Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
American Politics Research http://apr.sagepub.com/ The Empirical Implications of Electoral College Reform Darshan J. Goux and David A. Hopkins American Politics Research 2008 36: 857 DOI: 10.1177/1532673X08324213 The online version of this article can be found at: http://apr.sagepub.com/content/36/6/857 Published by: http://www.sagepublications.com Additional services and information for American Politics Research can be found at: Email Alerts: http://apr.sagepub.com/cgi/alerts Subscriptions: http://apr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://apr.sagepub.com/content/36/6/857.refs.html >> Version of Record - Nov 5, 2008 What is This? Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 The Empirical Implications of Electoral College Reform American Politics Research Volume 36 Number 6 November 2008 857-879 © 2008 Sage Publications 10.1177/1532673X08324213 http://apr.sagepub.com hosted at http://online.sagepub.com Darshan J. Goux David A. Hopkins University of California, Berkeley The events of 2000 inspired renewed debate in America over the manner in which presidents are chosen, with many critics of the existing system favoring the alternative of direct popular election. These advocates of reform often argue that the strategic environment shaped by the electoral college distorts the democratic process by encouraging candidates to focus on a few politically competitive states while virtually ignoring the majority of the nation. Using empirical evidence, we evaluate the claim that the electoral college is antidemocratic due to its effect on campaign strategy. Furthermore, we consider the possible alternative strategies adopted by candidates under a hypothetical national popular presidential election. We conclude that many of the consequences of electoral college reform, unacknowledged by its proponents, would not necessarily make presidential elections substantially more democratic, and we call for additional empirical research on a topic that has heretofore been dominated by theoretical debate and speculation. Keywords: electoral college; political campaigns; elections; presidency; political parties; mobilization; battleground states A ugust 4, 2004, was just another day on the campaign trail for President George W. Bush and his Democratic challenger, Senator John F. Kerry of Massachusetts. Bush delivered a stump speech in a riverfront park emphasizing his administration’s economic policies and record on fighting terrorism, whereas Kerry, hosting a public forum in a convention center, criticized the president for a loss of manufacturing jobs and pledged to reduce the American military presence in Iraq. The two candidates’ campaign events that day appeared remarkable only for their proximity—held within a half mile of each other in the medium-sized midwestern city of Davenport, Iowa (Balz & Goldstein, 2004). In a nation of nearly 300 Authors’ Note: Please address correspondence to David A. Hopkins, Institute of Governmental Studies, 109 Moses Hall #2370, Berkeley, CA 94720-2370; e-mail: [email protected]. 857 Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 858 American Politics Research million people spread over 3.5 million square miles, Bush and Kerry found themselves competing for votes in the same neighborhood. Although campaign aides insisted that the candidates did not intend their visits to occur simultaneously, it was hardly a pure coincidence that both men chose to spend the day in Davenport. Capturing the White House requires winning a majority of electoral votes, which the states allocate in a winner-take-all fashion to the candidate finishing first in the statewide popular vote.1 The American system of presidential selection, therefore, encourages candidates to concentrate their campaign resources in the minority of states considered winnable by either party (such as Iowa in 2004). Residents of these swing or battleground states receive a great deal of attention—candidate appearances, television advertisements, and mobilization efforts—from both sides during the months before the election, whereas the rest of the nation, deemed noncompetitive, experiences relatively little direct campaign activity. This characteristic of the electoral college has gained increasing notice from journalists and scholars alike in recent elections.2 To critics, already dissatisfied with the electoral college’s ability to deny the presidency to a candidate who finishes first in the overall national popular vote, the battleground state phenomenon further reinforces the institution’s antidemocratic nature. They argue that the electoral college, when combined with the winner-take-all allocation of state electors, gives disproportionate weight to the votes cast within politically competitive states. Because candidates focus their campaign resources on courting the residents of battleground states, those lucky voters reap benefits unavailable to the less fortunate citizens of states deemed safe and therefore ignored. That Davenport, Iowa, should be judged sufficiently important to the outcome of the election to warrant public appearances by both major presidential candidates on the same day, whereas the entire states of New York, Texas, and California receive no significant candidate attention during the entire general election campaign, strikes these observers as fundamentally unfair. This newly popular critique of the electoral college deserves serious consideration. Many previous criticisms of the institution’s antidemocratic attributes focused on scenarios that, although highly consequential, have occurred rarely (the election of the popular-vote loser, the election of the president by the House of Representatives) or never (the wholesale defection of electors to a rival candidate) in American history. In contrast, the massive effect of the electoral college on campaign strategy and candidate behavior is not only readily apparent and undoubtedly significant but dependably arises in every election. Even incumbent presidents tend to Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 859 govern with the electoral college in mind during their first term in office (Doherty, 2006). If the battleground state phenomenon truly represents a significant distortion of the electoral process, it bolsters the case for fundamental reform of the American system of presidential selection. We seek to evaluate several of the key claims advanced by critics of the electoral college who cite its (in their view) detrimental effects on campaign strategy as justification for reform or abolition of the institution. For this reason, we intend not only to determine whether the cited shortcomings of the existing electoral college are substantiated by available evidence but also whether they would be ameliorated by the introduction of direct popular election. This second goal admittedly requires some informed speculation, but we believe reasonable assumptions can be made about the nature of candidate behavior under this alternative electoral system. Perhaps unsurprisingly, our findings are somewhat mixed. On a few specific points, empirical evidence suggests that the battleground state phenomenon does work to the benefit of some classes of voters at the presumed expense of others. More broadly, however, we conclude that the strategic targeting of voters in specific geographic areas by presidential campaigns would continue even if the electoral college were abolished. Observers who interpret the degree of campaign attention received by a voter as an indicator of that voter’s electoral importance are likely to remain unsatisfied with the process of presidential selection even under a system of direct popular election. Strategy-Based Critiques of the Electoral College The electoral college has long exerted a powerful influence on the calculations of political actors. Yet much of the previous academic debate has not focused on candidate strategy. Prominent works have centered on the circumstances surrounding the adoption of the electoral college at the Constitutional Convention (Dahl, 2003), on theoretical concerns such as the proper role of “the federal principle” in choosing presidents (Best, 2004), on states’ relative voting power in a presidential election (Longley & Dana, 1992), and on a long-running but (to date) inconclusive series of analyses attempting to measure the direction and degree of partisan bias in the electoral college (e.g., Berthoud, 1997; Destler, 1996; Garand & Parent, 1991; Grofman, Brunell, & Campagna, 1997). Objections to the electoral college on the grounds that it discriminates against voters who do not reside in battleground states seem to have Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 860 American Politics Research become more frequent in recent years, perhaps due to the popular view of contemporary America as divided into staunchly red and blue bastions. Such polarization, to the extent that it has actually occurred, relegates most of the nation to the sidelines in presidential elections, leaving little competitive territory for the parties to fight over. Rakove (2004, p. 23) argued, “The logic of the electoral system and the sophistication of modern polling methods enable campaigns to concentrate resources and attention on the relatively small number of . . . ‘battleground states.’ Far from diffusing political activity across the extended republic, electoral logic allows it to be narrowly targeted to key communities and particular interests within them.” For these reasons, some advocates of reform argue that the electoral college consistently deprives most Americans of a meaningful voice in the selection of their president. As Peirce (2004, p. xi) put it, “Is it too much to ask: Couldn’t we all have a vote that counts?”3 According to this view, votes count only if they are cast within states contested by both major parties. Bennett (2006, p. 163) explained, “It is obvious that a voter in a state that is closely divided among presidential candidates effectively casts a weightier vote than does one in a similarly populous but politically lopsided state.” It is certainly obvious that residents of battleground states are the primary targets of candidate persuasion efforts. Whether this attention results in an effectively weightier vote is somewhat less clear, requiring some elaboration (and empirical confirmation) of the benefits exclusively enjoyed by these voters. We turn to this question next. Moreover, we caution against the assumption that abolition of the electoral college would give all Americans a vote that counts under this definition. There is good reason to believe that campaigns will continue to target particular geographic concentrations of voters and not others even under a nominally national popular election; we return to this argument below. What’s Distinctive About Battleground State Voters? Claims that the battleground state phenomenon renders presidential elections undemocratic or unfair rest on the assumption that residents of battleground states differ from their counterparts elsewhere in some politically relevant way. If the battleground state population were a fully representative sample of the national population, and if battleground state voters did not derive some benefit—instrumental, psychic, or otherwise—from their status as campaign targets, then the electoral college’s effect of restricting campaign activity to a subset of states would not seem to constitute a significant flaw. We seek to determine whether and to what extent the residents of battleground Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 861 Table 1 Targeted States in 2004 State Candidate Visits States in Which Both Campaigns Purchased Advertising Ohio Iowa Pennsylvania Wisconsin Florida Minnesota Michigan New Mexico West Virginia Colorado New Hampshire Nevada Oregon Maine Washington Other States With at Least One Candidate Visit Missouri North Carolina New Jersey Arizona Arkansas Hawaii Maryland South Carolina Texas Wyoming 45 31 30 28 23 19 17 12 11 10 10 7 5 4 0 7 5 4 3 1 1 1 1 1 1 Note: Candidate visits refers to visits by any of the four presidential and vice presidential nominees during the period September 3 to November 1, 2004, in which the candidate made at least one public appearance (state visits for the sole purpose of appearing on television programs or attending fundraising events are excluded). Source: Candidate visit data from Democracy in Action, George Washington University, http://www.gwu.edu/~action/2004/chrnfall.html (last accessed July 25, 2008). Advertising data from the Wisconsin Advertising Project, University of Wisconsin, http://www.polisci .wisc.edu/tvadvertising/ (last accessed July 25, 2008), and Daron R. Shaw (2006). states are politically distinctive, using the 2004 presidential election as a basis for our empirical inquiry. Table 1 presents measures of state-by-state activity by the Bush and Kerry campaigns during the 2 months in 2004 between the end of the Republican National Convention (the later of the two party conventions) and Election Day, Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 862 American Politics Research demonstrating that contemporary campaigns indeed focus their activities on a limited number of politically competitive states while paying less attention to the rest of the nation. Of the 277 total state visits during this period, more than 90% were concentrated in just 12 states, with the top 5 (Ohio, Iowa, Pennsylvania, Wisconsin, and Florida) accounting for well over half of all visits. Television spots followed a similar pattern; advertisements sponsored by both campaigns or national parties aired in only 15 states between the end of the conventions and the election. The statewide election results mirror the campaign strategists’ targets. Each of the 12 most competitive states, in which the margin of victory was 5 percentage points or less, had been campaign battlegrounds during the previous 2 months, bombarded with paid media messages and hosting at least five candidate visits.4 Of the 15 states in which both campaigns sponsored television advertising, only in Maine and West Virginia did one candidate receive at least 55% of the two-party vote. Is this subset of contested states unrepresentative of the nation as a whole, thus systematically benefiting particular social groups at the expense of others? Table 2 compares the demographic profile of the battleground states in 2004 to that of all other states and to the total national population. On most characteristics, residents of battleground states differ quite mildly from their counterparts in noncompetitive states. They are slightly more likely to have a high school diploma and a household income of less than US$35,000, but fewer have a college or graduate degree or make at least US$100,000. Battleground residents are slightly older than the general population and are marginally more likely to be military veterans or union members. Other than an overrepresentation of White Protestants in the battleground states, the distribution of religious affiliations also differs little between the two categories. The biggest differences between the battleground state population and the rest of the nation occur in the domain of race. Battleground states contain a lower proportion of racial minorities—Blacks, Latinos, Asians, and Native Americans. The relative Whiteness of electoral battlegrounds in 2004 is due to the lack of partisan competition across much of the Deep South, where African Americans constitute a significant portion of the population and in states with large Latino communities such as California, Texas, and New York. In contemporary politics, at least, the electoral college may work to underrepresent racial minorities within the electorate targeted by presidential candidates. This is a potentially damning indictment of the institution, especially considering the extensive history of electoral discrimination in the United States. Is the voting strength of minority groups unacceptably diluted by the electoral college? Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 863 Table 2 How Representative Are the Battleground States? 2004 Social Attribute Income less than US$35K Income between US$35K and US$100K Income more than US$100K High school diploma College degree (BA) Graduate degree 65 years or older Military veteran Union member Black Latino Asian Native American Mainline Protestant (White) Evangelical Protestant (White) Black Protestant Roman Catholic Mormon Jewish Muslim Unaffiliated Battleground States (%) Other States (%) Total United States (%) 41.8 41.3 41.5 47.5 10.7 73.1 20.4 7.2 18.5 14.0 8.1 7.9 7.1 2.0 0.7 21.1 24.5 6.9 24.1 0.9 1.8 0.4 15.8 45.7 13.0 68.7 21.6 8.0 16.0 12.0 7.1 12.9 12.7 4.5 0.8 16.0 23.0 10.8 24.6 1.4 2.3 0.6 13.6 45.7 12.3 70.1 21.3 7.7 16.7 12.6 7.4 11.4 11.0 3.7 0.8 17.5 23.4 9.6 24.5 1.3 2.2 0.5 14.3 Note: All figures represent the proportion of the adult (18 and older) population, except income figures, which represent the proportion of households. We consider the battleground states to be the 15 states exposed to campaign advertising on television in the fall of 2004 (see Table 1). Source: Data from 2000 U.S. Census, U.S. Bureau of the Census, http://www.census.gov (last accessed July 25, 2008), except union membership data from U.S. Bureau of Labor Statistics, http://www.bls.gov/news.release/union2.t05.htm (last accessed July 25, 2008), and religious affiliation data compiled by John C. Green, University of Akron, available at http://www.beliefnet .com/politics/religiousaffiliation.html (last accessed July 25, 2008). Some critics believe so. One argues that “few Republican presidential candidates under the electoral college have been willing to compete aggressively for African Americans’ votes” (Edwards, 2004, p. 98), presumably because Black voters are underrepresented in battleground states. But it is Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 864 American Politics Research unclear whether a nationwide popular vote would lead to a significantly different outcome. Research relating to congressional redistricting has demonstrated that the voting records of Republican officeholders do not become more sensitive to the political agenda of most African Americans even in southern states where the Black population exceeds 30% (Cameron, Epstein, & O’Halloran, 1996; Lublin, 1997). If African Americans do not receive sufficient attention from both parties during electoral campaigns, it is likely because they are not considered swing voters, whether or not they reside in swing states. Latinos are even more disadvantaged by the campaigns’ focus on battleground states, as Table 2 indicates, yet the perception that they are comparatively open to persuasion led to a remarkably prominent initiative by the Republican Party to court their support in the last two presidential contests. To be sure, the underrepresentation of racial minorities in the states contested by both campaigns in 2004 presents a potentially serious challenge to defenders of the electoral college. The extent to which it results in electoral inequality in practice depends, we believe, on the value of any benefits that accrue exclusively to the inhabitants of battleground states. We explore that question further in the next section. The Elusive Benefits of Battleground State Residence Critics of the electoral college often assert or imply that residence in a contested state bestows a number of desirable electoral rewards on citizens. For example, Bennett (2006, p. 163) argued that “candidates for president lavish time—and promises—on politically competitive states to the virtual exclusion of those where the outcome is foreordained,” suggesting that the residents of electoral battlegrounds enjoy concrete benefits denied to their less fortunate counterparts elsewhere. Although candidates certainly spend the bulk of their time (and advertising dollars) in these states, few examples of such promises made by candidates to particular states are immediately apparent. Perhaps the support of candidates of both parties for the embargo against Cuba qualifies as a state-specific promise designed to win favor with the refugee population in South Florida, as does the perennial declared opposition to the storage of nuclear waste in Nevada’s Yucca Mountain. In general, though, presidential elections in the United States hardly seem dominated by such parochial matters. Even Edwards (2004, pp. 101-102) found little mention of state interests in the campaign rhetoric of presidential candidates. Evidence that electoral targeting influences policy decisions Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 865 Table 3 Turnout Rates in Battleground States, 2000 and 2004 Battleground states All other states United States total Turnout Rate 2000 (%) Turnout Rate 2004 (%) 57.3 52.0 54.2 66.7 56.8 59.8 Note: Figures represent percentage of voting-eligible population casting ballots for president. Battleground states in 2000 were Arkansas, Delaware, Florida, Illinois, Iowa, Kentucky, Louisiana, Maine, Michigan, Minnesota, Missouri, Nevada, New Hampshire, New Mexico, Ohio, Oregon, Pennsylvania, Tennessee, Washington, West Virginia, and Wisconsin. Source: Turnout data compiled by Michael P. McDonald, George Mason University, available at http://elections.gmu.edu/voter_turnout.htm (last accessed July 25, 2008). Battleground state data from Shaw (2006). by incumbent presidents is likewise limited (see McCarty, 2000; Mebane & Wawro, 2002 for exceptions). One clear consequence of the battleground state phenomenon is the targeted mobilization efforts in contested states funded by candidates, parties, and interest groups on both sides. Political scientists have long demonstrated a link between voter mobilization activities and rates of political participation, including electoral turnout (Gosnell, 1927; Green & Gerber, 2004; Rosenstone & Hansen, 1993). Presumably, campaign targeting leads to higher turnout in battleground states than in the rest of the nation. As Table 3 demonstrates, the voter turnout rate in the battleground states was significantly higher than in the rest of the nation in both 2000 and 2004. In addition, the difference in turnout between battleground and noncontested states increased from the first election to the second. These findings suggest that a greater proportion of potential voters are mobilized in battleground states, either by targeted registration and get-out-the-vote drives or simply by the increased attention and campaign activity within the state. However, this differential turnout can be explained only in part by campaign effects. Many states with historically high turnout rates, principally due to lenient voter registration regulations (Wolfinger & Rosenstone, 1980), also became electoral battlegrounds in 2000 and 2004. Minnesota (76.8% voter-eligible turnout rate), Wisconsin (76.8%), Maine (72.3%), and New Hampshire (70.3%) all allow same-day voter registration, whereas Oregon (69.9%) conducts its elections entirely by mail-in ballot. At the same time, states with low aggregate levels of education, such as much of the South, consistently lag behind the rest of the nation in voting turnout. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 866 American Politics Research The campaigns largely ignored these states in 2004, considering them electorally safe for Bush. McKee’s (2002) analysis based on the 2000 election concluded that differential voter mobilization accounted for about 3 percentage points out of the total 5 percentage point difference in turnout rates between battleground and noncompetitive states. Battleground state residents do appear, therefore, to benefit from the mobilization of voters by presidential campaigns, although campaign effects do not account for the entire difference in turnout rates between contested and safe states. Does the disproportionate attention directed at them by presidential candidates make inhabitants of contested states more efficacious? As noted above, many critics of the electoral college draw a contrast between endlessly courted battleground state voters and the less fortunate residents of noncompetitive states who, they argue, may well interpret the lack of campaign activity in their communities as a sign that they have no meaningful voice in the outcome of the election, that their votes do not really count, and that the political system may therefore not be as responsive to their concerns. If this claim is true, the electoral college provides important psychic benefits to the minority of battleground state residents at the expense of potential voters elsewhere in the nation. Table 4 presents public opinion data drawn from the 2004 National Election Study. Respondents are divided into two groups based on their home state; those living in one of the 15 states identified in Table 1 as targets of television ad campaigns during the fall of 2004 were classified as battleground state residents. As one might expect, residents of battleground states are much more likely than others to predict a close electoral outcome in their state (t = 7.1, p < .001). Even so, nearly half of the inhabitants of uncontested states also foresee a competitive race, suggesting that many respondents fail to receive the message that their state is being conceded to one party or the other. We wonder whether the disproportionate attention given to battleground state residents indeed causes citizens elsewhere in the nation to feel as if their votes have less weight. Unsurprisingly, residents of campaign battlegrounds report significantly more persuasion and mobilization efforts, as campaigns, parties, and interest groups reliably concentrate their get-out-the-vote efforts on politically competitive states. Yet this activity does not appear to bolster the efficacy of battleground state residents. Two National Election Study items measuring feelings of efficacy produce near-identical results among both groups of respondents: a question measuring agreement with the statement, “People like me don’t have a say in what government does” and an item asking Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 867 Table 4 Mobilization and Efficacy in Battleground States, 2004 Battleground (%) Nonbattleground (%) Do you think the presidential race will be close here in (state) or will [candidate named/one candidate] win by quite a bit? Will be close 69.0 46.9 Win by quite a bit 31.0 53.1 ta = 7.10*** As you know, the political parties try to talk to as many people as they can to get them to vote for their candidate. Did anyone from one of the political parties call you up or come around and talk to you about the campaign this year? Yes 63.6 36.3 No 36.4 63.7 ta = –8.7086*** Other than someone from the two major parties, did anyone (else) call you up or come around and talk to you about supporting specific candidates in this last election? Yes 25.2 14.2 No 74.8 85.8 ta = –4.47*** Some people say that no matter who people vote for, it won’t make any difference to what happens. Others say that who people vote for can make a difference to what happens. Where would you place yourself? (5-point scale recoded 0-1, where 0 = makes no difference, 1 = can make a difference) 0.74 0.76 (.01) (.01) ta = –1.34 “People like me don’t have any say about what the government does.” Do you agree strongly, agree somewhat, neither agree nor disagree, disagree somewhat, or disagree strongly with this statement? Agree strongly/somewhat 47.1 47.2 Disagree strongly/somewhat 52.9 52.8 ta = 0.03 How much do you feel that having elections makes the government pay attention to what the people think—a good deal, some, or not much? A good deal/somewhat 92.1 92.7 Not much 8.0 7.3 ta = 0.37 Note: N = 1,212. Source: American National Election Studies (2004). a. Cell entries represent a two sample t test with equal variances. Although several of the dependent variables reported are dichotomous, the large sample size means the binary distribution is approximated to the normal distribution. *p < .05. **p < .01. ***p < .001, two-tailed. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 868 American Politics Research whether elections make the government “pay attention to what the people think.” Similarly, residents of noncompetitive states are as likely as voters in targeted states to agree that who one votes for can make a difference to what happens in the nation. Our findings are consistent with Wolak (2006), who found no evidence that battleground state residents have higher levels of campaign interest or discuss politics at greater rates than other voters. Although targeted mobilization efforts in battleground states seem to produce higher turnout, the concentration of campaign efforts within a subset of the nation does not lead to different levels of efficacy in the two groups of states, either by raising interest and engagement within electoral battlegrounds or by depressing investment in the political system elsewhere in the nation. How much envy should most Americans feel toward the residents of battleground states who attract so much attention from candidates and interest groups every 4 years? Most people prefer not to be bothered by telephone calls or knocks at the door urging them to vote, become annoyed by traffic jams and other inconveniences caused by candidate appearances in their community, and harbor an intense dislike of political advertising (Memmott, 2004; Pisik, 2004). Voters forced to endure such determined courting every 4 years do not even feel significantly more efficacious than residents of safe states, despite the additional electoral power they supposedly enjoy. We conclude that the advantages that accrue to battleground state voters courtesy of the electoral college are largely limited to an elevated level of voter turnout as a result of targeted mobilization efforts by candidates, parties, and interest groups, with additional instrumental or psychic benefits appearing modest at best. How Would Electoral College Reform Affect Campaign Strategy? Political reforms, when instituted, sometimes fail to resolve the problems they were designed to address. In other cases, existing deficiencies are rectified, but new, even more damaging flaws are introduced. For these reasons, advocates of reform strengthen their case when they augment their criticisms of the current system with a convincing argument that their preferred alternative will, in fact, represent an improvement over the status quo. The electoral college is, to be sure, an imperfect institution, and many of the criticisms leveled at it are entirely appropriate. But more attention should be Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 869 paid to the likely empirical consequences of its reform or abolition. In particular, critiques of the electoral college who focus on the battleground state phenomenon, lamenting the tendency of presidential campaigns to court the residents of certain states while ignoring others, overlook the probability that candidates would continue to target their resources toward particular geographic subgroups of voters even under a system of direct popular election. We cannot predict for sure how potential changes to the current system of presidential selection would affect candidate behavior. But we may begin what should be an energetic discussion among scholars of American presidential elections about the probable real-world effects of electoral college reform—an exercise that seems noticeably lacking in previous studies. Perhaps individuals on either side of the normative debate will rethink their positions when confronted with empirical evidence or perhaps exposure will harden their existing beliefs. In either case, we consider the effort a valuable one. Under a system of direct popular election, candidates no longer have reason to acknowledge state boundaries. All votes are of precisely equal mathematical value no matter where they are cast. Yet the severe practical limitations placed on presidential campaigns would endure even after reform. Resources, principally time and money, would remain limited—in fact, might seem even more so given the vastly expanded field of play. Abolition of the electoral college would not bring an end to campaigns driven by strategic calculation. Candidates cannot realistically compete everywhere or engage every voter; inevitably, each campaign would decide, just as now, to direct resources toward some identifiable populations and not others. Modern campaigns are largely waged through the media, whether free (news organizations) or paid (advertising spots). In 2004, candidates, parties, and affiliated groups spent more than US$600 million on television advertising, concentrated almost entirely in battleground states (Edsall & Grimaldi, 2004). It is likely that media markets, not states, would be relevant geographic subdivisions under a system of direct election. Even today, campaign consultants think in terms of media markets when buying advertising time or seeking local news coverage, but the specific markets emphasized are determined by calculations of state-level competitiveness (Althaus, Nardulli, & Shaw, 2002; Franz, Rivlin, & Goldstein, 2006; Shaw, 2006). This is not a recent practice; strategists as far back as Kennedy’s 1960 presidential campaign used media market boundaries to organize within-state activities.5 Absent the constraints imposed by the electoral college, other factors will dictate which markets—and, by extension, which voters—receive the most attention from candidates. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 870 American Politics Research There are over 200 media markets in the United States.6 The cost of advertising varies widely from one to another, based not only on population size within the market but also on other demographic characteristics that affect the attractiveness of the audience to potential advertisers. In the fall of 2004, the most expensive media market for prime-time advertising was Los Angeles, which cost US$3,204 per ratings point (in other words, to reach 1% of the households in the market), whereas US$11 bought a prime-time ratings-point worth of advertising in Glendive, Montana, the nation’s smallest market (see Table 5). Even taking the vast population difference into account, Los Angeles remains a pricier market than Glendive because it contains higher proportions of young residents and viewers with high incomes—a more valuable target audience for most commercial advertisers. How would the campaigns choose which media markets to target? Evidence indicates that they would follow much the same strategy as they do today: attempt to maximize the number of persuadable, or swing, voters targeted per dollar of advertising cost (Arterton, 1993; Shaw, 2006). If the electoral college is abolished, candidates seeking the swing vote need not confine themselves to competitive states. In general, the more populous an area, the more swing voters it has; campaigns today employ polling and analysis of previous election returns to precisely estimate the number of persuadable voters in any given geographic location, targeting campaign resources accordingly. To provide a simulation of campaign strategy, we have developed our own swing measure. First, we calculate a normal vote for each media market in the United States by averaging the two-party vote in each of the five presidential elections from 1988 to 2004. We then compute the mean absolute deviation from that figure or the average amount by which the vote in a particular election differs from the 20-year average— the swing. Multiplying the swing percentage by the number of voters in the media market in 2004 gives us an estimate of the number of swing voters. Table 6 divides the nation’s media markets into three categories of roughly equal population, using the measurement of swing voters described above and data on market cost from a standard industry report. These findings demonstrate that advertisers pay a premium to reach the residents of large metropolitan areas. The top 10 markets contained 31.6% of the swing voters nationwide (and just 29.4% of all voters), and a prime-time ad campaign cost an aggregate US$14,859 per ratings point (or exposure to 1% of households) in these markets. For a substantially lesser amount of US$10,737 per ratings point, a campaign could reach 33.6% of the nation’s voters (and 35.8% of its swing voters) by sponsoring ads in each of the smallest 155 markets across the nation. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 871 Table 5 Media Markets by Cost (Fall 2004) and Normal Democratic Vote (1988-2004) 10 Most Expensive Media Markets Los Angeles, CA New York, NY-NJ-CT-PA San Francisco-Oakland-San Jose, CA Chicago, IL-IN Boston-Manchester, MA-NH-VT Washington, DC-VA-MD-WV-PA Philadelphia, PA-NJ-DE Dallas-Fort Worth, TX Atlanta, GA-AL-NC Houston, TX 10 Least Expensive Media Markets Butte-Bozeman, MT Eureka, CA Zanesville, OH Twin Falls, ID Ottumwa-Kirksville, IA-MO Casper-Riverton, WY Great Falls, MT North Platte, NE Helena, MT Glendive, MT Cost-per-Point (Prime Time) Normal Democratic Vote US$3,204.00 2,981.00 1,685.00 1,485.00 1,263.00 1,175.00 1,106.00 1,097.00 868.00 777.00 54.1 59.5 66.7 58.5 59.5 56.3 57.5 40.7 44.9 43.2 25.00 25.00 25.00 23.00 22.00 20.00 19.00 17.00 13.00 11.00 48.3 55.9 43.5 33.2 51.1 37.2 44.9 30.0 47.9 41.2 Note: The normal Democratic vote is the average Democratic percentage of the two-party vote in the five elections from 1988 to 2004. Source: SQAD Media Market Guide (2004, Summer), Tarrytown, NY: SQAD. Table 6 Advertising Strategy: The Swing Voter Scenario Media Markets (DMAs) Top 10 Next 40 Bottom 155 Percentage of Total Vote, 2004 Percentage of Swing Vote, 2004 Cost per Ratings Point Cost per Ratings Point per 10,000 Swing Voters 29.4 37.0 33.6 31.6 32.6 35.8 US$14,859 11,370 10,737 US$113.4 84.1 72.3 Note: DMAs = designated market areas. Source: Election data compiled by authors. Cost of media calculated from figures provided for the cost of prime-time advertising in the fall of 2004 in each media market in the SQAD Media Market Guide, Summer 2004, published by SQAD, Inc., Tarrytown, NY. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 872 American Politics Research Figure 1 Media Market by Voter Population, 2004 Figures 1 and 2 confirm the high degree of congruence between the most populous and most expensive per capita media markets. The dark-shaded markets in Figure 1 are the largest 10 markets in the nation, home to about a third of the American population. They correspond with the large population centers of Boston, New York, Philadelphia, Washington, Atlanta, Detroit, Chicago, Minneapolis, Los Angeles, and San Francisco. Mediumsized markets are concentrated in slightly smaller metropolitan areas: Cleveland, Tampa, St. Louis, Buffalo, Nashville, and comparable places. The smallest markets are concentrated in thinly populated areas: upper New England, much of the Deep South, and across the interior West. In Figure 2, media markets are once again divided into terciles, this time by the cost per capita of one prime-time ratings point (equal to reaching 1% of the market’s population). The geographic pattern is much the same, confirming that advertisers pay a premium to reach television viewers in large metropolitan areas, rendering the audience in smaller population centers more cost-effective for political campaigns to target. We should not therefore assume that campaigns will necessarily concentrate most of their advertising resources in each of the largest metropolitan areas under a system of direct popular election. More likely, campaigns Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 873 Figure 2 Per Capita Advertising Costs by Media Market, 2004 will make extremely complex calculations about where their precious budgets will have the biggest positive effect. Given its relatively low advertising rates and abundant supply of swing voters, Florida would likely remain a top campaign target in a postreform world. The major markets in California and Texas, in contrast, command very high advertising rates compared to the number of voters they reach, and one or both campaigns could decide that limited funds would be best deployed elsewhere. Abolition of the electoral college certainly does not guarantee that residents of any particular large metropolis will experience substantially more direct attention, at least in terms of paid media, than they do under the existing system. Other forms of campaign activity may follow somewhat different patterns. Candidate appearances might well focus on the largest media markets, although the relative novelty of a visit to a moderately sized market could stimulate more press and popular attention. Less common, however, would be the multiday bus and train trips through small cities and towns that have become a staple of recent campaigns; such excursions would probably not reach enough voters to be worth the time. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 874 American Politics Research What if both campaigns were to emphasize mobilization of the party base over appeals to swing voters? In this scenario, the candidates are not competing for the same population. Rather than expecting both campaigns to make similar calculations, and thus mirror each other strategically, we assume that each side would target the geographic areas in which it enjoys electoral dominance, with the hope that a higher turnout in those locations would increase its national margin in the popular vote. This potential state of affairs should not reassure those observers who believe that American politics has become unacceptably partisan and ideologically polarized in recent years. Democratic candidates would presumably work to maximize turnout in coastal California and the urban Northeast, whereas Republicans would mobilize across the Deep South and interior West. More voters might be exposed to one campaign message, but fewer to both, further deepening partisan geographic divisions. In general, the Democratic base vote is heavily concentrated in large cities and inner suburbs outside the South, whereas the Republican base is more evenly distributed across geographically dispersed areas: outer suburbs, smaller cities, and towns. The location of the Democratic base in the most populous metropolitan areas becomes a relative liability for the party in targeting these voters by paid media. Large, Democratic leaning media markets tend to be more expensive per voter than the smaller, pro-Republican markets. In Table 7, media markets are divided into three roughly equipopulous categories: the Democratic and Republican base markets, where the normal vote for the prevailing party’s presidential candidate averaged at least 55% during the 19882004 period and the remainder group of competitive markets. The cost-perpoint and cost-per-point-per-voter ratios are highest among the Democratic base and lowest in the predominantly Republican areas, reflecting disproportionate Democratic loyalty in large media markets with high per capita advertising rates. To the extent that television advertising is effective as a mobilization tool, Republicans could spend their money more effectively by targeting the smaller media markets where they enjoy greater political strength. A comparison of Figure 3 with Figures 1 and 2 reveals that the Democratic base is indeed concentrated in the populous, high-cost markets of the Northeast, Midwest, and Pacific Coast. Republicans can reach their comparably sized base with more cost-effectiveness by running advertising campaigns in the small, inexpensive media markets common across the South and interior West. Although each of these markets is much smaller than the average Democratic base market, they add up to a roughly equal population when combined and command much lower advertising rates. These scenarios strongly suggest that the unequal geographic distribution of campaign activity would not end with the abolition of the electoral college. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 875 Table 7 Advertising Strategy: The Base Vote Scenario Media Markets (DMAs) Democratic base Competitive Republican base Percentage Percentage Percentage of of Total of Democratic Republican Vote, 2004 Normal Vote Normal Vote 37.6 30.8 31.6 43.4 30.8 25.8 31.8 30.7 37.6 Cost per Ratings Point Cost per Ratings Point per 10,000 Voters US$14,951 11,186 10,829 US$3.3 3.0 2.8 Note: DMAs = designated market areas. Source: SQAD Media Market Guide. Figure 3 Media Market by Partisanship, 1988-2004 To be sure, the identity of the voters or location of the communities that are deemed most critical to electoral victory can be expected to change if the institution governing the process of selecting presidents is fundamentally reformed. But presidential campaigns would still struggle with the need to deploy limited resources within a vast and populous nation; lacking the ability to engage everywhere, they will be forced to choose their battles. In some ways, Democrats might reap an advantage from reform; in some ways, Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 876 American Politics Research Republicans would gain an edge; and other consequences might well arise that cannot be currently foreseen. It seems likely that the geographic scope of partisan conflict would broaden somewhat from its current concentration within one third of the nation but would still fall well short of a true national election. Such an expansion would presumably increase the cost of presidential elections and therefore provide an advantage to the candidate or party with the largest financial resources. Historically, Republicans have raised more funds than Democrats, although this tendency has started to change in recent years. Whichever candidate held a funding advantage in any particular election would benefit disproportionately from abolition of the electoral college. It is difficult to tell how closely these predictions match the visions of reform proponents. Most prefer to concentrate on the alleged deficiencies of the current system of presidential selection than to describe in detail how the popular vote alternative would lead to different, better campaigns in the future. Although a simple national popular election would, in theory, give all citizens an equally weighted vote, campaigns would be compelled by strategic imperatives to target voters in certain geographic areas more than others to maximize scarce resources. Those who currently view such behavior as an unacceptable consequence of the electoral college could well be equally dissatisfied with the postreform world. Conclusion Although the possibility of significant reform or abolition of the electoral college via constitutional amendment remains vanishingly small, the recent proliferation of state level legislation and ballot initiatives seeking to modify the allocation of electoral votes suggests that change may still be in the offing. By awarding their electors to the national popular winner, a few populous states could, in effect, replace the electoral college without the need for ratification by three quarters of the states as provided for in the Constitution. Whether or not such efforts succeed, debate over the electoral college will doubtless continue in the pages of academic journals, within the news media, and on the floor of legislatures across the nation. The newfound scholarly focus on the observable consequences of the electoral college’s institutional features, such as its profound effects on the strategic behavior of candidates, is a welcome development that promises to add a strong empirical component to the discussion. But reform advocates who justify their position by citing the practical—as opposed to the theoretical— Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 877 imperfections of the current system, such as its (supposedly damaging) effect on campaign strategy, must make their case for the practical superiority on that count of their favored alternative. We do find evidence in favor of several claims made by critics of the electoral college, although other assumptions do not appear to survive empirical scrutiny. On the whole, we conclude that the flaws caused by the electoral college’s effects on candidate behavior are somewhat overstated by its detractors, whereras the alternative process of a direct popular election might well prove less revolutionary than many assume if it were indeed enacted into law. Notes 1. Maine and Nebraska allocate two electoral votes to the candidate finishing first in the statewide vote and one electoral vote to the winner of each congressional district. In practice, neither state has ever split its vote in the electoral college. 2. See Goux (2006) for evidence indicating that the news media coverage of “battleground states” has increased substantially since the 1960s. 3. Emphasis in original. 4. These 12 states are in increasing order of margin between the candidates in the two-party vote: (1) Wisconsin, (2) Iowa, (3) New Mexico, (4) New Hampshire, (5) Ohio, (6) Pennsylvania, (7) Nevada, (8) Michigan, (9) Minnesota, (10) Oregon, (11) Colorado, and (12) Florida. 5. See, for example, memo from Guild, Bascom, and Bonfigli, August 29, 1960, Robert F. Kennedy Pre-Presidential Papers, Box 38, John F. Kennedy Presidential Library. For the 1988 and 1992 Bush campaign plans see report, “1992 Guide to Targeting: Minimum Vote Analysis Goals and 1988 Vote at the ADI Level,” April 28, 1992, Robert Teeter Collection, Box 85, George Bush Presidential Library. 6. Within the advertising industry, media markets are known as designated market areas, or DMAs, and are defined by Nielsen Media using counties (parishes in Louisiana) as the constituent units. We exclude Alaska from our analysis because it is not divided into counties. References Althaus, S. L., Nardulli, P. F., & Shaw, D. R. (2002). Candidate appearances in presidential elections. Political Communication, 19, 49-72. Arterton, F. C. (1993). Campaign ’92: Strategies and tactics of the candidates. In G. C. Pomper, F. C. Arterton, R. K. Baker, W. D. Burnham, K. A. Frankovic, M. R. Hershey, & W. C. McWilliams (Eds.), The Election of 1992: Reports and interpretations (pp. 74-109). Chatham, NJ: Chatham House. Balz, D., & Goldstein, A. (2004, August 5). Campaigns cross paths in Midwest. The Washington Post, p. A1. Bennett, R. W. (2006). Taming the electoral college. Stanford, CA: Stanford University Press. Berthoud, J. E. (1997). The electoral lock thesis: The weighting bias component. PS: Political Science and Politics, 30, 189-193. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 878 American Politics Research Best, J. A. (2004). Presidential selection: Complex problems and simple solutions. Political Science Quarterly, 119, 39-59. Cameron, C., Epstein, D., & O’Halloran, S. (1996). Do majority–minority districts maximize black substantive representation in congress? American Political Science Review, 90, 794-812. Dahl, R. A. (2003). How democratic is the American constitution? (2nd ed.). New Haven, CT: Yale University Press. Destler, I. M. (1996). The myth of the electoral lock. PS: Political Science and Politics, 29, 491-494. Doherty, B. J. (2006). The politics of the permanent campaign: Presidents, fundraising, and the electoral college. Unpublished doctoral dissertation, University of California, Berkeley. Edsall, T., & Grimaldi, J. V. (2004, December 30). On Nov. 2, GOP got more bang for its billion, analysis shows. The Washington Post, p. A1. Edwards, G. C., III. (2004). Why the electoral college is bad for America. New Haven, CT: Yale University Press. Franz, M. M., Rivlin, J. & Goldstein, K. (2006). Much more of the same: television advertising pre- and post-BCRA. In Malbin, M. (Ed.), The election after reform: Money, politics and the Bipartisan Campaign Reform Act. Lanham, MD: Rowman and Littlefield. Garand, J. C., & Parent, T. W. (1991). Representation, swing, and bias in U.S. presidential elections, 1872-1988. American Journal of Political Science, 35, 1011-1031. Gosnell, H. F. (1927). Getting out the vote: An experiment in the stimulation of voting. Chicago: University of Chicago Press. Goux, D. J. (2006). A new battleground? Media perceptions and political reality in presidential elections, 1960-2004. Paper presented at the annual meetings of the American Political Science Association, Philadelphia, PA. Green, D. P., & Gerber A. S. (2004). Get out the vote! How to increase voter turnout. Washington, DC: Brookings Institution. Grofman, B., Brunell, T. L., & Campagna, J. (1997). Distinguishing the difference between swing ratio and bias: The U.S. electoral college. Electoral Studies, 16, 471-487. Longley, L. D., & Dana, J. D. (1992). The biases of the electoral college in the 1990s. Polity, 25, 123-145. Lublin, D. (1997). The paradox of representation: Racial gerrymandering and minority interests in Congress. Princeton, NJ: Princeton University Press. McCarty, N. M. (2000). Presidential pork: Executive veto power and distributive politics. American Political Science Review, 94, 117-129. McKee, S. C. (2002). Was turnout significantly higher in the battleground states in the 2000 presidential election? Paper presented at the annual meetings of the American Political Science Association, Boston, MA. Mebane, W. R., & Wawro, G. J. (1993). Presidential pork barrel politics. Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL. Memmott, M. (2004, September 27). Voters in Ohio give political ads a thumbs down. USA Today, p. 15A. Peirce, N. R. (2004). Foreword. In G. C. Edwards III, Why the electoral college is bad for America. New Haven, CT: Yale University Press. Pisik, B. (2004, November 2). Florida voters say “enough” to ads, calls, mailings. The Washington Times, p. A06. Rakove, J. N. (2004). Presidential selection: Electoral fallacies. Political Science Quarterly, 119, 19-37. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012 Goux, Hopkins / Empirical Implications of Electoral College Reform 879 Rosenstone, S. J., & Hansen, J. M. (1993). Mobilization, participation, and democracy in America. New York: Macmillan. Shaw, D. R. (2006). The race to 270: The electoral college and the campaign strategies of 2000 and 2004. Chicago: University of Chicago Press. Wolak, J. (2006). The consequences of presidential battleground states for citizen engagement. Political Research Quarterly, 59, 353-361. Wolfinger, R. E., & Rosenstone, S. J. (1980). Who votes? New Haven, CT.: Yale University Press. Darshan J. Goux is a PhD candidate in the Department of Political Science at the University of California, Berkeley. Her research focuses on electoral politics and the relationships between elite decision-making, voter behavior, media norms, and political institutions. David A. Hopkins is a PhD candidate in the Department of Political Science at the University of California, Berkeley. He specializes in the political geography of the United States and its effects on candidates, elections, and political institutions. Downloaded from apr.sagepub.com at UNIV OF SOUTHERN CALIFORNIA on January 24, 2012