Psychometrics of the Violence Nonlinear Dynamics Scale

The trajectory of daily partner violence generally demonstrates nonlinear dynamics, and this nonlinearity is important to patient outcomes, as it is sometimes a stronger predictor of outcomes than violence frequency or severity. However, measurement of such dynamics is difficult, requiring complete time series data of sufficient length to yield stable measures. The purpose of this study was to develop a pencil-and-paper instrument to estimate violence nonlinearity and assess its psychometrics.

Adult women (N = 143) who experienced violence in the previous month were enrolled from 6 primary care clinics. Baseline surveys assessed factors known to correlate with nonlinearity (partner's control strategies, violence appraisal, hope, social support, coping style) and violence dynamics using a 30-item instrument based on traditional characteristics of complex adaptive systems. Participants completed daily assessments of the previous day's violence using interactive voice response via telephone for 8 weeks, with data collection occurring between August 2013 and March 2015. Three different measures of nonlinearity were computed: LZ complexity (algorithmic complexity), approximate entropy (lack of regularity), and Lyapunov exponent (sensitivity to initial conditions).

Using factor analysis and reliability measures, the final 10-item Violence Nonlinearity Dynamics Scale (VNDS) was identified. The VNDS was found to have both internal consistency (0.817) and split-half reliability (0.796). In addition, the instrument demonstrated concurrent (correlating with both the combined nonlinearity factor score [r = 0.267] and Grassberger-Procaccia entropy [r = 0.338]) and construct (correlating with 9 of 13 previously identified nonlinearity correlates) validity.

The VNDS has both reliability and validity and could facilitate the inclusion of nonlinearity assessment in both intimate partner violence research and clinical work.