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.

Notes<br />

1 In this study, murder, attempted murder, aggravated assault, robbery, and sexual <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses were deemed<br />

‘serious’ <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses.<br />

2 Intake informati<strong>on</strong> includes the <str<strong>on</strong>g>of</str<strong>on</strong>g>fender’s pers<strong>on</strong>al and residential characteristics, and informati<strong>on</strong> about<br />

the instant <str<strong>on</strong>g>of</str<strong>on</strong>g>fense and prior criminal history.<br />

3 Full details about the experimental design and how the sample was selected, assessed for eligibility, and<br />

randomly assigned may be found in Barnes et al. (forthcoming).<br />

4 An alternative approach could be to use incarcerati<strong>on</strong> status as a proxy for cost. <str<strong>on</strong>g>The</str<strong>on</strong>g> disadvantage is that<br />

we must use either actual incarcerati<strong>on</strong> data for our sample, or assume the types <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses that might<br />

result in a sentence <str<strong>on</strong>g>of</str<strong>on</strong>g> impris<strong>on</strong>ment. Assumpti<strong>on</strong>s may be too subjective given the discreti<strong>on</strong> involved in<br />

sentencing (although state sentencing guidelines could assist), and full incarcerati<strong>on</strong> data are not available<br />

for our sample. In particular, it is possible that some post-random assignment sentencing decisi<strong>on</strong>s are still<br />

pending given the relatively short period <str<strong>on</strong>g>of</str<strong>on</strong>g> time since the experiment ended. While assigning victim status<br />

to each <str<strong>on</strong>g>of</str<strong>on</strong>g>fense type also involves assumpti<strong>on</strong>s, the level <str<strong>on</strong>g>of</str<strong>on</strong>g> subjectivity is likely c<strong>on</strong>siderably lower than it<br />

would be for incarcerati<strong>on</strong>.<br />

5 Most <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses in the dataset are derived from the Pennsylvania C<strong>on</strong>solidated Statutes, Secti<strong>on</strong> 18 (Crimes<br />

and Offenses).<br />

6 <str<strong>on</strong>g>The</str<strong>on</strong>g> risk ratio is simply pE=1 /p E=0 (where p = probability and E = dichotomous exposure status), whereas<br />

the odds ratio is (p/(1-p) E=1 )/(p/(1-p) E=0 ). <str<strong>on</strong>g>The</str<strong>on</strong>g> odds ratio tends to overstate our ‘natural’ interpretati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

relative outcomes: if the exposed group has a 50% risk <str<strong>on</strong>g>of</str<strong>on</strong>g> disease and the unexposed group has a 25% risk,<br />

the risk ratio is clearly 2 (the exposed group is twice as likely to get the disease than the unexposed group),<br />

but the odds ratio is 3 (the odds <str<strong>on</strong>g>of</str<strong>on</strong>g> disease in the exposed group are three times those <str<strong>on</strong>g>of</str<strong>on</strong>g> disease in the<br />

unexposed group), which seems greater. <str<strong>on</strong>g>The</str<strong>on</strong>g> risk ratio also remains stable regardless <str<strong>on</strong>g>of</str<strong>on</strong>g> the size <str<strong>on</strong>g>of</str<strong>on</strong>g> the risk;<br />

the magnitude <str<strong>on</strong>g>of</str<strong>on</strong>g> the odds ratio is closer to the risk ratio when the probability <str<strong>on</strong>g>of</str<strong>on</strong>g> disease in each group is<br />

small, and further away when it is large. Following the example above, if the risks were reduced to 20%<br />

and 10% respectively, the risk ratio would still be 2 but the odds ratio would fall to 2.25.<br />

7 In additi<strong>on</strong>, an adjusted risk ratio that is substantially different from the unadjusted risk ratio (a 10-15%<br />

difference is a comm<strong>on</strong>ly-used rule <str<strong>on</strong>g>of</str<strong>on</strong>g> thumb) suggests that the stratifying variable is a c<strong>on</strong>founder.<br />

8 <str<strong>on</strong>g>The</str<strong>on</strong>g> race indicator variable was populated with data from two different sources, with <strong>on</strong>e source selected<br />

as the default. However, serious discrepancies arose because the categories <str<strong>on</strong>g>of</str<strong>on</strong>g> race in the two original<br />

sources were substantially different.<br />

9 On the other hand, it is also possible that n<strong>on</strong>-low risk probati<strong>on</strong>ers are in fact more serious <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders, and<br />

are more likely to be incarcerated as a result.<br />

10 That past behavior is <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the str<strong>on</strong>gest predictors <str<strong>on</strong>g>of</str<strong>on</strong>g> future behavior is <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the best documented<br />

findings in criminological research (e.g., Nagin & Paternoster, 1991; Nagin & Farringt<strong>on</strong>, 1992).<br />

172

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

Saved successfully!

Ooh no, something went wrong!