A popular tactic for fighting a ballot initiative is to claim it violates the single subject rule, leaving it to a panel of judges to decide its fate and possibly prevent it from reaching the voters. To understand the role of the single subject rule in initiative politics, then, we have to understand how judges make their decisions. The challenge facing judges, as we discussed in our previous posts, is that the definition of a “subject” is inherently subjective and infinitely malleable. In the absence of an objective definition, how do judges make their decisions?
To assess what factors influenced how judges decide, we examined 765 votes in 154 single subject cases during the decade 1997–2006 by the supreme and intermediate appellate courts in California, Colorado, Florida, Oregon, and Washington. These states are heavy initiative users, and contain a mix of aggressive and restrained stances toward the single subject rule.
Because there is no objective definition of a subject, we were interested in the possibility that judges are led, perhaps subconsciously, to find a single subject violation when they don’t like the substance of an initiative. In addition, because aggressive enforcement of the single subject rule requires judges to take a stronger position on the number of subjects than restrained enforcement, we were curious if single-subject decisions are more connected with political preferences in states with aggressive judicially created single-subject tests compared to states with restrained tests.
As we explain below the fold, we found some very interesting empirical results.
For each single subject case, we identified the participating judges and collected information on their personal characteristics, terms, and ideological orientation. We also collected information on the content of the initiatives that were under review.
The key step in our analysis was to identify the partisan orientation of judges and initiatives. For judges, we classified them as Democrats or Republicans based on the party of the governor that nominated them if we did not have direct evidence on their party affiliation. For initiatives, we classified them as “conservative” or “liberal/progressive” based on standard descriptions (and we had a category of “other” for initiatives that did not have an obvious ideological orientation). There are more sophisticated ways to measure judicial orientation in the literature, but our approach has the virtue of transparency, and it turns out to have a lot of explanatory power even if crude.
Overall, 30 percent of the judges we studied were classified as Republicans and 70 percent were Democrats. In contrast, 57 percent of the initiatives under review were classified as conservative in their policy orientation (such as a law denying government benefits to illegal immigrants) compared to 24 percent that were classified as liberal (such as pro-environmental regulation measures), with 19 percent in the “other” category (such as gaming measures).
We then compared voting behavior on initiatives that the judge was likely and unlikely to be sympathetic with: a judge was said to “agree” with the initiative if (1) the judge was a Democrat and the initiative was liberal/progressive, or (2) the judge was a Republican and the initiative was conservative; otherwise the judge was said to “disagree.” If judges enforce the single subject rule based on an objective standard, there would no correlation at all between votes and a judge’s personal sympathies toward the initiative.
However, we found that judges are more likely to uphold an initiative if they “agree” with it in the sense above than if they “disagree” with it. Research on other types of cases has found a correlation between a judge’s votes and his or her political orientation, so finding that partisanship predicts voting in single subject cases is not surprising. What is surprising is the magnitude of the effect, and the degree to which an aggressive enforcement approach appears to increase partisanship.
In states with a restrained approach to enforcement of the rule (California and Washington in our sample), we found that judges voted to uphold the initiative 88.3 percent of the time when they agreed with it compared to 80.6 percent of the time when they disagreed, a modest difference. In contrast, in states with aggressive enforcement (Colorado, Florida, and Oregon in our sample), judges voted to uphold initiatives they agreed with 83.2 percent of the time, almost the same approval rate as in restrained states, but voted to uphold initiatives they disagreed with only 41.1 percent of the time. Thus, in states with aggressive enforcement judges were 42.1 percent less likely to approve an initiative they disagreed with than an initiative they agreed with.
This is an enormous difference in voting behavior based on partisanship, the magnitude of which seems unprecedented in the literature on judicial voting, and which strongly points to the centrality of subjective considerations when courts attempt to apply the single subject rule aggressively.
We double-checked these simple comparisons using more sophisticated multivariate regression techniques (and also explored several other hypotheses about what motivates the votes of judges). We attempted to control for other factors that are likely to influence decisions, such as the text of measure (for example, how many words it contains), the level of the court (supreme versus intermediate appellate), and characteristics of the judge (such as years until the judge next stands for election). The basic pattern is robust to all of these potentially confounding factors: Whether or not a judge agrees with the initiative policy is a strong predictor of his or her voting behavior in aggressive states, but not in restrained states.
Readers wanting the detail and nuance should turn directly to our paper.
The bottom line is that partisanship matters – a lot – in states with aggressive enforcement but not in states with restrained enforcement. This undercuts the view that judges are able to apply the single-subject rule objectively according to judicially constructed tests embodied in single subject doctrine. Tomorrow, in our final blog post, we will turn to the normative implication of this finding.