Abstract
In this note, we revisit the work of Carson and Crespinto examine the effect of different redistricting plans on competitiveness in U.S. House elections. Similar to the previous results, our probit estimates on an expanded dataset that includes redistricting cycles from 1972 to 2012 reveal that commission and court-drawn districts experience more competition on average than those drawn by legislatures. These results provide additional support for the hypothesis that one way to increase the competitiveness of congressional elections is to allow extra-legislative bodies to draw congressional district boundaries.
Introduction
Unable to come up with a redistricting plan for the 1966 congressional elections, the Maryland General Assembly left it to the courts to draw new district boundaries. After the frustration with the previous course of action, the state government decided to experiment with a relatively new method and appointed a joint committee that consisted of delegates, state senators, and administrative representatives, to devise new congressional districts following the 1970 census. (Meyer 1971) Although the legislature maintained the authority to approve the final maps, the goal was for the committee to do the vast majority of the work. This represented an attempt to take the initial planning outside of the normal legislative process and formed the basis for a prototype redistricting commission. In the following decade, additional states such as Connecticut, Hawaii, Maine, and Montana also tried to move redistricting, partially or in whole, toward an extra-legislative process. This trend continued and picked up speed in each subsequent round of redistricting.
Recently, Carson and Crespin (2004) studied if the level of competition for commission-, court-, and legislative-drawn plans differed following the 1990 and 2000 rounds of redistricting. In this note, we revisit and update the previous work by including the results from the 2012 elections as well as two additional years, 1972 and 1982. Below, we briefly examine the literature related to redistricting methods and competition followed by a description of our data and methods. Next, we present our findings from the entire post-Wesberry era that show when redistricting moves away from the normal legislative process, elections become more competitive. Finally, we offer a brief discussion about the broader implications of our findings.
Literature Review
Carson and Crespin (2004) pose the following question: Does the method employed for redrawing congressional boundaries, whether legislative-, commission-, or court-drawn, influence the level of competition in a given district? They argue that as redistricting is pulled outside of the normal legislative process, the goals and institutional constraints vary, which leads to different levels of competitiveness. They estimate a probit model using a binary measure of competition as the dependent variable, as well as utilize a number of controls including indicator variables for southern states, open-seat races, and the presence of a quality challenger, and measures of district partisanship and campaign expenditures. After analyzing data from the 1992 and 2002 House elections, these authors conclude that legislative-drawn districts experience lower levels of competition compared with other line-drawing methods.
Since Carson and Crespin (2004), questions remain concerning redistricting methods and their impact on competition and politics in general. Much research has followed in this vein. For instance, Grainger (2010, 545) finds evidence that “legislatively drawn districts have been, on average, less competitive than panel-drawn districts” in California since the 1960s. Also, Lindgren and Southwell (2013, 13) find that after controlling for partisan and contextual factors, “independent commissions resulted in more competitive districts, as measured by margin of victory and challenger win rate” in elections from 2002 to 2010.
Other works have taken a more nuanced approach to evaluating redistricting methods. Murphy and Yoshinaka (2009) posit a dynamic redistricting thesis and find that redistricting methods can foster competitiveness in some districts while protecting other members from potential competition. Karch, McConnaughy, and Theriault (2007) take a more detailed look at evaluating redistricting commissions. In particular, these commissions can take two different paths—“partisan” where the process is dominated by legislative insiders or “good government,” where outside interest groups exercise more influence. Though these are not necessarily mutually exclusive, the method chosen bears different implications for the future of that district, with the former method being driven more so by concerns of the controlling party. McKenzie (2012, 799) investigates court decisions on redistricting plans between 1981 and 2007 and concludes that federal judges “are neither neutral arbiters nor crass partisans” but instead seem to make decisions under “constrained partisanship” when considering congressional boundaries.
Finally, other scholarship has tested the potential implications of redistricting. McCarty, Poole, and Rosenthal (2009) test the relationship between redistricting and polarization and argue that “neutral” plans do not reduce levels of polarization any more than partisan-based plans would. However, they do attribute the 2008 Republican seat share increase to successful partisan gerrymandering. Yoshinaka and Murphy (2011, 435) demonstrate that the implications of redistricting are not limited to partisan composition. They suggest that certain plans “are able to produce significant instability between elections, especially for opposing-party incumbents.”
It has become clear that redistricting continues to be an important element of American congressional politics. Therefore, in keeping with the progress of this literature, we will seek to revisit Carson and Crespin’s (2004) analysis to assess the impact of redistricting method on electoral competition. Unlike previous scholarship, which tended to examine one or two election cycles, we will include each round of redistricting since the U.S. Supreme Court ordered changes in the 1960s. Also, the type of redistricting method employed may have to be given more consideration when predicting election winners. Therefore, failure to account for this could result in an overestimation of the effect of incumbency or the presence of a quality challenger, at least in elections following the creation of a new congressional map.
Data and Method
We examine the influence of commission- and court-drawn plans using the data found in Carson and Crespin (2004) as well as data from the 1972, 1982, and 2012 congressional elections. To determine who was responsible for drawing congressional districts, we used the descriptions in the appropriate version of Congressional Quarterly’s Congressional District book for the first four cycles (1973, 1983, 1993, and 2003). For the fifth cycle in our analysis, data were obtained from credible sources such as a state department’s website or a state’s leading newspaper. Since there are many variations in the type of commissions, we used an inclusive definition of the term. 1
Similar to Carson and Crespin, we employ an indicator variable coded 1 for competitive races and 0 otherwise as our dependent variable and expect to find more competition in plans drawn outside of the normal legislative process. A competitive race is classified as such if the winning candidate received less than 60% of the two-party vote in that election. 2
Table 1 depicts the relationship between the dependent variable and redistricting method for each congressional election following redistricting. From this, we see that legislative-drawn districts have a lower percentage of competitive races than both court-drawn and commission-drawn districts—generally conforming to our expectations. To further test our hypotheses, we estimate a probit model with the same predictors as Carson and Crespin (2004).
Number of Competitive House Races by Redistricting Method.
Our independent variables of interest are indicator variables for court-drawn and commission-drawn districts, using legislative-drawn districts as the reference group. A positive and significant coefficient will provide evidence in support of our hypothesis. Next, we include yearly fixed effects to control for any year-to-year differences that may emerge in the data. Furthermore, we employ the same controls as Carson and Crespin (2004). These include indicators for southern states, 3 open seats, and quality challengers, as well as a measure for the amount of money spent in a given race and the partisanship of a given district. Controlling for southern states allows us to avoid any confounding results that may be derived from the unique political and cultural environment of the area present during the time frame of this analysis. Also, due to the hegemony enjoyed by one party in these states, we expect southern states to hold less competitive congressional races, all else equal. The same logic applies to the district partisanship measure. This variable is measured as the absolute value of the percentage of the two-party vote received by the Republican presidential candidate in the entire nation minus the percentage of the two-party vote received by the same candidate in a given district. This measure provides some insight into how favorable a district is to one party or the other. Therefore, as this variable increases, we expect races in the district to be less competitive on average. We also include two indicator variables for whether or not the district was hosting an open-seat election and whether or not a quality challenger was running. Finally, we include the same spending measure as Carson and Crespin (2004). This is the logged total amount of money spent by both the Republican and Democratic candidates. As spending increases, we expect competitiveness to increase as well. A full description of the data as well as information onthe method employed by each state can befound in tables A1, A2, and A3 in the Appendix.
Results
Table 2 presents the results from our replication and extension. Here, we are able to replicate Carson and Crespin (2004) exactly. Also, all variables are significant and in the expected direction in the expanded model. Relative to legislative-drawn districts, court-drawn and commission-drawn districts experience more competition on average, all else equal. Figure 1 shows the probability of a district being competitive for each type of plan while holding each continuous variable at its mean and each indicator at its median. Moving from a legislative-drawn district to one that was redrawn by the courts results in an increase of 8.18 percentage points in the predicted competitiveness. The shift from legislative- to commission-drawn increases the probability of being competitive by 9.67 percentage points. There is no significant difference, however, between court and commission plans. 4
Probit Estimates for Competition in U.S. House Elections.
Note. Standard errors in parentheses.
Predictive margins with 95% confidence intervals.
Returning to Table 2, we see that competition appears to be decreasing over time, relative to elections held in 1972. This could possibly be due to a number of explanations, including an increased polarized environment. For instance, as the two major parties become more homogeneous, it is possible that we as the electorate are witnessing a change from competitive elections to more competitive primaries. This is commonly seen in conservative districts in the Deep South, where the winner of the Republican primary is essentially guaranteed to win the general election. Although this was once true for the Democratic Party, our entire analysis occurs after the passing of the 1965 Voting Rights Act. This caused a split in the Democratic Party that, in turn, created more competition at the polls in the earlier years of our analysis. Over time though, Southern Democrats either migrated to the Republican Party or began to look more like mainstream Democrats.
We also see that increased spending, an open seat, and the presence of a quality challenger all increase competition. Meanwhile, there tend to be fewer competitive races in southern states than non-southern states. Similarly, more partisan districts experience fewer competitive races than their less partisan counterparts. These results provide an interesting insight into the current state of congressional elections and redistricting plans. As time progresses and overall levels of competitiveness decline, the need for non-legislative-drawn districts increases if we wish to see greater levels of competition.
We also see a decrease in the standard errors of each estimate. This is likely due to the addition of three new election cycles and the resulting increase in sample size. However, we also see an increase in the effect each of these variables has on the overall level of competition. This is an interesting result that could indicate more volatility in the competitiveness of congressional elections. This may also reflect strategic behavior on the part of office-seekers. For instance, quality challengers may only enter a race if he or she knows the opponent is not a very well-funded incumbent. Another possible explanation is that the introduction of three additional election cycles has simply allowed the relationship to become more pronounced. Future work will hopefully address this issue more systematically.
To address the original question of “does the method used to redraw congressional lines affect partisan competition in the following elections,” the short answer is a simple yes. By extending the 2004 analysis to include five redistricting cycles while employing the same model with the same covariates, we find continued support for the hypothesis that “more competitive elections occur when courts and commissions are directly involved in the redistricting process, as opposed to when redistricting is handled only in the state legislative process” (p. 455). Based on their analysis, Carson and Crespin (2004) conclude, “[a]ssuming that one of the goals of court- and commission-drawn plans is to produce more competitive elections, then our findings suggest that this objective is being met” (p. 462). The results presented here on the expanded dataset provide no evidence to the contrary.
In addition to our replication and extension, we explore how general ideological trends toward polarization in state legislatures as well as the introduction of computer-based redistricting influenced competition. Both of these took off following the 1980 redistricting cycle. 5 Therefore, we separate the previous analysis into two models, one limited to elections held in 1972 and 1982 with the other containing election from 1992, 2002, and 2012, comparing the effects of different redistricting methods on the number of competitive House races. Table 3 presents the results from our estimates before and after the implementation of computer technology for the purpose of redrawing congressional lines.
Probit Estimates for Competition Pre- and Post-1982.
Note. Standard errors in parentheses.Computer software was available for the round of redistricting following the 1980 census but was rare and expensive (Altman, Karin, and McDonald 2005).
Here, we see that before the introduction of computer-based technology, court- and commission-based redistricting plans were not significant predictors of electoral competition. However, by 1992, non-legislative-drawn plans, on average, resulted in higher levels of competition than elections held in legislative-drawn districts. The introduction of this technology allowed line-drawers to apply a much more detailed approach to constructing new districts, which appears to have influenced the overall levels of competition. This relationship is also clearly depicted in Figure 2.
Predictive margins with 95% confidence intervals.
In the first panel, we see that the confidence bands overlap for all three methods of redistricting. This indicates that there is not a significant relationship between which line-drawing strategy was employed and the level of competition experienced in the 1972 and 1982 elections. However, in the second panel, we see that legislative-drawn districts witness significantly lower levels of competition than both court- and commission-based plans. This provides evidence in support of our hypothesis that technological advances coupled with increased polarization have resulted in legislatures drawing congressional districts that are significantly more partisan and therefore less competitive.
Conclusion
In the 2012 election, only 27 incumbents, or 6.2% of the members of the U.S. House, lost in the general election. The most recent election was actually much more competitive compared with the one that followed the last round of congressional redistricting. In 2002, when only eight incumbents were defeated, politicians and pundits alike bemoaned the lack of competition. Hirsch (2003) called the 2002 election the “most incumbent friendly in modern American history.” Meanwhile, Earl Blumenauer (D-OR) and now former representative Jim Leach (R-IA) wrote an article that appeared in the New York Times appealing for a shift from the current practice of allowing elected officials to draw new district boundaries toward a system where commissions are tasked with changing congressional districts (Blumenauer and Leach 2003). They argue that “commissions offer the best hope for taking partisanship out of the redistricting process.” These sentiments are echoed in recent comments made by Representative Reid Ribble (R-WI) including, “We’re at a place now in this country where voters are not picking their representatives anymore. Representatives, through the gerrymandering process and redistricting, are picking their voters.” (“Some Lawmakers Want to Put Politics Aside” 2013)
If states wish to increase competition in their congressional districts, utilizing commissions (or courts) when redistricting is an option that should be considered according to the analysis provided here. By testing for the levels of electoral competitiveness over five redistricting cycles, we gain empirical leverage over the question of interest. Our results provide further evidence to the conclusion reached in Carson and Crespin (2004): that on average, elections in court-drawn and commission-drawn districts are more likely to be competitive compared with their legislative-drawn counterparts. While these results may not be entirely counterintuitive, they are encouraging in that they suggest that commission- and court-drawn plans can lead to more competitive elections, as they were designed to do.
At the same time, our analysis has implications beyond the underlying competitiveness of House races. This scholarship also has implications for democracy in America. Knowing that state legislators are able to control the level of competition in a district, should the power to redraw congressional lines be removed from their hands? If so, would it make sense to utilize commissions to reduce the amount of partisan influence? As outlined earlier, much work has been devoted to understanding the redistricting process and its influence on other areas of political science, but there is still a need for more research on this fundamentally important issue in the context of democratic representation. If states are able to alter the balance of power in the House through their redistricting plans, then how they draw the seats can have far reaching policy consequences well into the future. Our findings also have notable implications for scholars studying the incidence of careerism in the House as well as the strategic nature of candidate entry and exit decisions. In addition, the type of redistricting method employed may have to be given more consideration when predicting election winners. Therefore, failure to account for this could result in an overestimation of the effect of incumbency or the presence of a quality challenger, at least in elections following the creation of a new congressional map. Only future research will be able to provide us with answers to these critical questions.
Footnotes
Appendix
Redistricting Plans by Year.
| 1972 | 1982 | 1992 | 2002 | 2012 | |
|---|---|---|---|---|---|
| Legislative | 34 | 29 | 27 | 26 | 33 |
| Court | 6 | 10 | 8 | 6 | 3 |
| Commission | 1 | 5 | 8 | 11 | 7 |
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.