[Violence scale and severity index: a methodological proposal for measuring violence by the partner in Mexican women]

To construct and validate a scale to assess violence by the male partner against women. An index of severity of the emotional and physical damage was also designed to assess the intensity of the violent actions against women.

The sample consisted of a total of 26 042 women who participated as respondents in the National Survey on Violence against Women (ENVIM per its abbreviation in Spanish) conducted in Mexico during 2003. Respondents were all users of health services provided by the Mexican government. The questionnaire was organized into 17 sections, one of which was a 27-item scale to assess partner violence. The purpose of this scale was to measure the type (physical, emotional, sexual and financial) and degree of violence based on severity. A severity index was constructed based on two procedures: 1) the validity, reliability, and factor analyses of the scale and 2) the assessment of severity by expert judges who assigned a value to each item of the scale.

The validity and reliability results indicated this scale has adequate internal validity (Cronbach's Alpha = 0.99). The factor analysis with Varimax rotation yielded a four-factor solution. The factors were: 1) Psychological violence; 2) Physical violence; 3) Severe physical violence; and 4) Sexual violence. The combination of the four factors accounted for 62.2% of the variance in the scale. Using the ratings from the judges a table of values for each of the violent actions described was obtained. The scores assigned by the judges ranged from 0 to 354. Results revealed a prevalence of 21% partner violence in the last twelve months. The prevalence of psychological violence was 18.5%; of physical violence 10.1%; severe physical violence 6.7% and sexual violence 7.0%.

The scale of violence described in this article is a very useful and reliable instrument to assess marital violence against women. It is suggested that this instrument be used in other settings to compare results with different samples.