Beliefs about Relationship Violence Against Women Scale and Gender Stereotypes and Beliefs: Development and Confirmatory Factor Analysis

Background: Although many instruments have been developed to measure relationship violence (Flood, 2008), most are focused on physical violence, few on sexual/psychological violence (Strauchler et al., 2004), and many lack common socialized beliefs and stereotypes in many patriarchal societies. Therefore, three studies describe the development and confirmatory factor analysis (CFA) of Beliefs about Relationship Violence against Women Scale (BRVWS) and Gender Stereotypes and Beliefs (GSB). 

Methods: We collected data on three studies using a cross-sectional survey design. Item development (56 items using a Likert-type scale from strongly disagree = 1 to strongly agree= 5) includes comprehensive literature review, focus groups, expert review, and pilot testing. Examples of items include "because a man is the head of his household, he has a right to hit his woman;" "a woman should be sexually submissive to her man."

The sample (N=210) for study 1 in Nigeria was 53.3% male (n=112) and 46.7% female (n=98). The mean age was 28.71 years (SD=9.63). The sample (N=202) for study 2 in Nigeria was 63.9% male (n=129) and 36.1% female (n=73). The mean age was 26.49 years (SD=5.82). The sample (N=404) for study 3 includes the United States (n=127, 31.40%), South Africa (n=122, 30.20%), and Nigeria (n=155, 38.40%): 73.8% were females (n=298) and 26.2% were males (n=106). The mean age was 35.43 years (SD=10.47).

Exploratory factor analysis (EFA) (principal axis factoring method) was used to extract the factor structure. The final factors retained were influenced by empirical/theoretical relevance, criteria of eigenvalue greater than 1, and the scree plot. Varimax rotation was used to rotate factor loadings for interpretations. Items that loaded at greater than .40 on their respective factor were retained in the final structure.

Construct and concurrent validity, social desirability/common method bias were examined using scales ranging from Illinois rape myth acceptance scale (McMahon, 2010) to Buss-Perry Aggression Questionnaire(Buss & Perry, 1992).

CFA using maximum likelihood estimation method (Stata-13 software) and measurement invariance was performed on the identified factors (study 2-3). Recommendations for acceptable model fit follow stipulation by Kline (2006): RMSEA ≤ .05 or between .05 and .08, CFI >.90, and SRMR < .10.

Results: The EFA (study-1) identified six theoretically meaningful factors: three factors for BRVWS (13 items) - physical violence, psychological manipulation and control, and sexual violence against women; and three factors for GSB (16 items) - beliefs about sexual submissiveness of women, emotional stereotypes about women, and sexual stereotypes about men. Cronbach’s alpha ranged from .79 to .90.

BRVWS/GSB correlated moderately (≥.30) with hostility, physical aggression, relationship distress, victimization experience, and rape myth acceptance.

CFAs demonstrated acceptable fit for BRVWS/GSB in Nigeria: RMSEA .072(BRVWS) and .067(GSB); CFI .937(BRVWS) and .937(GSB); SRMR .060(BRVWS) and .054(GSB); and TLI .917(BRVWS) and .926(GSB) (study-2) and modified cross-cultural fit for GSB (2-Factor): RMSEA .080; CFI .956; and TLI .953 (Scalar Invariance, study-3).

Conclusions and implications: Preliminary findings suggest that BRVWS and GSB are suitable for understanding gender-based violence beliefs and stereotypes, the knowledge of which is crucial for cross-cultural practice and policies on violence against women.