12.03.2014 Views

The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk

The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk

The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

directi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> effects is as we would expect given prior research. In the RCTs, assignment<br />

to intensive supervisi<strong>on</strong> made no difference to the prevalence <str<strong>on</strong>g>of</str<strong>on</strong>g> rearrest or rec<strong>on</strong>victi<strong>on</strong><br />

(mean OR for arrests = .93; p ≤ .72; mean OR for c<strong>on</strong>victi<strong>on</strong>s = .98, p ≤ .80). We also<br />

see no significant effect <str<strong>on</strong>g>of</str<strong>on</strong>g> ISP in the quasi-experiments, although note that the raw effect<br />

sizes (especially for c<strong>on</strong>victi<strong>on</strong>s) show moderate reducti<strong>on</strong>s in recidivism associated with<br />

ISP and the small number <str<strong>on</strong>g>of</str<strong>on</strong>g> studies in these categories reduce the likelihood <str<strong>on</strong>g>of</str<strong>on</strong>g> a finding<br />

being statistically significant (mean OR for arrests = .83, p ≤ .10; mean OR for<br />

c<strong>on</strong>victi<strong>on</strong>s = .60, p ≤ .10). <str<strong>on</strong>g>The</str<strong>on</strong>g> forest plot for c<strong>on</strong>victi<strong>on</strong> outcomes (Figure 1.4) suggests<br />

that there is a lot <str<strong>on</strong>g>of</str<strong>on</strong>g> uncertainty in this model: the c<strong>on</strong>fidence interval around the mean<br />

effect size is clearly very large.<br />

Our analyses also indicate an increase in technical violati<strong>on</strong>s associated with ISP.<br />

Across the RCTs, intensive supervisi<strong>on</strong> was associated with a 54 per cent increase in the<br />

odds <str<strong>on</strong>g>of</str<strong>on</strong>g> a technical violati<strong>on</strong> (mean OR = 1.54, p ≤ .06). A smaller, n<strong>on</strong>-significant<br />

increase <str<strong>on</strong>g>of</str<strong>on</strong>g> 29 per cent was observed across the quasi-experiments (mean OR = 1.29, p ≤<br />

.22) but is again based <strong>on</strong> a much smaller subset <str<strong>on</strong>g>of</str<strong>on</strong>g> studies. Finally, we found no effect<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> ISP for the subset <str<strong>on</strong>g>of</str<strong>on</strong>g> studies reporting drug related effects (mean OR = 1.14, p ≤ .10).<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> mean effect sizes should be interpreted with cauti<strong>on</strong> given the very small number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

events in some <str<strong>on</strong>g>of</str<strong>on</strong>g> the studies (see Figure 1.7).<br />

Table 1.2 indicates substantial heterogeneity across studies in four <str<strong>on</strong>g>of</str<strong>on</strong>g> our seven<br />

analyses, as evidenced by the highly significant Q statistics (RCT arrests: Q = 61.55, p <<br />

.001; RCT technical violati<strong>on</strong>s: Q = 120.11, p < .001; quasi-experiment technical<br />

violati<strong>on</strong>s: Q = 18.91, p ≤ .004; quasi-experiment c<strong>on</strong>victi<strong>on</strong>s: Q = 22.74, p < .001). This<br />

indicates that there is more variability between studies than we would expect from the<br />

28

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!