Data for each of the statistical models was analyzed using logistic regression, employing a standard logit equation:

logit P = b₀+ b₁X₁ + b₂X₂ + ...+ b_kX_k

where P is the dependent variable, X₁-X_k are the independent variables, and b_o-b_k. are coefficients generated by the model.

To test the crime control motivation of prosecutors, the bivariate dependent variable �DA Discretion� was tested against a group of independent variables broken down into three categories, each representing a separate subgrouping of related components. The first of the three subgroups, Model 1, consists of variables related to the increased likelihood of recidivism by the offender. It is expected that prosecutors are more likely to use prosecutorial discretion when the likelihood of recidivism is reduced. The bivariate dependent variable is the use of prosecutorial discretion, as measured by the decision to strike a prior strike. The independent variables include: time period between offenses (based upon the number of years between the last strike and current offense), remote prior strikes (as coded by the prosecutor evaluating the case using subjective criteria), no recent criminal history (coded by the prosecutor, signifying a clean felony record in recent years), prior strikes from same case (cited if strikes arose from the same incident), and drug abuse (as measured by previous drug convictions). The hypothesis, variable list, and expected directions of the logit coefficients for Model 1 are summarized in Figure 1.

or go to...
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11
12 | 13 | 14 | 15 | 16 | 17 | 18 | 19