Note: All figures are cumulative (i.e., they reflect the respondents' total experiences from their first NYS interview as an adolescent through their last interview either as an adolescent or an adult). The figures are zero-order correlations (Pearson's r values), which means that each figure reflects a linear relationship between just two variables (e.g., prevalence of violent victimization and prevalence of felony assault), without accounting for any of the other variables. The figures in parentheses are neither statistically significant nor marginally significant (with p < .05 representing statistical significance and p < .10 representing marginal significance); all other figures (those not in parentheses) are at least marginally significant.

Prevalence refers to whether a respondent had a particular problem. Frequency refers to the number of times the respondent had the problem. Frequency data have been adjusted mathematically (by logarithmic transformation) to reduce their skewness (i.e., lack of symmetry in the distribution); the adjustment also gives more weight to lower frequencies, because lower frequency estimates of offending have been shown to be more reliable than higher frequency estimates in the NYS data (Huizinga and Elliott, 1986).