2 51 PREDICTIVE PROPERTIES DYNAMIC INSTRUMENTS No significant effect was found for year of publication (b1 = -.02093, SE = .01267, Z = -1.652, p = .099, k = 14). Table 2.1. Effect sizes for research question 1: Predictive properties of dynamic risk assessment instruments by recidivism type (van den Berg et al., 2018, p.184) Recidivism Type Fixed-Effect Random-Effect Q I2 (%) N K L Cohen’s d 95% CI Cohen’s d 95% CI Sexual Incl. outlier (Allan & Dawson, 2002) 0.71 0.74 [.63, .79] [.66, .82] 0.70 0.76 [.61, .79] [.63, .90] 25.5 59.9 17.7 63.3 5,699 5,877 41 42 22 23 Violent 0.37 [.30, .44] 0.43 [.29, .57] 42.7 70.0 10,368 27 14 Any 0.64 [.56, .72] 0.64 [.56, .73] 9.8 7.6 3,405 20 10 Note. K = Number of studies; L = Number of unique samples Values in bold are statistically significant. p < .01 2.4.1.3 Predictive properties for any recidivism The fixed-effect weighted Cohen’s d on any recidivism was 0.64, 95% CI [0.56, 0.72], N = 3,405, based on 10 unique samples. The Q indicated the variability across studies was not significant (Q = 9.8, p = .37). The I2 (7.6%) suggested low variability among studies. 2.4.1.3.1 Moderators As shown in Appendix C3 in the online supplemental materials, the Qbetween analyses revealed no significant effects for categorical moderators. The fixed-effect metaregression of the continuous moderators revealed no significant effects either; year of publication (b1 = -.00624, SE = .01468, Z = -0.425, p = .671, k = 10), sample size (b1 = .00011, SE = .00020, Z = 0.545, p = .586, k = 10), and follow-up time (b1 = .00004, SE = .00091, Z = 0.044, p = .965, k = 10). 2.4.2 INCREMENTAL VALIDITY Only studies reporting incremental validity of dynamic risk assessment instruments over static risk assessment instruments were included in this meta-analysis. The average follow-up interval was 80.5 (range = 36–240) months for studies predicting sexual recidivism, 54.4 (range = 36–240) months for studies predicting violent (including sexual) recidivism and 73.6 (range = 9–240) months for studies predicting any recidivism. Observed recidivism rates were 13.0% for sexual recidivism (13 samples, n = 3,747), 11.0% for violent (including sexual) recidivism (11 samples, n = 9,019), and 33.6% for any recidivism (8 samples, n = 2,511). Table 2.2 presents the effect sizes for the incremental validity of dynamic risk assessment instruments developed for adult male
RkJQdWJsaXNoZXIy MjY0ODMw