Variables that measure significantly different from zero include no history of weapons possession or use, documented history of mental illness and proof problems for prior strikes. Prosecutors are over ten times more likely to strike a strike for those defendants who do not have a history of weapons use or who have a documented case of mental illness. Although part of the high odds statistics can be traced to the small number of cases correlated with each of these factors (a total of 22 for no weapons use and 6 for mental illness), its significance to the evaluation process cannot be completely discounted. A defendant who has committed the minimum of three offenses without the aid of a weapon presents a lessened degree of dangerousness, which is a commonly cited mitigating factor. Using similar reasoning, a defendant that has a documented case of mental illness is less culpable than a three-strikes defendant who does not.

The variable proof problems is unique because it documents those cases in which the prosecutors feel that they are unable to get a three-strikes conviction in court due to evidentiary problems. David Neubauer found that prosecutors are more likely to plea bargain cases that are likely to end in an acquittal, thus the quality of evidence can be a significant predictor in the use of discretion (Neubauer 1974). Here, the odds of having a strike stricken are increased twenty times when prosecutors cite proof problems. Part of this can be explained by the small case numbers; out of 11 cases in which proof problems were noted, 9 resulted in a prior strike dismissal, although it also makes logical sense that if prosecutors perceive that the strikes will be thrown out by the court anyway because of evidentiary problems, they will go ahead and make the dismissal on their own.

Finally, the multi-faceted variable mitigating personal characteristics was cited infrequently by prosecutors. Out of 256 cases, it was used only 8 times to justify dismissing a strike (judges in San Diego County, on the other hand, used this 28 times to justify their dismissal actions). Given its relative disuse, it is not surprising to find that this variable is both non-significantly different from zero and in the wrong direction.

In Model 3, the final sub grouping tests the use of discretion against variables related to the seriousness of the defendant�s criminal offenses. The dependent variable, the decision to strike a strike, is measured with the following independent variables: total severity of current offenses (the severity score of all current offenses is totaled), total severity of strike offenses (the severity score of all previous strike offenses is totaled), number of prison priors (prison priors that qualify as sentencing enhancements are included), minor current offense (coded by the prosecutor if current offense is a minor felony), mitigating prior strikes (coded by the prosecutor if criminal conduct in prior strikes is less than what is normally presumed for this offense), and mitigating current offense (coded by the prosecutor if the defendant�s involvement in the current offense is less than what is normally presumed for this offense). The hypothesis, variable list, and expected directions of the logit coefficients for Model 3 are summarized in Figure 5.

Of all of the legal variables that can impact the use of discretion, seriousness of the crime consistently rank as the most significant. Offenders who had harmed their victim were less likely to have their cases dismissed (Holmes, Daudistel, and Farrell 1987) or receive a reduced sentence (Smith 1984). The decision to dismiss a case was also found to correlate with offense severity and the seriousness of the first charge (Gottfredson and Gottfredson 1986). In this model, the decision to strike a prior strike as based upon the seriousness of the criminal record is highly significant at the 99.9999% level (X² = 7996)

Individual variables that reach the level of statistical significance from zero include number of prison priors, de minimis offense, and mitigating facts of prior strikes. One of the strongest variables within this group is the variable measuring the number of prison priors. The likelihood of the prosecutor striking a strike in a given case decreases by 41% for each additional prior prison commitment on the offender�s record. One of the subjective evaluations, de minimis offense, usually cited when the third strike is a �wobbler� offense, increases the likelihood of the prosecutor striking a strike over seven and a half times when compared to other three-strike cases. Additionally, when the facts of the prior strikes are mitigating or not as serious in nature as the charge might indicate (for example, the defendant is convicted for residential burglary, yet the victim is his own mother and the offense is perpetrated in the home he shares with her), the likelihood of the prosecutor striking a strike is five times as great.

The fact that neither the current offense severity variable nor the strike offense severity variable is significant is surprising. The literature is rich with prior studies that confirm the importance of offense seriousness on the use of discretion. Part of the problem may lie with the severity scale, since criminal offenses are not presented in a perfect ratio order, however similar scales have been used in other studies without any significant consequences. Rather, what is likely driving the non-significance here is the fact that most of the three-strike offenders have similar offenses.

Data presented in Tables 5 and 6 reveals that most of the three-strikes defendants are actually committing offenses that rank low on the severity scale. As noted earlier, most of the current offenses being prosecuted as third strikes fell below the serious crime threshold and over three-fifths of the three-strike defendants committed an offense equivalent to residential burglary as their most serious strike offense. The remaining 40% of strike offenses were also congregated at the lower end of the strike severity scale. Only 6% of the three-strike defendants committed an offense equal to or worse than voluntary manslaughter. Second-strike numbers are even more dramatic; approximately 83% of all three-strike offenders committed strike offense no greater than the equivalent of residential burglary for their second strike. Considering that most of the current and strike offenses measured here are tightly packed around a small range of scale rankings, the non-significance in the analysis reflects the fact that prosecutors are less likely to distinguish between cases based upon offense seriousness. When deciding whether or not to strike a strike.

Model 4

As a test for data robustness, each of the sub-models analyzing the use of prosecutorial discretion is included in a larger model. Although the number of independent variables triples, their individual importance to the use of discretion remains essentially the same. Variables that were previously significantly different from zero and remain so in this final model include prison priors, de minimis current offense, remote prior strikes, no recent criminal history, mitigating prior strikes, no history of weapons use, documented history of mental illness, and proof problems. The only newly significant

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