Confirmatory Factor Analysis and Gender Invariance of Acceptance of Rape Myths Scale in Nigeria

Background: Research on beliefs associated with violence against women has increased over the last decades (Fulu et al., 2013) and new scales are increasingly developed and modified for examining sexual violence against women (see Flood, 2008). Nevertheless, many scales continue to suffer from widespread use due to sparse information about their structural properties and measurement equivalence across societies. Regrettably, scales developed in one society may not hold the same meanings for respondents in a different society and interpretations of findings from utilizing such scales across countries are often greeted with great skepticisms (Byrne & Watkins, 2003). The present analyses therefore examine confirmatory factor analysis (CFA) and gender invariance of rape myth acceptance scale in Nigeria. Specifically, the analyses examined whether the scale holds similar factor structure, whether the latent means can be compared, whether respondents interpreted items similarly or ascribe the same meaning to them across gender in Nigeria. 

Methods: Data (N=454) from two cross-sectional surveys in Nigeria were merged for the analyses: 39.6 % were females (n=177) and 60.4% were males (n=270). The mean age was 31 years (SD=8). The 22-item Illinois Rape Myth Acceptance (IRMA) scale (McMahon & Farmer, 2011) captured four dimensions of rape-supportive beliefs, namely “She asked for it,” “He didn’t mean to,” “It wasn’t really rape,” and “She lied.” Response choices, using a Likert-type scale, ranged from strongly agree =1 to strongly disagree=5. Examples of items include "if both people are drunk, it can’t be rape” and “rape happens when a guy’s sex drive goes out of control.” Cronbach’s alpha is as follows: “She asked for it” .81, “He didn’t mean to” .78, “It wasn’t really rape” .72, and “She lied” .79. CFA using maximum likelihood estimation method (Stata-14 software) and measurement invariance across gender was examined. For the CFA, RMSEA, CFI, and SRMR were examined to determine acceptable model fit (Kline, 2006). Non-significant change (Δ) in chi-square, RMSEA values falling within the confidence interval, and the ΔCFI being smaller than -0.01 informed decisions about measurement invariance across gender (Cheung & Rensvold, 2002; Timmons, 2010). 

Results: CFA demonstrated reasonable fit for rape myth acceptance scale: RMSEA .069; CFI .88, and SRMR .066. Measurement invariance across gender was equally established. The change in chi-square was nonsignificant for metric factorial invariance (Δχ² =23.75, df = 18, ns) and the RMSEA values fell within the confidence interval of values for configural invariance. Instead of scalar invariance, partial scalar invariance (three unconstrained items) (Δχ² =24.61, df = 19, ns) was established: the ΔCFI between scalar invariance and partial scalar invariance was smaller than -0.01. 

Conclusions and implications: Preliminary findings indicated that the hypothesized model of 22-item acceptance of rape myths is plausible in Nigeria. With minor modifications, latent means was comparable across gender, as men and women appeared to ascribe the same meanings to IRMA’s construct and interpreted the items similarly. Therefore, IRMA is suitable for examining rape-supportive beliefs in Nigeria. Its future use in the region may generate comparable knowledge with other regions where validation has been established.