Logistic regression model of prediction of exposure to violence against Roma women in Roma families in isolated Roma settlements of Međimurje county


  • Aleksandar Racz Zdravstveno veleučilište
  • Borut Rončević
  • Milan Milošević



Based on the conducted field research on the incidence of violence against Roma women in Roma families, as well as their beliefs and attitudes about violence against Roma women in Roma families conducted on the cause of 350 Roma women living in 12 isolated Roma settlements in Međimurje County, by using logistic regression analyzes, three predictive profiles of Roma women as victims of gender violence in Roma families were made.

The application goal of the research is aimed at helping health and social services if they could further shift the focus of their work towards those Roma women as beneficiaries who are at higher risk of exposure to violence.

In the first model, which included only socio-demographic characteristics of respondents according to Wald's test of significance of coefficients, older age (OR 1,848, p-0,007), material deprivation (OR 2,345, p-0.021) and the earlier age of birth of the first child (OR 7.962, p-0.004) were singled out as statistically significant from the predictors of Roma women's exposure to domestic violence within Roma families.

Extending the model with attitudes and prejudices about violence against women with significant predictors already highlighted, such as younger age (OR 1,760, p-0,015), earlier age of first child (OR 8,210, p-0,003) and earlier age of starting a family (OR 0.305, p-0.048), as the strongest predictor, beliefs and attitudes about violence were singled out (OR 1,957, p-0,002). Roma women with a higher degree of acceptance of prejudice about violence are also more likely to be victims of violence.

Finally, the introduction as an independent variable and exposure to childhood violence by individual forms of abuse (physical, psychological, sexual and economic) resulted in a prediction model that proved to be statistically significant (p <.01), which of all three models best predicted values ​​of the dependent variable, where the predictors were significant earlier age of birth of the first child (OR 7,955, p-0,041), earlier age of starting a family (OR 0,106, p-0,017), higher number of children in care (OR 3,087, p-0,049 ) and greater exposure to physical (OR 14,560, p-0.001), economic (OR 4,728, p-0,049) and psychological (OR 8,088, p-0,000) childhood violence. Exposure to domestic violence is such a strong predictor that by introducing it into the model, attitudes about violence, ie a greater tendency to accept prejudices about violence against women and greater material deprivation, lose their predictive power, which directly confirms the assumption that Roma women victims of childhood violence a much higher risk of being a victim of adult violence.