Wiesner 4
Analogous analyses with onset of arrests as predictors did not reveal significant
predictive effects for either early or late onset of arrests.
Prediction of times being fired. Approximately 70% of the young men did not
report being fired from a job across the nine-year period. The maximum number of times
being fired was 5 (1% of the sample). Prospective effects of adolescent deviant behaviors
and young adult risk factors on the outcome measure were investigated using the Poisson
regression model (LIMDEP version 8.0.10; Econometric Software, 1996-2003) which is
well-suited for the analysis of count variables. Predictive effects were controlled for SES,
age, exposure time (total number of months in the labor market), and parental antisocial
behavior at age 9/10 as a measure of early family risk.
The likelihood ratio test for goodness-of-fit indicated that the full Poisson model
was significantly better than the intercept-only Poisson model (Chi
2
(8)
= 45.67, p < .001).
Furthermore, the likelihood ratio test for overdispersion indicated that the Negative
Binomial model provided a significantly better fit to the data than the Poisson model
(Chi
2
(1)
= 14.84, p = .0012). Zero-inflated extensions of both models were also tested, but
either failed to converge or did not provide a better fit to the data. Hence, the Negative
Binomial model was chosen as the final model. A comparison of observed and predicted
probabilities revealed that it closely described the data. The likelihood ratio test for
goodness-of-fit indicated that the full Negative Binomial model was significantly better
than the intercept-only Negative Binomial model (-LL
Full
= -178.45, -LL
Restricted
= -193.25,
Chi
2
(8)
= 29.61, p < .001). Results indicated that only three variables had significant
predictive effects, namely age (b= 0.55, SE = 0.26, p < .05), adolescent substance use (b=
1.02, SE = 0.27, p < .001), and low educational attainment (b= 0.60, SE = 0.27, p < .05).
Being older, reporting a higher frequency of adolescent substance use, and having left