Saturday, January 17, 2009
Preservation Hall (New Orleans Marriott)
* noted as presenting author
PURPOSE: The intimate partner violence (IPV) literature identifies control as an essential construct in understanding IPV against women, yet lacks a specific measure of control in IPV. The lack of measurement underlies a critical gap in the literature: the relationship between contexts of control and causes and consequences of IPV. To address this gap, this paper presents the development and initial validation of a scale to measure factors that contribute to control of women intimate partners in a variety of contexts. METHODS: Our review of the relevant literature identified a basis for five controllability dimensions: social, physical, financial, cultural, and institutional. The data collection instrument contained demographic questions and the five subscale item pools. Scale items were developed based on literature and refined through expert interviews. After obtaining approval from the institutional review board, a purposive sample of 241 male and female adult research participants was obtained from a public four-year university (N=101), a public community college (N=109), and community events (N=31) in a city in the Southeastern U.S. Data were assessed for missing values and appropriately managed. Analyses included assessment of validity with one-way ANOVA, reliability using Cronbach's alpha and the standard error of measurement, and factor structure through exploratory and confirmatory factor analyses. FINDINGS: The global scale and three subscales Power, Dependence, and Resources possess good psychometric properties and apply to individuals in varied contexts. Reliability and validity were initially assessed for the entire scale with five subscales. However, based on an inconclusive factor structure, an exploratory factor analysis (EFA) was conducted. We found a marginally reliable three-factor structure for the controllability scale: Power (Cronbach's alpha=.66; SEM=.42), Dependence (alpha=.73; SEM=.47, and Resources (alpha=.69; SEM=.32). The global scale reliability was good (alpha=.83; SEM=.22), indicating a strong global measure of this innovative construct. Discriminant and convergent construct validity were established using participant race and a single-item indicator of control, respectively. A one-way ANOVA on race with Tukey's post hoc comparisons was not significant, supporting discriminant construct validity. Higher participant ratings of how much someone controlled their lives were hypothesized to be convergent with overall and subscale controllability. The ANOVA was significant for overall controllability (F=11.5, p<.001) and the power (F=11.6, p<.001), dependence (F=3.4, p<.05), and resources (F=5.0, p<.01) subscales. Six expert reviewers concluded the three-factor items had satisfactory content validity. The three-factor model had a good overall fit with chi-square-degree of freedom ratio 1.53 (p=.000) and fit indices from .87 to .93. There was no significant departure from model fit (RMSEA=.045, p=.84). IMPLICATIONS: The broad applicability of the controllability scales and their ability to assess factors that facilitate control in diverse contexts is vital to understanding IPV control. Based on our results, the global controllability and subscales show promise for use in research applications seeking to understand how abusive intimates control their partners. In addition, the 29-item global scale shows promise for use in applied settings. The Controllability scale contributes to the literature on IPV by providing the first measure of multi-dimensional situational factors that facilitate control in intimate relationships.