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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

e put into assessing and addressing those clients’ needs. In order to do so, the lowest<br />

risk <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders in the agency needed to be assigned to large caseloads with minimal<br />

supervisi<strong>on</strong>, so that <str<strong>on</strong>g>of</str<strong>on</strong>g>ficers would be freed up to work more closely with high-risk<br />

clients. This represented a departure from the initial ‘<strong>on</strong>e size fits all’ approach to<br />

supervisi<strong>on</strong> previously used by the agency (see Fig. 2.1). A risk predicti<strong>on</strong> model was<br />

developed to assess which <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders were at low and high risk <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fending. <str<strong>on</strong>g>The</str<strong>on</strong>g> Low<br />

Risk Experiment then randomly assigned <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders predicted to be low risk to a lowintensity<br />

model <str<strong>on</strong>g>of</str<strong>on</strong>g> supervisi<strong>on</strong> (‘LIS’) or the normal model <str<strong>on</strong>g>of</str<strong>on</strong>g> supervisi<strong>on</strong> as described<br />

above (supervisi<strong>on</strong> as usual: ‘SAU’). <str<strong>on</strong>g>The</str<strong>on</strong>g> results <str<strong>on</strong>g>of</str<strong>on</strong>g> the experiment indicated that LIS<br />

can safely be used with low-risk <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders without increasing the risk <str<strong>on</strong>g>of</str<strong>on</strong>g> serious<br />

recidivism. APPD next plans to test the allocati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> high-risk <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders to high-intensity<br />

or regular supervisi<strong>on</strong>.<br />

Forecasting model<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> statistical model used to forecast the 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 is described in<br />

full in Berk et al. (2009). Random forests methods were applied to a dataset <str<strong>on</strong>g>of</str<strong>on</strong>g> all<br />

probati<strong>on</strong> and parole cases in Philadelphia between 2002 and 2004 to predict the risk <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

being charged with a new serious crime 2 within two years <str<strong>on</strong>g>of</str<strong>on</strong>g> the probati<strong>on</strong> or parole case<br />

start date. <str<strong>on</strong>g>The</str<strong>on</strong>g> predicti<strong>on</strong> was based <strong>on</strong>ly <strong>on</strong> the type <str<strong>on</strong>g>of</str<strong>on</strong>g> data that would be available to<br />

probati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g>ficers at intake. 3<br />

At the request <str<strong>on</strong>g>of</str<strong>on</strong>g> APPD, the model was designed to stratify<br />

61 per cent <str<strong>on</strong>g>of</str<strong>on</strong>g> cases as low risk, with the remainder either high risk (approximately 10 per<br />

cent) or neither low nor high (approximately 30 per cent) (Fig. 2.1). APPD also deemed<br />

67

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

Saved successfully!

Ooh no, something went wrong!