TABLE VIII.
JTDC Study 3 – Treatment Effect on Measures of Recidivism within Given Number of Months since Release; Sample of Youth With Complete 18-Month Follow-Up Data
| Months since release | 2 | 4 | 6 | 8 | 10 | 12 | 14 | 16 | 18 |
|---|---|---|---|---|---|---|---|---|---|
| Panel A: Effect on probability of readmission by X months | |||||||||
| Effect of participation (IV) | − 0.1342** | − 0.1893*** | − 0.1984*** | − 0.2010*** | − 0.1764** | − 0.1689** | − 0.1911*** | − 0.1598** | − 0.1643** |
| (0.0682) | (0.0734) | (0.0735) | (0.0722) | (0.0704) | (0.0690) | (0.0683) | (0.0673) | (0.0669) | |
| Control complier mean | 0.345 | 0.563 | 0.681 | 0.738 | 0.733 | 0.768 | 0.793 | 0.790 | 0.795 |
| Panel B: Effect on number of times youth readmitted within X months | |||||||||
| Effect of participation (IV) | − 0.1633* | − 0.3295*** | − 0.3411** | − 0.4739*** | − 0.5353*** | − 0.5286** | − 0.6911*** | − 0.7031*** | − 0.7045*** |
| (0.0835) | (0.1235) | (0.1493) | (0.1746) | (0.1956) | (0.2148) | (0.2331) | (0.2464) | (0.2522) | |
| Control complier mean | 0.387 | 0.816 | 1.041 | 1.386 | 1.578 | 1.734 | 1.995 | 2.117 | 2.196 |
| Panel C: Effect on number of times youth readmitted or rearrested within X months | |||||||||
| Effect of participation (IV) | − 0.1819 | − 0.3323* | − 0.2656 | − 0.4036 | − 0.5434* | − 0.5681 | − 0.6757* | − 0.6610 | − 0.6595 |
| (0.1168) | (0.1803) | (0.2247) | (0.2702) | (0.3191) | (0.3574) | (0.4019) | (0.4382) | (0.4615) | |
| Control complier mean | 0.653 | 1.346 | 1.855 | 2.486 | 3.063 | 3.497 | 3.998 | 4.420 | 4.803 |
Notes. n = 2,693. Table presents results from linear probability model with dependent variable indicating whether youth returned to JTDC within X months, how many times they returned, and whether they returned or were rearrested within X months; average marginal effects from probit model are similar. Some individuals have multiple spells and provide multiple data points in the sample. We present robust standard errors clustered on the individual. LATE operationalizes participation as spending more than 5% of a stay in a CBT unit. All regressions include baseline covariates and day-of-admission fixed effects. * p < .10, ** p < .05, *** p < .01.