RT Article
T1 Crime, time, and punishment: An exploration of selection bias in sentencing research
JF Journal of quantitative criminology
VO 1
IS 1
SP 103
OP 126
A1 Zatz, Marjorie S. 1955-
A2 Hagan, John
LA English
YR 1985
UL https://krimdok.uni-tuebingen.de/Record/176714380X
AB The sentencing decision reflects the culmination of a long series of processing and, thus, selection decisions, with cases leaving the system at each decision point. Accordingly, the substantive implications of bias due to sample selection are of particular concern for sentencing research. In an effort to assess the existence and manifestations of selection bias, the sentencing decision is modeled for three samples, each of which was selected from different stages of the justice process. Event-history data on felony arrests in the State of California over a 3-year period are used, along with a relatively simple analytic technique which reduces such bias. Results indicate that bias is introduced when censored observations are excluded from the analyses. Also, the effects of certain exogenous variables on sentence length differ, depending upon the selection criteria. Of these, the influence of pleading guilty rather than going to trial is especially interesting. Overall, our findings are consistent with the possibility that selectivity bias has concealed effects of sentence bargaining in some earlier studies.
K1 censoring
K1 event history
K1 Plea Bargaining
K1 Sentencing
K1 Selection Bias
DO 10.1007/BF01065251