Acceptability of dating violence among late adolescents: the role of sports participation, competitive attitudes, and selected dynamics of relationship violence

Adolescence, March 22, 2008 by Michael J. Merten

Model 1 in Table 4 examines the four respondent-level characteristics of athletic participation, competitiveness, need to win, and respondent gender. No relationship was found between acceptability of violence and two of the three athletic variables, athletic participation and competition. However, the third athletic variable, need to win, has a significant relationship acceptance of violence in dating relationships ([beta] = .24, SE = .06, p < .001). Respondent gender also had a significant relationship with the acceptability of violence, with males being more accepting of violence than women. Using Raudenbush and Bryk's (2002) method to compute explained variance for respondent-level variables, these four respondent characteristics account for 8% of the variance in acceptability of violence scores.

Model 2 adds the first case-level variable, initiator act, to the equation. This variable is a significant predictor of the acceptability of violence in dating relationships ([beta] = .35, SE = .02, p < .001), with violent responses by the recipient to more aggressive acts initiated by the dating partner being viewed as more acceptable than violent responses to less aggressive initiator acts. The association between the case-level initiator act and acceptability of violence is independent of the association between the respondent-level factors and acceptability of violence previously tested, indicated by the stability of the respondent-level betas and [R.sup.2] when the case-level variable is added to the model. This single case-level initiator act accounts for an additional 5% of the variance in acceptability of violence. The percentage of explained variance accounted for by case-level characteristics was computed using a method derived by Snijders and Bosker (1999).

In Model 3 the second case-level variable is added to the model, recipient's reaction to initiator's act of violence. Recipient's reaction has a significant negative association with the acceptability of violence in dating relationships ([beta] = -.55, SE = .02, p < .001). The direction of the relationship indicates that the recipient's response becomes less acceptable the more severe the recipient's reaction. Recipient reaction accounts for an additional 15% of the total variance of acceptability of violence scores.

Model 4 adds the third case-level variable to the model, initiator-recipient gender combination. Results indicate that, controlling for level of response, a female recipient's violence toward a male initiator is more acceptable than a male recipient's violence toward a female initiator ([beta] = .73, SE = .03, p < .001). Initiator-recipient gender combination accounts for a further 10% of the variance in acceptability of violence.

These three case-level variables, reflecting important aspects of violent couple interactions, are independent of the respondent-level variables. This is indicated by the stability of the respondent-level betas and [R.sup.2] values as the case-level variables are successively added to the general model. Together they explain 30% of the variance in the acceptability of violence, which is three times that accounted for by the respondent-level athleticism variables.