Alex Keena

Measuring Partisan Bias

We claimed in Chapter 1 that there was a sharp increase in partisan bias as a result of the redistricting that followed the 2010 Census. In this and the following two chapters, we provide systematic evidence for this and ask why this has happened. Empirically, we have two main tasks. The first is to measure partisan bias at both the national and state level. This will allow us to test whether there has been a significant increase in partisan bias following the 2010 districting round. This is the task of this chapter. Assuming that there is partisan bias, the second task is to test various explanations of why the level of bias has changed and why bias occurs where it does. We have hypothesized that this is a thoroughly political phenomenon – state governments pursue partisan advantage when they are unconstrained by the courts. However, other explanations are possible. Chapter 4 considers nonpolitical alternative explanations. For example, it has been suggested that bias against the Democratic Party is inevitable because of demographic reasons, such as the fact that Democratic voters are concentrated in urban areas or the fact that it is necessary to draw majority-minority districts to comply with the Voting Rights Act. Chapter 5 considers political explanations for the patterns of bias we see. It asks whether we can explain the bias we observe in terms of the political motivations and capabilities of those in charge of drawing districts. We make use of the established methodology for measuring partisan bias in districting (see Gelman and King 1994b). This has been widely applied, tested, and peer reviewed. It is based on the idea of symmetry: If the Democrats get 52% of the votes and win 60% of the seats as a result, then if the Republicans were to win 52% of the vote in the next election they should get 60% of the seats. We can measure how much a districting plan violates symmetry by estimating how many seats a party would win in the event of different levels of popular support. While popular support varies from election to election, the pattern of support between districts remains quite stable.

The Unnoticed Revolution

Shortly after the 2012 elections, when all the votes had been collated, it was noticed that the Republicans had won 234 seats out of 435 in the House of Representatives, even though the Democrats had won a slight majority of the vote (50.6%). This in itself was not particularly remarkable. After all, the Republicans had won a House majority in 1952 and 1994 without winning a majority of the vote. These proved to be one-off events. However, we shall argue, what happened in 2012 was different. The result in 2012 was the result of systematic bias produced by the new districts adopted after the 2010 Census. Furthermore, eight years earlier the Supreme Court decision in the case Vieth v. Jubelirer (2004) made challenging a districting plan on grounds of partisan gerrymandering practically impossible. The combination of the Supreme Court effectively permitting partisan gerrymandering and the willingness of many state governments to draw districts for maximum partisan advantage has profound consequences. It will effectively determine control of the House of Representatives for the next decade; it provides a loophole for the egalitarian and democratizing electoral reforms the Supreme Court required in the 1960s; it means that state governments rather than voters can determine the character of a states congressional delegation; and it challenges the Madisonian principle that at least one part of government should be directly elected by the people. The Supreme Courts decision in Vieth v. Jubelirer (2004) drew little attention in the popular press or even academic circles – certainly there was nothing comparable to the response to the case of Citizens United (2010), which reduced restrictions on independent political expenditures. This is not completely surprising, as its significance was not obvious. It did not overturn the decision of the district court. In response to a complaint of political gerrymandering in the congressional districts of the State of Pennsylvania, the Supreme Court affirmed the decision of the lower court to not overturn the plan. The Supreme Court agreed on this decision 5–4, but it could not agree on a common opinion. Thus technically it did not overturn the courts previous finding in Davis v. Bandemer (1986a) that political gerrymandering was justiciable – something the Court could rule on.

Conclusion: Vieth, Majority Rule, and One Person, One Vote

The Supreme Courts decision in Vieth v. Jubelirer (2004) and the partisan districting that followed the 2010 Census raise fundamental questions about American democracy. They force us to reconsider the principles of majority rule, equal protection, and one person, one vote. In an immediate sense, the current level of partisan gerrymandering challenges the principle of majority rule, in that control of the House may depend as much on how the districts are drawn as on how people vote, and a minority may often dominate a majority. However, the consequences of the Supreme Courts ruling are more profound than who controls the House for the next decade. The Vieth decision undermines the one-person, one-vote jurisprudence of the 1960s that began with Baker v. Carr (1962). The decision respects the finding in Wesberry and Reynolds that all districts have to have equal population. However, this requirement means little or nothing if state legislatures are able to use partisan gerrymandering to achieve the same ends. The fact that state legislatures can to a substantial degree fix the outcome of House of Representatives elections also subverts the Madisonian principle that at least one branch of the legislature should be chosen directly by the people. This marks a major change in the effective constitution of the United States. However, before considering the broader significance of Vieth and the partisan bias that followed the 2010 districting round, we need to summarize the empirical findings of this book. We can confirm that partisan bias increased sharply in the 2010 districting round. We find that the bias of the House of Representatives in 2012 was around 9% (measured by asymmetry). Roughly speaking, this means that if there were a 50/50 vote split, one party would win 55% of the seats, while the other would get 45%. This compares to a 3% bias in the 2000 districting round. In many individual states, the bias is far, far greater. There are numerous states where the asymmetry is between 30% and 40%. Or, alternatively, one party can win between two-thirds to three-quarters of the seats with only half the votes. The immediate political implication of this bias is that the House of Representatives is likely to remain under Republican control for the next decade.

Answering Justice Scalia's Challenge to Equality: Does Equal

So far we have considered the effects and the significance of the Vieth v. Jubelirer (2004) ruling and the redistricting that followed it. We have found that it gives state governments a very powerful tool to manipulate election outcomes. Indeed, by drawing districts (in)appropriately, a state legislature can turn a minority of voters into an overwhelming majority of the states federal House delegation. Now, however, we turn to the case itself and, in particular, how the finding may be challenged in the future. We find that this hinges on a fundamental question of political philosophy: Does the United States Constitution require that the House of Representatives respect the majority rule principle? That is, is it required that a majority of voters be able to elect a majority of representatives? Writing for a plurality of the Court on Vieth v. Jubelirer (2004), Justice Scalia argued that the Court should not consider complaints of partisan gerrymandering because no standard exists for adjudicating such cases. That is to say, there is no standard that can be derived from the Constitution and then be practically applied to determine whether unconstitutional partisan gerrymandering has occurred. A fifth Justice (Justice Kennedy) agreed that no such standard currently existed but held out hope that one might be found. This defines the task for anyone who may wish to challenge the Vieth ruling. It is necessary to find a standard that is both required by the Constitution and provides a practical test that rules out partisan gerrymandering. The majority rule principle is an obvious candidate for such a standard of political fairness. It is clear that partisan gerrymandering violates the majority rule principle. As we saw in Chapter 3, in some states, it is necessary for one party to win 58% of the vote to win a majority of the states congressional delegation. Furthermore, as was also demonstrated in Chapter 3, we clearly have the technology to determine whether a districting plan has a systematic tendency to violate the majority rule principle – this was provided by Gelman and King (1994b). Majority rule is also a basic principle of political fairness. Indeed, it is often argued that it is synonymous with democracy.

Geographic Explanations for Partisan Bias

In the previous chapter, we found considerable evidence of partisan bias in the districting plans in many states that, when aggregated over the entire nation, gave the Republican Party a considerable advantage. Furthermore, we found that the level of bias in the 2010 districting round was approximately three times that in the 2000 round. However, it could be argued that this bias is not the result of deliberate partisan gerrymandering, but rather the inevitable results of geographical or demographic factors. For example, it could be argued that the observed bias is a result of the fact that Democratic voters tend to be concentrated in urban areas. This leads to Democratic votes being inefficiently distributed without the need for conscious gerrymandering. In this chapter, we consider the plausibility of such explanations. There are a number of ways partisan bias could be explained without the need for deliberate gerrymandering. First, as we just noted, it could result from the fact that Democratic voters tend to be concentrated in urban areas. As a result, the Democrats win districts in these areas by far greater margins than they need, leaving the Republicans to win a greater number of nonurban districts by smaller margins. Thus the kind of bias we described in the previous chapter comes about as a result of where people live and would be hard to avoid regardless of who drew the boundaries. Alternatively, it has been suggested that the desire to provide descriptive representation for minorities and the need to draw majority-minority districts to comply with the Voting Rights Act might reinforce the concentration of Democratic voters in overwhelmingly Democratic districts. Finally, increased bias might result from the fact that the technology for districting (in particular geographic information systems) has improved. This explanation is slightly different in that it does not claim that bias is not the result of partisan motives. However, it argues that the increased bias we observe is not the result of a change in the motives of districting authorities or the constraints they face, but rather a result of the fact that they are more able to gerrymander accurately. That is, the motives have not changed, but the tools available to fulfill those motives have improved.