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
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
the proporti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> false positives and false negatives expected in the final model to be<br />
operati<strong>on</strong>ally acceptable.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> false positives (<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders err<strong>on</strong>eously<br />
identified as low risk) was set at 5 per cent, and the proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> false negatives<br />
(<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders err<strong>on</strong>eously identified as high risk) was 20 per cent. A higher false negative<br />
rate was accepted given the lesser public safety c<strong>on</strong>cerns around this type <str<strong>on</strong>g>of</str<strong>on</strong>g> error.<br />
Experimental design<br />
Selecti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Cases<br />
APPD selected the West and Northeast regi<strong>on</strong>al supervisi<strong>on</strong> units as the sites from<br />
which experimental participants would be drawn. All cases active <strong>on</strong> probati<strong>on</strong> in these<br />
two units <strong>on</strong> July 27, 2007 were extracted. <str<strong>on</strong>g>The</str<strong>on</strong>g> random forests model was applied to<br />
each case to produce an individual risk assessment (some probati<strong>on</strong>ers had multiple<br />
cases) in the form <str<strong>on</strong>g>of</str<strong>on</strong>g> a ‘reliability score.’ <str<strong>on</strong>g>The</str<strong>on</strong>g> reliability score is a number between 0 and<br />
1. Cases with a reliability score above 0.5 were designated as low risk. From this<br />
assessment, 2,859 <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders were serving probati<strong>on</strong> terms for low-risk <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses. <str<strong>on</strong>g>The</str<strong>on</strong>g>y<br />
were pre-screened for eligibility for the experiment. Low-risk cases were excluded from<br />
the random assignment pool if any <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the following factors made them ineligible for<br />
low-intensity supervisi<strong>on</strong>:<br />
<br />
<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> case was due to expire within thirty days <str<strong>on</strong>g>of</str<strong>on</strong>g> the extracti<strong>on</strong> date.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fender was placed under the supervisi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> a specialized unit by court order<br />
after the extracti<strong>on</strong> date.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fender was in an existing low-risk caseload. 4<br />
68