Materials and methods
The data are collected from the study Young in Norway, which was launched in 1992 on the basis of a nationwide representative sample of adolescents in schools (response rate: 97 per cent). The next round of data collection took place in 1994 (response rate: 92 per cent). Respondents who still were in school in 1994 and who had assented to continued follow-up (91 per cent) received a questionnaire in 1999. Of these, 84 per cent responded (cumulative response rate: 68 per cent). Detailed information on the design of the study, data collection and permissions can be found elsewhere (7).
Most of the analyses are based on the 1999 study and are restricted to women who had been in a relationship during the previous six months (n = 1 250). In addition, the 1994 data are used for longitudinal analyses. In the following, I will refer to the measurement points as t2 and t1 respectively. At t2, the respondents were 20 – 25 years old.
Intimate partner violence (IPV) was assessed at t2 using two paired questions about the respondent’s and the partner’s physical assaults (slaps in the face, kicking, biting and/or blows with a fist). The questions were derived from a modified short version of the Conflict Tactics Scale (8). In most analyses, two dichotomous variables have been used. The first concerns whether the women had perpetrated the described forms of violence at least once. The second distinguishes between non-reciprocal and reciprocal IPV.
Independent variables. Other items based on the Conflict Tactics Scale concerned verbal aggressiveness (verbal abuse, insults) and statements or actions undertaken with a malicious intent. Avoidance in conflict situations (sulking, refusal to speak) was also assessed. The variables were dichotomised with the cut-off point placed at six or more incidents during the last six months, which implied that the top 10 – 15 per cent of the distribution was singled out.
Whether the partner was perceived as supportive was operationalised as having ticked «partner» for the following items: Who can make you feel better when you’re really down and out? (item 1), whom would you’ve asked for advice or help if you had a personal problem? (item 2), or if you had been arrested by the police for illegal activi-ties? (item 3). A total of 13 response categories were provided, and multiple answers were possible. Perceived acceptance by the partner was identified in a similar manner, although with the aid of only one item: Who accepts you fully and totally, your good as well as bad sides? Since these measurements referred to ongoing relationships, they were only used in analyses of women who actually were in an intimate relationship at the time. The same applied to analyses of the women’s perceptions of the quality of their sex life and the relationship as a whole. Here, responses were given on a five-point scale, and a positive assessment was defined as having ticked either «very good» or couldn’t have been better».
Aggression-related personality traits were assessed at t2 with the aid of six items from the State Trait Anger Expression Inventory (9). The respondents reported how well various statements described themselves by crossing off on a scale ranging from 1 («applies very poorly») to 4 («applies very well»). Three items measured angry temperament (e.g., «I’m short-tempered») (Chronbach’s α = 0.87), and another three concerned anger suppression (e.g., «I often harbour grudges against others») (Chronbach’s α = 0.61). Two sum scores were constructed and subsequently dichotomised with the cut-off point placed at an average score of 3.0 or above. Self-assessed aggressiveness was assessed at both t1 and t2 using an item from Bem’s Sex Role Inventory (10). The variable was dichotomised, so that respondents who answered that it was «completely» or «fairly» correct that they had this characteristic were singled out.
Eight dichotomous variables from t1 concerned involvement in problem behaviour during the last year. For low-prevalence behaviour (fistfights, vandalism, bullying, shoplifting/petty theft) the cut-off point was placed at no versus 1+ incidents. For more prevalent norm violations, the cut-off point was placed between 0 – 1 and 2+ (fare dodging in public transport, telling lies), or between ≤ 5 and 6+ incidents (unwarranted absence from school). A measure for verbal aggression, which was constructed on the basis of two items pertaining to acting-out behaviour towards teachers (ferocious arguing, profanities and/or verbal abuse) was also used (≤ 3 vs. 4+ incidents).