Note: The R_{L}^{2} (explained variation) for table 6 (0.127) indicates the strength of the relationship between adult success and the set of adolescent predictors, taken as a group. It is the percentage reduction in error of prediction. The R_{L}^{2} of 0.127 indicates that it is possible to reduce the error in predicting adult success by 12.7 percent if all of the adolescent predictors listed in the table are known.
^{*} The likelihood ratio criterion, *p*(lr), indicates the extent to which the predictive effects of the adolescent variables are statistically significant, either for the
model as a whole (global significance) or for a particular comparison (nonsuccess versus stable success, or unstable success versus stable success). One way to interpret *p*(lr) is to think of it as the probability that the observed effect is just a coincidence, as opposed to a real effect of the adolescent predictor on the adult outcome. Effects are considered statistically significant if *p* __<__ .05 or marginally significant if *p* __<__ .10; if p is greater than .10, one cannot be certain whether the effect in the table represents a real impact of the predictor on the outcome, as opposed to random variation or coincidence.

^{†} Odds ratios represent the factor by which to multiply the odds of a particular outcome for each predictor (e.g., being a victim of violence in adolescence multiplies the odds of nonsuccess as opposed to stable success in adulthood by a factor of 1.376).

^{‡} Cumulative frequency.

^{§} Cumulative prevalence.

^{||} Prevalence.

^{¶} The intercept is the expected value of the outcome (actually, the natural logarithm of the odds of the outcome) if the values of all the predictors are zero.