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.
<str<strong>on</strong>g>of</str<strong>on</strong>g> low-intensity supervisi<strong>on</strong> based <strong>on</strong> risk <str<strong>on</strong>g>of</str<strong>on</strong>g> serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fending, because there may be<br />
public and political anxiety about reducing criminal justice interventi<strong>on</strong> to adjudicated<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders. Thus, for low-intensity probati<strong>on</strong> to maintain credibility, it is arguably more<br />
sensible to dem<strong>on</strong>strate that few serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders slipped through the net than to show<br />
how many n<strong>on</strong>-serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders had their supervisi<strong>on</strong> requirements reduced. From the<br />
model standpoint, we must ensure a high positive predictive value, which indicates the<br />
proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> predicted low-risk <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders who were actually low risk, and high<br />
specificity (proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders who received a n<strong>on</strong>-low risk predicti<strong>on</strong>).<br />
Large values for these two measures indicate a low rate <str<strong>on</strong>g>of</str<strong>on</strong>g> false positives (predicted lowrisk<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders who commit serious crimes).<br />
C<strong>on</strong>versely, we expect to see lower<br />
sensitivity and lower negative predictive values, because the sample c<strong>on</strong>tains many false<br />
negatives. Most n<strong>on</strong>-low risk probati<strong>on</strong>ers do not go <strong>on</strong> to be serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders in the<br />
two-year follow-up. Serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses are rare events in our sample, and n<strong>on</strong>-low risk<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders are not necessarily high risk. 9<br />
Low values <strong>on</strong> these two measures are more<br />
acceptable because there is no harm when an <str<strong>on</strong>g>of</str<strong>on</strong>g>fense is not committed, regardless <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
risk predicti<strong>on</strong>. However, we must also keep the purpose <str<strong>on</strong>g>of</str<strong>on</strong>g> the model in mind: the<br />
diversi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> a majority <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders in APPD’s caseload to low-intensity supervisi<strong>on</strong>. If<br />
sensitivity is too low (too few n<strong>on</strong>-serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders received low-risk predicti<strong>on</strong>s), that<br />
goal will not be fulfilled.<br />
Table 3.5 shows the results <str<strong>on</strong>g>of</str<strong>on</strong>g> these tests using the model definiti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> severity.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> positive predictive values are high at all thresholds, indicating a low rate <str<strong>on</strong>g>of</str<strong>on</strong>g> false<br />
positives in general. This is promising, but will be driven by the very low sample<br />
prevalence <str<strong>on</strong>g>of</str<strong>on</strong>g> serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fending post-risk assessment. <str<strong>on</strong>g>The</str<strong>on</strong>g> present cut-<str<strong>on</strong>g>of</str<strong>on</strong>g>f point <str<strong>on</strong>g>of</str<strong>on</strong>g> 0.5<br />
158