The Case Against Partisan Gerrymandering

Superficially, the question in Gill v. Whitford—the blockbuster case about partisan gerrymandering in Wisconsin, to be heard next month by the Supreme Court—is a legal one. Is the test for gerrymandering adopted by the trial court “discernible,” or rooted in the high court’s precedent, and “manageable,” or consistent in the results it would produce? Lurking beneath this question, though, is a more fundamental debate about the nature of voting and representation in modern American politics. The position the Supreme Court takes in this debate will likely influence its decision more than any legal argument.

In its amicus brief, the Wisconsin legislature tells a rosy tale of voter and legislator behavior. Wisconsin voters do not “blindly support one party or the other,” the brief contends. Rather, they often split their tickets, or change their allegiances from one election to the next, based on “issues that matter to the electorate” and “the quality of the candidates and their campaigns.” Legislators, similarly, are highly responsive to their constituents’ preferences. Competitive races “force the winning candidate to adopt more moderate, centrist positions,” while “a landslide may allow that candidate to move further from the center.”

This is indeed how American politics worked in the not-too-distant past. Between the 1964 and 1972 presidential elections, for example, there was an almost 50-point net swing in Republicans’ favor. In this era, Congress was also full of conservative Southern Democrats and liberal Northeastern Republicans. Party was plainly a poor predictor of voters’ and legislators’ choices.

If this story were still true today, there would be little reason for the courts to police partisan gerrymandering. Consider voters: If they ignore party labels and cast their ballots based on issues and candidates, gerrymandering would not be durable enough to warrant judicial intervention. An unfair outcome in one election could easily disappear in the next one. Or take legislators: If they mirror their constituents’ opinions, it barely matters how many Democrats or Republicans are elected. Regardless of its partisan composition, the legislature would pass essentially the same laws.

On the other hand, the plaintiffs in Whitford (whom I help represent) argue that party matters a great deal, both to voters and to legislators. Partisan affiliation, in this view, is the most potent driver of voter behavior. Few voters split their tickets or change their minds from one year to another. Likewise, most legislators are highly polarized. There is not much difference in terms of voting record between legislators from competitive districts and their co-partisans from safe ones.

If this account is accurate, partisan gerrymandering would be much more troubling. It would be durable because voters would be unlikely to switch their votes in sufficient numbers to flip many districts from one party to the other. It would also be democratically distorting, because the extra Democrats or Republicans elected due to the practice would tend to toe the party line. A gerrymandered legislature would thus pass different laws—ones less reflective of the will of the people—than a fairly apportioned body.

So, which side has the better of this debate? Is partisanship a mere sideshow, as the Wisconsin legislature would have it, or is it the main event in modern American politics? The evidence strongly supports the latter perspective.

Start with the prevalence of split-ticket voting. The below charts plot the number of Republican assembly votes in each ward in Wisconsin’s 2012 and 2016 elections versus the number of Republican presidential votes in those same two elections. If split-ticket voting were common, there would be a weak connection between these variables. Conversely, if voters generally backed the same party at all levels of government, the link would be strong. The charts reveal that the link is not just strong; it is virtually ironclad. In both 2012 and 2016—the year when Donald Trump’s candidacy supposedly broke all the usual rules—there was a near-perfect correlation between assembly votes and presidential votes. In 2012, the correlation was 0.984. In 2016, it fell all the way to … 0.979.

Now turn to the consistency of voters’ preferences over time. The next chart plots the number of Republican assembly votes in each ward in Wisconsin’s 2016 election versus the corresponding number in 2012. If voters’ views were volatile, shifting based on the issues and candidates in each election, these variables would be tied only weakly. On the other hand, if voters typically pulled the lever for the same party each year, the relationship would be strong. Again, the chart shows an extraordinarily tight relationship. The correlation between 2012 and 2016 assembly votes is a whopping 0.975.

As for legislators, data exists for not only the Wisconsin State Assembly but also state houses around the country. In a recent article, political scientists Devin Caughey, Chris Tausanovitch, and Chris Warshaw examined how state house members’ voting records are linked to their margins of victory in the previous election. They concluded that the margin of victory makes almost no difference for purposes of representation. A Democrat (or Republican) from a swing district is nearly as liberal (or conservative) as one from a safe district. The impact of party dwarfs that of district composition.

This evidence suggests that partisan gerrymandering is durable because voters’ partisan preferences are mostly fixed. One way to measure if that’s so is to look at the “efficiency gap,” a common quantitative measure of gerrymandering. (To learn more, read this paper that I co-authored.) One of the experts in Whitford, political scientist Simon Jackman, compared district plans’ initial efficiency gaps with their average efficiency gaps over their lifetimes. He found that plans that are skewed in a party’s favor in the first election tend to remain tilted in that party’s favor for the rest of the decade. Gerrymandering is indeed highly persistent.

The evidence also suggests that gerrymandering distorts the democratic process by enabling more of a party’s loyalists to be elected. To confirm this, Caughey, Tausanovitch, and Warshaw examined how an efficiency gap in a party’s favor affects the ideological midpoint of the legislature and the laws actually passed by the body. They concluded that a pro-Democratic efficiency gap pushes representation and policy in a liberal direction, while a pro-Republican efficiency gap makes them more conservative. Just by changing how district lines are drawn—without persuading a single voter—gerrymandering warps the entire political system.

The underlying debate in Whitford, then, has a clear winner and a clear implication. Far from being tangential, party is currently the most important driver of voter and legislator behavior. This fact causes today’s gerrymandering to be both durable and democratically subversive. These characteristics, in turn, cry out for judicial intervention. Without it, in gerrymandered states, the government will persistently flout the public will.