NCJ Number
174146
Journal
Journal of Interpersonal Violence Volume: 13 Issue: 1 Dated: February 1998 Pages: 58-72
Date Published
1998
Length
15 pages
Annotation
Three studies of the validity of self-reported perpetration of violence and victimization by violence among adolescents found that one cannot trust absolute frequencies of violent incidents obtained from self-reports.
Abstract
In the first study, participants were 426 students (50.1 percent female) in grade 11 classes at nine schools in Simcoe County, Ontario. The questionnaire used in this study contained 26 items about male-to-female aggression among teenage peers; boys reported perpetration, and girls reported victimization. In the second study, participants were 79 students (51.9 percent female) in grade 11 classes at all four high schools in a city of 25,000. Questionnaires and administration were the same as in the first study, except half the questionnaires used a 6-month time frame for all items, and half used a 12-month time frame. In the third study, participants were 182 students (50.6 percent female) in non-grade 11 classes at one school that requested activities for its remaining students; whereas, grade 11 students participated in the larger research project. All questionnaires asked about the perpetration and victimization of nonphysical and physical aggression that involved boys and girls and sexual aggression that involved opposite-sex peers. Half the questionnaires asked about the past 1 month, and the other half asked about the past 12 months. If the self-reports in these studies were valid, variations in time frame should have produced large differences in reports, but they did not, and sex/victim perpetrator should have had little effect on incidence estimates, but had a large effect. Although these findings place severe limits on the validity of these reports, they do not necessarily mean the participants' reports were meaningless. More serious items tended to yield greater time frame effect. Moreover, the effects of sex were sensible and systematic. Males' and females' reports of prevalence correlated, suggesting that reports were affected by actual event frequencies. Correlations between victims' and perpetrators' reports and analyses of internal consistencies gave no basis to conclude that short time frame estimates were more reliable (and therefore more valid) than long time frame estimates. These results cast doubt on fundamental assumptions of recent research on interpersonal violence. 2 tables, 1 figure, and 29 references