that no correlations were found for the final versions of the REMV and REMP scales. Using four versions of the REMV and REMP scales (combining victim and rapist gender), Osman (2011) has also found that individuals may feel empathy for both the victim and the offender, given that no correlations were found between the scales in any condition.
Gender differences were further found regarding empathy. Using the RES, Deitz and colleagues (1982) and Deitz, Littman, and Bentley (1984) have shown that women are more empathetic towards the victim than men. In fact, one of the most relevant findings in this area of human behaviour is the existence of gender differences regarding the ability to express empathy (Ching & Burke, 1999; Deitz et al., 1982; Deitz, Littman, & Bentley, 1984; Weir & Wrightsman, 1990), with women suggested to be more empathetic towards the victim and men towards the perpetrator (Mellor, Fung, & binti Mamat, 2012; Sakallı-Uğurlu, Yalçın, & Glick, 2007).
The same sort of evidences was found using the REMV scale (Smith & Frieze, 2003): in the original study, women expressed higher rates of empathy for the rape victim and men scored higher on empathy for the perpetrator. The scale items are gender-neutral, not allowing us to infer about role of the victim or perpetrator gender on empathy (but they may be adapted). However, it is likely that participants imagined a female victim and a male offender, based on cultural and biological roles that expose female victims to male offenders (Crawford & Popp, 2003; Hannon, Kuntz, Van Laar, Williams, & Hall, 1996; Osman, 2011). Thus, according to social stereotypes and bearing in mind that rape is a crime with higher rates of female victims, women are usually assumed as the victim and men as the offender (Check & Malamuth, 1983). Simultaneously, women are more likely to fear rape than men and they are not assumed to have the physical strength to harm others (Rozee, 2008).
Regarding the victim and rapist gender, Osman (2011) has shown that subjects are more empathetic towards female victims, as well as towards those raped by a man rather than a woman. Hence, female victims and male rapists seem to elicit responses that differ from those reported for male victims and female rapists. This is even more important if we consider that when men are raped by a woman this crime is taken less seriously than a male sexual aggression against women (Anderson & Lyons, 2005; Davies, Pollard, & Archer, 2006; Oswald & Russell, 2006), and there is less certainty about the crime (Hannon et al., 2000).
Individuals with personal victimization were further shown to have greater empathy for a victim than those without rape experience (Ching & Burke, 1999; Deitz et al., 1982; Osman, 2011; Smith & Frieze, 2003). Women with victimization experience were revealed to have higher scores in empathy for a female victim, but victimized men reported no differences in empathy levels regarding a male or female victim (Osman, 2011).
Empathy also appears to be related with victim responsibility assignments, which is one of the most studied variables in relation to the first construct. For example, Smith and Frieze (2003) corroborated the findings by Deitz and colleagues (1982), Deitz, Littman, and Bentley (1984), showing that perceived victim responsibility was correlated negatively with rape-victim empathy for both males and females. Thus, these results provide empirical evidence for the effects of empathy on observers’ attitudes towards rape victims: empathy for the victim decreases the level of blame attached to the victim, whereas victim blame increases with perpetrator empathy, proving that blame is inversely proportional to empathy.
The interrelations between empathy and other variables were also canvassed by other studies reporting that empathy is positively related to perceived similarity with rape victims (e.g., Deitz et al., 1982; Deitz, Littman, & Bentley, 1984; Feldman, Ullman, & Dunkel-Schetter, 1998; Miller, Amacker, & King, 2011) and inversely related to rape myth acceptance (Gerger et al., 2007; Miller, Amacker, & King, 2011). For example, when testing a causal model regarding interrelations among predictors of rape victim blaming, Miller, Amacker, and King (2011) have found that rape victim empathy has positive associations with sexual assault history and perceived similarity, and negative relations with rape myth acceptance.
Overview and Hypotheses [TOP]
In line with the above-mentioned, rape victim empathy seems to influence one’s dispositional judgments and might explain likely gender differences regarding rape perceptions. Nonetheless, there is no currently validated instrument to measure this psychological construct in the Portuguese language. For that reason, the purpose of this study is to contribute to assess the validity of the REMV scale (Smith & Frieze, 2003) in a Portuguese sample. We expect to achieve this goal through reliability and exploratory analyses, along with the examination of theoretical relations between both subscales (during and after rape), as well as between the REMV and REMP scales (convergent validity). The following hypotheses were developed: (1) the REMV subscales (during and after rape) are positively correlated; (2) REMV and REMP are not correlated; (3) women display significant higher mean scores for REMV and men for REMP.
Method [TOP]
Subjects [TOP]
A total of 254 Portuguese individuals (127 women and 127 men) participated in the study. The convenience sample used consisted of subjects ranging in age from 18 to 57 years (M = 26.93, SD = 8.20). Most participants were single (78%) and completed 12 years of schooling (57%).
Measures [TOP]
Rape-Victim Empathy Scale (REMV) — This scale (Smith & Frieze, 2003) assesses how participants feel about a rape victim and consists of 18 items (α = .92) that measure empathy during (α = .87) and after rape (α = .90). Originally, participants indicated their level of agreement with each item on a 7-point Likert-type scale (1 = strongly disagree to 7 = strongly agree), where higher scores indicated higher empathy (items 14, 15 and 16 were reverse-scored). To further improve its psychometric properties, the Portuguese version of the scale (Appendix 1) was transformed into a 7-point scale. The REMV scale (and the REMP scale) was translated to Portuguese by bilingual specialists using the back-translation technique. The original scales and translations were further re-evaluated and edited by field experts, and modifications were made to ensure the accuracy of the final version. A pre-test was conducted in a sample of 24 individuals, which led to minor adjustments before the final version.
Rape-Perpetrator Empathy Scale (REMP) — This scale was developed by Smith and Frieze (2003) with the purpose of measuring how likely participants were to empathize with a rape offender. It contains 18 (α = .81) items originally assessed on a 5-point scale. The measure was adapted to a 7-point scale (1 = strongly disagree to 7 = strongly agree) that assesses empathy during (α = .79) and after rape (α = .85). Some items were reverse-scored so that higher scores indicated higher empathy for all items.
Socio-Demographic Information — In this section, participants were asked to report their age, nationality, gender, marital status, and education level. Three control questions were included through which respondents were asked to identify the aim of the study and asked if they had been rape victims or knew any rape victim.
Procedure [TOP]
Data collection was achieved through a self-report questionnaire. Participants were approached in several public settings and informed that the study was part of a project investigating human behaviour. The questionnaire was applied individually, following informed consent procedures and the assurance of confidentiality of the disclosed data. After the inquiry, participants were fully debriefed about the aims of the study.
Results [TOP]
Reliability and Item-Level Analyses [TOP]
The data were analysed using IBM SPSS 21.0. The information regarding mean scores, standard deviation, corrected item-total correlation and Cronbach’s alpha if item deleted are shown in Table 1.
Table 1
Item Descriptive Statistics and Reliability
| Item | M | SD | Corrected item-total correlation | Cronbach’s alpha (α) if item deleted | Skewness
|
|
|---|---|---|---|---|---|---|
| Statistics | Std. Error | |||||
| 1a | 2.65 | 1.69 | 0.26 | 0.84 | 0.75 | 0.15 |
| 2 | 2.54 | 1.65 | 0.40 | 0.83 | 0.83 | 0.15 |
| 3 | 3.93 | 1.53 | 0.36 | 0.83 | -0.25 | 0.15 |
| 4 | 4.64 | 1.67 | 0.63 | 0.82 | -0.54 | 0.15 |
| 5 | 4.22 | 1.80 | 0.61 | 0.82 | -0.26 | 0.15 |
| 6 | 4.34 | 1.68 | 0.63 | 0.82 | -0.30 | 0.15 |
| 7 | 3.93 | 1.90 | 0.52 | 0.82 | -0.11 | 0.15 |
| 8 | 4.43 | 1.49 | 0.64 | 0.82 | -0.48 | 0.15 |
| 9a | 4.39 | 1.83 | 0.22 | 0.84 | -0.41 | 0.15 |
| 10 | 5.06 | 1.49 | 0.65 | 0.82 | -0.73 | 0.15 |
| 11 | 5.18 | 1.47 | 0.50 | 0.82 | -0.95 | 0.15 |
| 12 | 5.56 | 1.32 | 0.47 | 0.83 | -0.95 | 0.15 |
| 13 | 5.52 | 1.40 | 0.59 | 0.82 | -1.31 | 0.15 |
| 14a | 3.89 | 1.81 | 0.06 | 0.85 | -0.06 | 0.15 |
| 15a | 5.98 | 1.37 | 0.16 | 0.84 | -1.59 | 0.15 |
| 16a | 5.25 | 1.58 | 0.21 | 0.84 | -0.70 | 0.15 |
| 17 | 5.29 | 1.42 | 0.33 | 0.83 | -1.01 | 0.15 |
| 18 | 4.59 | 1.60 | 0.57 | 0.82 | -0.67 | 0.15 |
aThe items with superscripts present corrected item-total correlations below 0.30.
Regarding item-level analyses, all the item standard deviations were higher than 1.0, ranging from 1.32 for item 12 to 1.90 for item 7, meaning that responses differed mostly on this item. Most item-total correlations were between 0.33 and 0.65, but items 1, 9, 14, 15, and 16 had low item-total correlations (below 0.30, according to Nunnally and Bernstein 1994). These items were removed, resulting in a 13-item scale with a Cronbach’s alpha of 0.87. Reliability for the subscales also presented acceptable internal consistency both for REMV during (α = 0.79) and after rape (α = 0.85). These values are above the 0.70 level recommended by Nunnally (1978). On the basis of these results, 13 of the 18 items were included in the subsequent factor analysis.
Factor Structure [TOP]
The factor structure of the REMV scale was analysed through an exploratory factor analysis for the 13 items. The KMO value of the data was 0.85 (above the 0.60 value recommended by Martínez-Arias, 1995) and the Bartlett’s test was significant (χ2 = 1413.503; DF = 78; p < 0.001). The exploratory factor analysis, using the principal components estimation method with varimax rotation and the criterion of eigenvalue higher than 1.00, produced a two-factor solution accounting for 55% of the total variance (40.70% for Factor 1 and 14.29% for Factor 2). Both factors were identified, respectively, as REMV post rape and during rape, with eigenvalues of 5.29 and 1.86. The standardized factor loadings for the 13 items composing these two factors are presented in Table 2.
As the purpose of the exploratory factor analysis was the identification of the preliminary factor structure underlying the scale, we chose to retain items with a factor loading of 0.40 or above (Stevens, 2002). The first factor was composed of eight items, and the second of nine and all loadings were higher than 0.41, indicating that though related they were reasonably independent. Some items (7, 10 and 18) cross-loaded in both factors, with the major loadings belonging to the original dimension. Item 8 cross-loads in both factors, with a higher loading in a factor to which it does not originally belong; however, as the loading is above 0.40 and the semantics of the item reflects empathy after rape, we chose to maintain it in the original factor. Therefore, all items were accepted and remained in the original factor. Factor 1 (empathy after rape) is composed of items 2 to 7, whereas Factor 2 (empathy during rape) is composed of items 8, 10, 11, 12, 13, 17, and 18.
Variable Relations [TOP]
The REMV subscales (during and post rape) are shown to be correlated positively (r = .55, p < .001). These results lend support for hypothesis 1, indicating that as the value of one variable goes up in value, the other variable goes up too. Hypothesis 2 is also supported by our findings, since Pearson coefficient correlations confirm that no association exists between REMV and REMP (r = -.45, p = .46). REMP has also presented acceptable internal consistency for both the overall scale (α = .85) and subscales: during rape (α = .73) and post rape (α = .78).
Finally, regarding gender differences, mean scores for the REMV scale did not differ significantly (p > .05) according to the observer gender. There is a tendency, however, for women to score higher (M = 4.65, SD = .98) than man (M = 4.46, SD = .99). On the other hand, as expected, mean scores for REMP (M = 12.52, SD = 4.47) were found to be significantly different (t|252| = -4.468, p < 0.001) according to observer gender: men (M = 2.75, SD = .93) were more empathetic towards the perpetrator than women (M = 2.29, SD = .72). These results partially support Hypothesis 3, because significant gender differences were observed for REMP but not for REMV.
Discussion [TOP]
This study examined the psychometric properties of the REMV scale (Smith & Frieze, 2003) in a Portuguese sample, providing a number of useful outcomes in the context of the study sample. Five original items (1, 9, 14-16) were removed before the exploratory analysis, because they presented item-total correlations below 0.30 (Nunnally & Bernstein, 1994). Two items originally fitted in the during rape subscale (1 and 14), whereas the others belonged to the post rape subscale (9, 15 and 16). One possible explanation for these low values lies on the fact that three of these items were reverse-scored: “I find it difficult to know what goes on in the mind of a rape victim” (14), “I don’t understand how a person who is raped would be upset” (15), and “I can’t understand how someone who has been raped can blame their partner and not take some of the responsibility” (16). The reverse-scored items were more often answered atypically and decreased overall validity. It is recommended that appropriate correction procedures be applied (e.g., using non-reversed items, although response biases may represent a substantial problem) to reduce the difficulties.
Semantically, it is also unclear whether item 1 (“I find it easy to take the perspective of a rape victim”) fits in the during or post rape subscale, and individuals may find it difficult to put in the victim’s shoes but still be empathetic (because it is never easy to assume the other’s perspective in such a painful situation). The same assumption is applied to item 9 (“I know if I talked to someone who was raped I’d become upset”), given that the subjects had no contact with sexual victimization and may not be able to predict their own reactions.
It is recommended that appropriate correction procedures be applied in the future for all the items (e.g., using non-reversed items, although response biases may represent a substantial problem). As a result, only items with item-total correlations above 0.30 (Nunnally & Bernstein, 1994) were retained and included in the subsequent factor analysis. A final 13-item instrument was shown adequate levels of internal reliability.
Using a 13-item instrument, our findings further confirmed a two-factor solution nearly similar to the one found by Smith and Frieze (2003): empathy during (Factor 2) and after rape (Factor 1). The standardized factor loadings were all above 0.40 for both dimensions (Stevens, 2002). Item 8 cross-loads in both dimensions and presents the lowest loading in the factor to which it originally belongs (probably because it is a compound-complex sentence). This was not considered a substantial problem, given that it has a value above 0.40 and it cannot be changed to the during rape subscale since it semantically refers to a situation that occurs after rape. We further observed that other items (6, 10 and 18) cross-loaded in both factors, but the load values were higher in the original factor, not presenting a problem.
In general, the current exploratory study reveals that the 13-item Portuguese version of the REMV scale (Smith & Frieze, 2003) retains acceptable psychometric measures, supporting its dualistic nature. However, the lack of other studies on the psychometric properties of the scale prevents us from performing a comparative analysis regarding the reliability and validity of the instrument. In the future, confirmatory factor analysis should be conducted with the complete scale to determine whether these findings hold (adapting the items with major problems).
Additionally, the REMV scale presented theoretically consistent relationships with other variables. Validity evidence was found via correlational tests between both latent dimensions, with a moderate positive correlation being found between the during and post rape dimensions. This result was to be expected as both factors share mutual elements (they both refer to a unique relationship with a rape victim). This means that both dimensions represent rape victim empathy, but this correlation is not redundant. The correlation observed in this study is similar to the one found in the original scale (Smith & Frieze, 2003), providing further evidence on the scale’s validity.
The REMV scale was also found not to be correlated with REMP (Smith & Frieze, 2003), indicating that these theoretical constructs are interdependent and they should be measured separately. Our results are consistent with past findings (Osman, 2011; Smith & Frieze, 2003) and lend support to the assumption that REMV and REMP are not mutually exclusive, because feeling empathy for the victim does not mean that one cannot feel empathy for the rapist (and vice versa).
Regarding gender differences, women were found to score higher in REMV than men, but these results are not statistically significant. Taking into account our preference for ingroup members (Tajfel, Billig, Bundy, & Flament, 1971; Tajfel & Turner, 1986), which may be based on factors such as perceived similarity (Batson & Shaw, 1991; Cialdini et al., 1997) with a rape victim, women were expected to present higher levels of empathy for the victim. Likewise, according to defensive attribution theory (Shaver, 1970), when an adverse event occurs, individuals tend to assign less blame to those perceived as more similar. The perceived similarity with the victim may point to the likelihood of a similar misfortune, stirring individuals to defend themselves against the prospect of future victimization. Thus, considering that, typically, women are the victims of rape, they were expected to be more empathetic towards the victim (Osman, 2011; Smith & Frieze, 2003).
Considering that culture may influence empathy towards the victim (because the perception of gender roles and of rape may differ according to different cultures), our results might have been impacted by the specificity of the Portuguese sociocultural context. According to an optimistic perspective, the results obtained might reflect the greater awareness of the Portuguese individuals about the causes and consequences of this sort of crime. The mass media have a prevailing influence on people’s perceptions and, by inducing individuals to take the perspective of a rape victim, they may represent a way to reduce stigma, regardless of gender (Sakallı-Uğurlu, Yalçın, & Glick, 2007). The lack of significant differences between genders might also be linked with the general feeling of greater closeness towards the victim or with the perception that she is not morally responsible for the situation. The transformation of the relationships between men and women was one of the most important acquisitions in Western cultures. Stirred by the fall of the dictatorship regime, Portugal has known great social changes during the last decades of the twentieth century. As a consequence, today women are recognized the same rights as men (Silva, 2002). Assuming that individuals imagined a female victim, our results may reflect these social changes, meaning that the Portuguese culture is less conservative than the cultures of other studies on the topic (mostly American).
On the other hand, the mean scores for REMV were slightly above the scale’s central point for both genders. Hence, according to a less optimistic perspective, the results obtained in this study may be the consequence of certain impartiality regarding this crime. On the contrary of other countries where the topic has been quite discussed, such as the United States (e.g., Miller, Amacker, & King, 2011; Osman, 2011; Smith & Frieze, 2003) and the United Kingdom (e.g., Grubb & Tarn, 2012), in Portugal rape is still a topic that is not discussed openly (with a small number of empirical studies on the theme, and none regarding rape empathy and gender differences); thus, no further inferences can be drawn, meaning that much research is still needed regarding the influence of culture on rape perception and gender differences.
Concerning the REMP scale, as expected, male participants were shown to be more empathetic towards the perpetrator than female participants, supporting similar findings (Osman 2011; Smith & Frieze, 2003). The scale items are gender-neutral, but it is likely that participants pictured a female victim and a male rapist (because this is the social prototype of a rape situation) (Check & Malamuth, 1983; Osman, 2011). Additionally, until recently, Portugal was emerged in a patriarchal culture where women were expected to be submissive and men were educated to be tough and aggressive (Silva, 2002). As we know, social changes take time, meaning that the crime might have been perceived as the result of a sexual attraction (Deitz, Littman, & Bentley, 1984). According to the preference for ingroup members (Tajfel et al., 1971), namely based on perceived similarity (Batson & Shaw, 1991; Cialdini et al., 1997), men might have taken the perspective of a rapist who was not able to control himself, justifying, in a certain way, his lack of control over sexual impulses.
In future studies, participants should be asked to indicate which gender they imagined when rating the items (for both the victim and the rapist). Similar research might also prove useful to ascertain the psychological processes that contribute to victim empathy and its interrelations with situational and individual variables such as ambivalent sexism (Glick & Fiske, 1996), just world belief (Lerner & Miller, 1978), perceived similarity with the rape victim (Grubb & Harrower, 2009; Miller, Amacker, & King, 2011), social projection (Krueger, 2007), relational-interdependent self-construal (Cross, Bacon, & Morris, 2000), rape myth acceptance (Burt, 1980), and other categories of social cognition. The testing of predictive models regarding interrelations among the variable above-mentioned should also be considered along with assignments of blame to rape victims and perpetrators.
Thorough research is needed in order to develop more adequate intervention programs focused on reducing the likelihood of social stigmatization towards rape victims. Using a sample of fraternity men at a public university, Foubert and Newberry (2006) have shown that attitudes towards rape can be changed through prevention programs: after participating in such a program, participants were found to report higher empathy towards rape victims and lower rape myth acceptance, likelihood of raping, and likelihood of committing sexual assault. This is even more important when considering that empathy promotes helping behaviours. Paciello and colleagues (2013) have confirmed this predictive power of empathy in helping attitudes and behaviours, even when the helpers are faced with potential high costs.
Therefore, the use of research action plans and the beliefs of professionals (e.g., law enforcement officers, educators and health professional) working in the field should also be considered. Professionals would benefit from thorough information on the topic, because using the knowledge of perceptual biases might mitigate assignments of blame to rape victims (in general and specific contexts such as rape education programs or rape trial decisions). As a final recommendation, the gender of the victim, rapist and observer, as well as the victimization experience, should also be included in future research on the topic (with studies similar to the one by Osman, 2011).
In conclusion, this study provides further support to the importance of considering REMV and REMP as two separate constructs. The REMV scale (Smith & Frieze, 2003) was shown to be a reliable and valid instrument to measure how likely individuals are to empathize with a rape victim. Therefore, it appears that the conceptualization REMV as a twofold concept embodies a promising road towards understanding the processes underlying rape perceptions.
