We used a likelihood ratio test to compare the fit of this model using an exchangeable covariance structure, which restricts the variance in pre and post intercepts to be equal, to one with an unstructured covariance structure, which allows the variance to differ pre- and post-Prop 47 implementation, as a test of the change in county variance. The latter model is the most flexible and was used to generate the policy effects on the marginal probabilities of felony conviction. For county-specific estimates of outcomes pre vs. post-Prop 47, we generated empirical Bayes estimates of county random pre- and post-Prop 47 implementation intercepts from the models with unstructured covariance and calculated the linear combinations of fixed and random effects corresponding to pre- and post-Prop 47 periods. We also used fixed effects models with county dummy variables, interacted with the pre-post Prop 47 variable to generate marginal probability estimates of within-county pre-post change. A large proportion of arrests had no disposition . We assumed these cases were either never presented to the district attorney or the district attorney did not file charges, and therefore assigned an outcome of no conviction. We compared case characteristics for those with and without dispositions and found support for this assumption. Those without dispositions were less severe cases, and therefore less likely to result in conviction. For example, a larger proportion had no concurrent arrests . They were also less likely to have concurrent felony arrests . It is possible, however, that some of these cases resulted in convictions and the disposition was never reported by the court. We therefore conducted a sensitivity analysis that assumed the most severe extreme: that all sale/transport cases with missing dispositions received a felony conviction, and that all Prop 47 cases with missing dispositions received felony convictions if they occurred in the pre-Prop 47 period or if they included a concurrent felony arrest in the post period. Prop 47 arrests without concurrent felony arrests in the post period are unlikely to have received a felony conviction,cannabis growing facility since the Prop 47 offense was at that point classified as a misdemeanor. Changes in the outcomes of Prop 47 arrest events must be considered in the context of potential changes in the composition of arrestees .
When comparing the full sample prior to propensity score matching, a greater fraction of post-Prop 47 arrest events had concurrent arrests of other types , suggesting a decline in arrests when drug possession was the sole offense. Post-Prop 47 arrestees also appeared to differ in terms of criminal histories, with more prior arrests . The propensity score matched sample had better covariate balance. For sale/transport arrests, pre and post groups were much more similar . Though they were not compared statistically, we find the population arrested for these offenses appears quite different from those arrested for Prop 47 drug offenses. Prop 47 offenders had more numerous but lower level prior arrests and convictions. Racial differences were notable as well, with larger racial disparities among sale/transport arrests .After Prop 47 was adopted, Prop 47 drug convictions declined in both those arrested for Prop 47 drug offenses and those arrested for sale/transport. For those arrested for sale/transport, the decline in convictions overall was approximately equal to the decline in Prop 47 convictions. Although for the Prop 47 group, the percentage point decline in Prop 47 convictions was slightly larger than the decline in overall convictions, suggesting that it may have become more common to convict these arrestees of other crime categories. The question then becomes, which type of convictions replaced the Prop 47 convictions, if any, and did these replacements counteract potential reductions in felony convictions for Prop 47 offenses? Looking first at the Prop 47 group, there was a 14 percentage point decline in felony convictions , from 21.2% to 6.9%. This approximates the percentage point decline in felony Prop 47 convictions , suggesting felony convictions for other concurrent offenses did not increase. This is corroborated by the break-down of convictions for other offenses, which shows increases in convictions for misdemeanor drug and misdemeanor “other” categories, but not felony offenses . For example, the most common concurrent felony offense was felony property, present for 9.9% and 9.8% of Prop 47 drug arrests in pre and post periods, respectively. Yet the proportion of arrestees who received convictions for felony property offenses did not rise: 2.5% pre and 2.1% post.Declines appear to be driven by declines in the fraction of these arrestees who were ultimately convicted of felony Prop 47 offenses.
The likelihood of any felony conviction in this group declined 7.1 percentage points, approximately the drop seen in the proportion convicted of a Prop 47 felony with no other felony drug conviction . In contrast, no significant change occurred in felony convictions for sale/transport, suggesting prosecutors did not pursue more convictions for these offenses as a means to maintain prior levels of felony drug convictions post-Prop 47.With regard to Prop 47 arrests, pre and post estimates for each county suggest counties where felony convictions were more likely pre-Prop 47 were reduced towards zero to a greater degree, such that post-Prop 47 outcomes were more similar across counties. Mixed models with random pre-Prop 47 intercepts, random coefficients for the policy effect, and an unstructured covariance structure allowing for a correlation between intercept and slope random effects, showed a significant, negative covariance between random effects . Aligning with the pattern depicted in Figure 3.2, this suggests counties where felony convictions were more likely in the pre-period also declined more towards the less punitive counties, reducing the variance across counties. The reduction in county differences is corroborated by variance estimates from models with county specific random intercepts for the pre and post periods . We find a larger variance in county random intercepts in the pre-Prop 47 period . The likelihood ratio test comparing an exchangeable covariance structure as a nested model indicated that the unstructured covariance structures which allowed pre- and post-Prop intercept variances to differ, was a better fit to the data . To put this in concrete terms, of the 56 California counties, prior to the policy the most punitive county had a conviction probability of .38 , whereas the least punitive county had a conviction probability of .04 . After Prop 47 was adopted,cannabis grow system the most punitive county had a conviction probability of .19 whereas the least punitive county had a conviction probability of .02 . Another way to conceptualize these results is in terms of how discrepant the statewide probability of felony conviction was from the least punitive county pre-Prop 47, and the extent to which that discrepancy changed post-Prop 47. Prior to Prop 47, the statewide probability of felony conviction was 17 percentage points higher than the least punitive county , meaning 81% of statewide felony convictions following Prop 47 drug arrests would not have occurred if prosecuted in the least punitive county.
Whereas after Prop 47, the statewide probability was just 3 percentage points higher than the least punitive county prior to passage.There was also significant variation across counties in the likelihood of felony conviction following a sale/transport arrest , ranging from 0.05 in Merced County to 0.51 in Calaveras County in the pre-Prop 47 period. However, mixed model results indicated the significant variance in the pre-Prop 47 period , did not decline post-Prop 47 . This suggests that, while people arrested for sale/transport were less likely to ultimately get a felony conviction after Prop 47 was adopted, this effect did not vary substantially across counties, and no county showed an increase in felony conviction probability for sale/transport arrests. In other words, it does not appear that more punitive counties altered plea bargaining practices for sale/transport arrests to retain pre-Prop 47 levels of felony convictions, as this would have resulted in an increase in variance in felony conviction probabilities for this category of arrest. Findings aligned with sensitivity analyses that assumed all cases with missing dispositions received felony convictions .In this study of the change in felony convictions in California counties after Proposition 47 reduced criminal penalties for drug possession, we found significant declines in the likelihood of a felony conviction following arrests for Prop 47 drug offenses and non-Prop 47 felony drug offenses . Prior to Prop 47, dramatic geographic inequalities in probability of felony convictions after drug possession arrests prevailed between counties, and these geographic inequalities were substantially reduced after adoption of Prop 47. The reduction in felony convictions aligns with reports from the Judicial Council of California on reductions in felony filings following Prop 47 passage , while providing new evidence that reductions led to declines in geographic disparities in felony convictions for drug arrests. By holding county-specific case characteristics constant across time, this study identified a reduction in the excess variation that was attributable to county practices. This impact likely reflects that Prop 47 eliminated prosecutorial discretion for how drug possession can be charged. While previous research has found that the county-specific interpretation and implementation of reforms tends to reinforce the preexisting prosecution and sentencing practices within the county , results from the current study do not indicate counties attempted to mitigate the effects of Prop 47 with felony filings for concurrent offenses, or reducing plea bargaining for sale offenses.
Several factors could explain why Prop 47 led to reductions in geographic disparities in case outcomes, when other reforms have not. Prop 47 was a voter initiative, and considering the influence of community priorities for law enforcement on charging policies and decisions, prosecution practices may be more responsive to these types of reforms. Secondly, Prop 47 called for reduced criminal penalties, whereas prior studies have evaluated reforms like three strikes laws which maximize punishment. Maximizing punishment is costly, whereas reducing it can assuage overburdened courts. Therefore, we may be more likely to see change resulting from reforms that call for lesser criminal penalties, and especially when that call comes from the public. Reducing variation in the likelihood of a felony conviction for two equivalent cases mitigates inequalities in criminal justice exposure due to unequal applications of the law. However, requiring that all drug possession offenses be prosecuted as misdemeanors also suggests that cases with different characteristics are now being treated more similarly. A defendant can still be convicted of a felony for concurrent felony offenses, so it is the effect of criminal history on case outcomes which we would expect to be minimized post-Prop 47. Criminal history is strongly associated with race/ethnicity, which may reflect biases and practices in drug law enforcement , while increasing the severity of punishment for subsequent drug offenses . There is evidence Prop 47 in fact reduced the effect of criminal histories in San Francisco, where prior to Prop 47, racial disparities in case dispositions and sentencing were attributable to more extensive pretrial detention and criminal histories among black defendants . When Prop 47 reclassified drug possession offenses to misdemeanors, these characteristics had lesser effects on case outcomes, and racial disparities declined. Further research could assess whether findings from San Francisco apply statewide. There are also implications for substance use disorder treatment. Prop 47 generated $103 million in savings in the first year, awarded through grants to counties to increase access to substance use disorder and mental health treatment, and education . Counties with few felony convictions pre-Prop 47 may have had greater support for and availability of drug diversion options which allow dismissal of charges for successful drug treatment completion. However,Prop 47 generated concerns that without the possibility of a felony conviction, the incentive to engage in treatment would be removed . Prior research has suggested that, as compared to volitional substance users, individuals with more severe substance use disorders tend to fail to meet the court’s conditions for diversion and ultimately receive harsher termination sentences . If this were the case, it would be logical that this group would opt out of diversion options now that the sentence for drug possession is less severe. Whether this is the case, and if so, understanding successful strategies counties have developed to increase access to needed treatment through other routes, would be valuable. CA DOJ’s Statewide Automated Criminal History System data is the most comprehensive data source available for studying criminal justice policy changes in the state, and has been used in significant studies of Prop 47, as well as other reforms such as Prop-36, which increased drug diversion following arrest .