This paper explores the racial origins and legacy of the Electoral College through historical and quantitative analyses. At the Constitutional Convention, the Electoral College served the interests of Southern slaveowners by perpetuating the advantage of the Three-Fifths Compromise. Following Reconstruction, Southern states that disenfranchised African Americans received an even larger voice in the Electoral College. Republicans tried to counter these moves by admitting unusually small states into the union. The ironic consequence of this non-racial decision is that, today, these states are heavily White and have citizens with higher levels of racial resentment. A MM-regression analysis of every election from 2000 on indicates the Electoral College has consistently awarded more votes per capita to states with Whiter populations and more racially conservative attitudes. The racially-disparate power exerted by these states makes it more difficult to enact racially-egalitarian policies. This paper adds a new perspective to the normative debate over the Electoral College, which typically focuses on democratic fairness and federalism concerns.
Donald Trump’s divisive campaign attracted voters motivated by racial resentment (Klinkner 2016) and culminated in an Electoral College inversion – a result in which a candidate with fewer popular votes still won the White House. The 2016 election also reignited the normative debate over the Electoral College. Scholars and pundits typically focus on whether this institution’s democratic unfairness outweighs federalism interests. This discussion is incomplete because it ignores the racial inegalitarianism embedded within a seemingly non-racial institution.
This article traces the origins and measures the effects of racial inequality within the operation of the Electoral College. The relationship between race and the apportionment of representation was a natural consequence of a winding sectional struggle for power. Sectionalism began as a competition between slave states and free states, but the Civil War did not end this broader conflict. Instead, the war transformed the sectional battle into one between the Republican Party (in the North, West, and occasionally the South) and the Democratic Party (in the South and occasionally in the West). Some of the sectional battles were fought on explicitly racial terms, while other conflicts were non-racial in their origins but created racially unequal allocations of power.
Delegates to the Constitutional Convention created the Electoral College as a natural complement to the Three-Fifths Compromise that rewarded southern states with disproportionate representation. Ironically, the outcome of the Civil War exacerbated the degree to which the Electoral College promoted racial inequality. While the Republican Party tried on various occasions to enforce new constitutional protections for African-Americans, but divided government and factionalization within their own party made it difficult to stop the rise of Jim Crow. The Fourteenth Amendment required the census to count formerly enslaved people as five-fifths of a person, but disenfranchisement techniques ensured Southern states received votes in the Electoral College based on their Black populations without having to let African Americans influence these or other elections. The Republican Party tried to counter the influence of the Solid South during periods of unified government by strategically admitting new states to the Union that were unusually small, heavily White, and reliably Republican. The composition of these Great Plains and Mountain West states was not the product of racial decisionmaking, but the result of the Homestead Act being more successful in encouraging Whites to settle the West.
Slavery and Jim Crow are gone, but the anachronistic Electoral College remains, and so do issues of racially disparate allocations of electoral votes, driven in part by the legacy of the Republican Party’s statehood admission strategy. A MM-regression model indicates states with Whiter populations have substantially more electoral votes per capita, even when controlling for ruralness. The results suggest that in 2016, a one-standard-deviation change in the White population share is associated with a 6.9% increase in the median state’s votes in the Electoral College. That finding is the equivalent of providing a state with eight electoral votes with another 0.553 votes simply for being somewhat Whiter than its peers. The analysis also indicates state citizenries with more racially conservative attitudes receive significantly more power under Electoral College arrangements.
As I document in the concluding section, the racial inegalitarianism of the Electoral College has done much more than offend America’s commitment to democracy and equality. It has altered the outcomes of multiple presidential elections and forced presidents to pay more attention to Southern interests than they otherwise deserved. In turn, this racially-distorted power structure has altered the direction of many civil rights issues, from giving diplomatic recognition to Haiti following its revolution to excluding African Americans from meaningful access to New Deal programs. Today, it is more difficult to make federal policy more racially egalitarian because states with more racially resentful citizenries exert disproportionate power in the Electoral College, as well as the Senate.
Alexander Hamilton spoke for many of the Founders when he lauded the Constitution’s method of executive selection. In Federalist 68, he (1788) wrote, “[I]f the manner of it be not perfect, it is at least excellent.” That enthusiasm quickly faded after the election of 1800 prompted the Twelfth Amendment, and the Electoral College has remained controversial ever since. Since the Founding, over 700 constitutional amendments had been introduced to reform or abolish the Electoral College (Slonim 1986, p. 35). The most frequent justification for an Electoral College amendment is to make the Constitution more democratic (e.g. Longley and Peirce 1996; Levinson 2006; Edwards 2011).
Some political theorists defend the Electoral College as an institution that promotes other constitutional virtues without unnecessarily undermining principles of democracy. Several scholars justify the Electoral College as a compromise designed to promote federalism (Hardaway 1994; Adkins and Kirwan 2002; Boylan 2008; Ross 2012). Best (1996, p. 17–18) casts the Electoral College as a republican institution that mediates the excesses of majoritarianism, an argument she bolstered with an unfortunate example: “Time was, when an arithmetical majority in this country supported slavery or at least racial discrimination.” Diamond (1992, p. 45) feared Electoral College reform would undermine constitutional veneration. He argued, “[A] long-standing constitutional arrangement secures, by its very age, that habitual popular acceptance which is an indispensable ingredient in constitutional legitimacy…”
Institutional rules are rarely perfectly fair, so if the Electoral College violated the principle of “one person, one vote” in a minor and random fashion, these calls to tolerate this institution might be compelling. However, the US Constitution singles out for particular disapprobation rules that undermine racial equality, and the justifications offered by the framers of the Reconstruction Amendments provide a powerful retort to defenders of federalism, republicanism, or constitutional veneration. Sections 2 and 5 of the Fourteenth Amendment and Section 2 of the Fifteenth Amendment call for federal oversight of any state that may be tempted to use federalism as a pretext to deny African Americans political equality. The Constitution’s commitment to republicanism also changed during this historical moment. Reconstruction Republicans justified excluding Southerners from retaking their seats in Congress until they ratified the Fourteenth Amendment because the South’s failure to protect the rights of African Americans undermined the Republican Guarantee Clause of Article IV (Amar 2006, p. 366–376). Finally, the Thirteenth, Fourteenth, and Fifteenth Amendments suggest that Americans were no longer willing to venerate those aspects of the original Constitution that undermined racial equality.
The last serious push for reform to the Electoral College occurred just after the passage of the Voting Rights Act. In 1969, the House of Representatives passed a constitutional amendment creating a direct election for president. The amendment enjoyed broad bipartisan support, including from President Nixon. One reason the amendment failed to receive enough support in the Senate was opposition from liberals who were concerned that a popular vote would dilute the political power of African Americans.
The argument assumed African Americans tend to live in large, politically competitive states, which would force candidates from both parties to appeal to their interests to build an Electoral College majority (Banzhaf 1968; Bickel 1971). Several scholars have pointed out these premises are not true – the heaviest concentration of Black voters are actually in the South, large states are not usually swing states, and African-Americans vote overwhelmingly for Democratic candidates (Edwards 2011, p. 140–141; see also Longley and Peirce 1996; Lineberry et al. 2002, p. 146–153). As Edwards (2011, p. 142) concludes, “The electoral college thus discourages attention to the interests of African Americans because they are unlikely to shift the outcome in a state as a whole.”
In the hopes of reshaping the debate over the Electoral College, the next three sections provide historical and quantitative analyses of the creation, alteration, and modern operation of this institution. In the concluding portion, I evaluate how the racial dynamics of the Electoral College have altered election and policy outcomes, often at the cost of civil rights initiatives.
Many political science accounts of the Founding fail to capture the racial and regional implications of the debate over executive selection. John Roche’s constitutional history (1961) dismissed the Electoral College as a hodge-podge solution on which the Founders did not deliberate, even though the convention debates included a discussion of slavery as a salient consideration on several occasions. This account also ignores the fact that American had prior experience with indirect elections: the Maryland Constitution of 1776 used an Electoral College-style system to choose members of its state senate. While Maryland leaders justified this institutional arrangement as a defense against demagoguery, it also had a racial legacy. When Maryland reformers sought a directly-elected senate in the 1830s, some aristocrats feared this change would make it easier to abolish slavery. In exchange, the state also passed a constitutional amendment requiring a unanimous vote in both chambers and across two legislative sessions before slavery could be eliminated (Friedman 2011, p. 6–7).
Several authors argue the Founders were not improvising but created the Electoral College to accommodate federalism concerns. However, they define federalism as balancing the interests of large states and small states, not slave states and free states (Hardaway 1994; Adkins and Kirwan 2002; Boylan 2008; Ross 2012). In a debate over legislative apportionment on June 28, 1787, James Madison rejected this argument by pointing out that three of the largest states – Virginia, Pennsylvania, and Massachusetts – had very little in common other than size. They had different economic interests, religious traditions, and did not tend to vote together in the existing Congress (Madison 1893, vols. I, p. 255; see also Rakove 1997, p. 52–55, p. 66–70).
Instead, Madison viewed the federalism fault lines in sectional terms. “[T]he great division of interests in the U. States,” Madison (as cited in Farrand 1911, vols. I, p. 486) stated on June 30, “did not lie between the large & small States; it lay between the Northern & Southern” because of “their having or not having slaves.” Furthermore, as elaborated below, the entire notion of small states and large states is potentially deceptive because it depends on how (and if) enslaved people are counted in a state’s population. The debate over the apportionment of representation – in the Senate, House, and selection of the president – was the first major stumbling block of the convention. While slavery was not the only factor in this conflict, it played an integral role.
On July 5, a committee proposed the “Great Compromise”: allocating representation in the Senate equally and based on population in the House. The committee report included a previous recommendation to count “three-fifths of all other persons” in the census. Gouverneur Morris of Pennsylvania then offered a Northern alternative for apportionment in the House: a formula that allocated representation on both population and the wealth of all forms of property. Under this plan, which the convention rejected on July 9, Northern wealth from industry and commerce could counter the value of Southern land and enslaved people (Rakove 1997, p. 70–74).
At the other extreme, Southern delegates tried on three occasions to strike the word “three-fifths” thereby counting enslaved people as equal to Whites (Dougherty and Heckelman 2008), p. 297). The convention also rejected these proposals. Thus, it became increasingly clear that the convention would have to compromise if it were to succeed. Gouverneur Morris (as cited in Madison 1893, vols. I, p. 322) lamented that while he “could never agree to give such encouragement to the slave trade as would be given by allowing them a representation for their negroes,” he also knew the South would never “confederate on terms that would deprive them of that trade.”
After the adoption of what would become the Three-Fifths Clause, the Convention eventually moved to consider the presidency. The major debate over presidential selection was whether Congress or some external entity should choose the chief executive. Given that most state constitutions at the time required legislative elections for governor, congressional appointment was appealing, but it also raised concerns about whether a president would be adequately independent. Pennsylvania’s James Wilson was the strongest supporter of a democratically elected president, and he found an ally in Madison. Madison (as cited in Farrand 1911, vols. II, p. 56) argued on July 19, “the people at large” were “fittest” to choose the president, and a direct election “would be as likely as any [selection method] that could be devised to produce an Executive Magistrate of distinguished Character.”
Opponents of direct presidential elections utilized three lines of argument. On May 31, Elbridge Gerry of Massachusetts (as cited in Farrand 1911, vols. I, p. 48) cautioned against trusting the people to make such an important decision, arguing “the evils we experience flow from the excess of democracy.” Gerry’s anti-democratic stance was understandable, considering his home state had just finished quelling Shays’ Rebellion, but few other delegates shared his views (Hardaway 1994, p. 75). This critique only suggests the creation of some form of indirect election, not one that artificially inflates the power of slave states. Indeed, Gerry was no Southern sympathizer.
Instead of arguing the people should not have the right to select the president, the more common argument considered whether it was practical to expect the people to make an informed choice of presidential candidates (Slonim 1986, p. 40–41). Because of the geographic expanse of the Republic and the Founders misguided assumption that neither political parties nor campaigns would exist to educate voters, many delegates feared the people would vote for whichever candidate was most familiar to them. In the first election, everyone expected that person to be George Washington, but in future elections, the most recognizable candidate would likely be someone from a voter’s home state. If this were true, the large states would have a natural advantage over the small states (Finkelman 2002, p. 1147–1150).
Hugh Williamson, a delegate from North Carolina, spoke for many skeptics of direct presidential elections, when he stated on July 17 (as cited in Farrand 1911, vols. II, p. 32), “There are at present distinguished characters, who are known perhaps to almost every man. This will not always be the case. The people will be sure to vote for some man in their own State, and the largest State will be sure to succede.” In the very next sentence, Williamson added a third, distinctly Southern argument, noting, “This will not be Virga. however. Her slaves will have no suffrage.” Williamson, who originally proposed the three-fifths rule (Rakove 1997, p. 73), was not willing to cede back the power advantage the South had gained in the House in the context of presidential selection.
Two days later, Madison (as cited in Farrand 1911, vols. II, p. 57) indicated a willingness to compromise. He suggested an indirect form of presidential election would solve the problem of voters being unfamiliar with the candidates as well as the Southern problem. “There was one difficulty however of a serious nature attending an immediate choice by the people,” Madison observed. “The right of suffrage was much more diffusive in the Northern than the Southern States; and the latter could have no influence in the election on the score of the Negroes. The substitution of electors obviated this difficulty and seemed on the whole to be liable to the fewest objections.”
Madison (as cited in Farrand 1911, vols. II, p. 111) returned to this point a week later: “The second difficulty [of direct election] arose from the disproportion of qualified voters in the N. & S. States, and the disadvantages this mode would throw on the latter.” Madison may have been referring to the fact that some Northern states had lower property requirements for White males to earn the right to vote, but he also hoped that a direct presidential election would incentivize Southern states to allow a higher proportion of their White, male population to vote (Amar 2006, p. 157). Of course, what the South would never do is give any political rights to its enslaved people.
Thus, the Electoral College was born. Hugh Williamson’s observation about slavery determining which states are considered large and small proved to be correct. The 1790 Census revealed Virginia had 10% fewer White citizens than Pennsylvania, the second largest state, but enjoyed 20% more votes in the Electoral College (Amar 2017, p. 70). When confronted with the Electoral College’s failures in 1796 and 1800, Congress chose to reform, rather than abandon this system. Congress proposed the Twelfth Amendment fully aware of the size of the slavery bonus. Massachusetts Congressman Samuel Thatcher (as cited in Amar 2017, p. 70–71) complained that “[t]he representation of slaves adds thirteen members to this House in the present Congress, and eighteen Electors of President and Vice President at the next election.”
The reason the Twelfth Amendment did not replace the Electoral College with another executive selection system has everything to do with the Louisiana Purchase. Many Southerners had been anticipating this opportunity as far back as the late 1780s. Describing the ratification debate in the Palmetto State, Shugerman (2002, p. 268) observed, “South Carolina’s leadership recognized that the three-fifths compromise offered the slave states considerable advantages if they could accumulate more slaves. Additionally, the representation system in the Senate provided an incentive to establish as many slave states in the West as possible in order to balance out the North.”
Following the ratification of the 13th Amendment, the labels “slave state” and “free state” disappeared, but the war did not settle the broader sectional struggle. Instead, the conflict now pitted Republicans, who were committed to implementing the Reconstruction Amendments, against Democrats, who resented the war’s destructiveness and transformation of the Constitution. Both parties needed strong regional bases of power to advance their agendas. As McCarty et al. (2002, p. 433) noted, “At war’s end, the Republican party’s…survival depended on expanding its sphere of influence. There were two directions it could go: south or west.” Some tactics in this new fight, like the refusal of Southern states to obey the commands of the Fifteenth Amendment following the demise of Reconstruction, were clearly race-based strategies. Others, such as the effect of the Homestead Act and the Republican Party’s admission of unusually small states, were products of racially-benign motivations. However, as the quantitative analysis in the next section indicates, even decisions with non-racial origins can create major racial consequences over time.
Following the Civil War, the Republican Party dramatically reshaped the Constitution, addressing voting discrimination in several ways. Scholars who focus on the Fifteenth Amendment often overlook other important constitutional innovations. For example, the Fourteenth Amendment contained two reforms to the Electoral College. Section Two changed the apportionment for the House (and therefore the Electoral College) formula to count freed slaves as whole persons, and it included a Penalty Clause designed to prevent Southern states from backsliding on voting rights.
The Penalty Clause requires Congress to reduce a state’s House seats in proportion to the number of adult White males whose voting rights are “denied… or in any way abridged, except for participation in rebellion, or other crime.” Section 2 fulfilled the Republican commitment to civil rights, but it also served a strategic purpose. As Smith (1970, p. 829) observed, “It is well known that the Republicans intended to perpetuate their party’s ascendency after the Civil War. Section Two of the Fourteenth Amendment was…an important part of their plan.”
The demise of Reconstruction led to prolonged periods of divided government, which kept civil rights off the national agenda. The interests within the Republican Party also began to splinter during this time. When Republicans regained unified control in 1889, Representative Henry Cabot Lodge (R-MA) introduced a bill to provide federal court-appointed officials to supervise voter registration and elections in the South. Opponents labeled the initiative the “Lodge Force Bill,” the same terminology used to describe Reconstruction laws controlling Southern elections as well as Jacksonian legislation designed to end South Carolina’s nullification of federal tariffs. Like his opponents, Lodge (as cited in Upchurch 2015, p. 95) also viewed his legislation as a continuation of antebellum sectional struggles. He declared, “freedom was national and slavery sectional. So it may be said with equal truth that honest elections are national and dishonest elections are sectional.”
The bill passed the House on a party-line vote, but it faced a Democratic filibuster in the Senate. Senator George Hoar (R-MA) (as cited in Welch 1965, p. 519), tried to rally his colleagues, accusing the new Southern constitutions of engaging in “contrivances” to prevent African Americans from voting. “You have tried everything else,” he said to his Southern colleagues, “try justice.” The rhetoric failed, as did the cloture vote (Welch 1965, p. 520–521).
After the defeat of the Lodge Bill, Congress immediately proceeded to the Apportionment Act (Welch 1965, p. 521). Civil rights advocates had made impassioned pleas to invoke the Penalty Clause during each House reapportionment debate in the late 19th century (Rosin 2013), but an increasingly factionalized Republican Party lacked the will to pursue any further civil rights efforts in this session. The civil-rights oriented wing of the Republican Party was more prepared to fight over reapportionment 10 years later, during another period of unified government. Representative Marlin Olmsted (R-PA) introduced a measure to investigate the disenfranchisement of African Americans in the South with the intent of invoking the Penalty Clause. While the measure passed, the committee could not gather the relevant data before the Apportionment Act became law (Jenkins, Peck, and Weaver 2010), p. 60).
Even without punishing Southern Democrats, reapportionment in 1901 favored the GOP. The Republican leadership was content to act without the Penalty Clause if it meant avoiding another Democratic filibuster. During the floor debate, Representative Richmond Pearson (R-NC) (as cited in Rosin 2013, p. 61) lamented, “The fourteenth amendment…has failed of its purpose… It has strengthened the hands it intended to curb and cripple the creatures it intended to aid. The fourteenth and fifteenth amendments, instead of being muniments of right, have been converted into two black signposts, pointing deluded believers to their doom.” Knowing he faced defeat in the next election, Pearson secured a diplomatic appointment from President Roosevelt.
The death of Reconstruction and the factionalization of the Republican Party made it difficult to maintain a foothold in the South. The next best option for maintaining dominance lay out west. First, Republicans encouraged westward settlement through the Homestead Act of 1862, which made plots of land available nearly free of charge. While the benefits of the Homestead Act were available to citizens regardless of race, very few African Americans participated. In 1900, when homesteading had peaked, African Americans owned only 272 homesteads in the nine states admitted between the end of the Civil War and the end of the century (Taylor 1999, p. 152).
A number of factors may explain the lack of African American settlement in these newly-created states. First, many African Americans may have been reluctant to claim the benefits of the Homestead Act in light of earlier land policy failures. In 1866, Congress passed the Southern Homestead Act, which was designed to give 160-acre plots of land to freedmen. Much of the land, however, was non-arable. When Congress repealed the law in 1876, most of the land remained unclaimed, and three-quarters of successful land claimants were White (Shanks 2005, p. 35–36). Second, while land under the Homestead Act may have been free, it was very costly for African American farmers to free themselves from their sharecropping obligations and move west (Berlin 2010, p. 134).
Congress, during periods of divided government, also did little to make the West politically hospitable to African Americans. When Congress considered Nebraska’s application for statehood in 1867, it required the Nebraska Constitution to include a provision forbidding racial discrimination in voting rights. Congress made no such requirement for any state admitted during the rest of the century. While it would be tempting to conclude that Congress believed the ratification of the Fifteenth Amendment solved this problem, Congress reimposed non-discrimination requirements in the state constitutions of Oklahoma, New Mexico, and Arizona in the early 20th century, during Republican unified government (Biber 2004, p. 131).
As more Americans moved west to the Great Plains and Mountain West, Republicans attempted to consolidate power by strategically admitting these territories as states. According to one study (McCarty et al. 2002, p. 446), during periods of divided government a Republican-learning territory had, on average, a 1.6% chance of being admitted. When Republicans controlled Congress and the White House, such a territory had a 14.4% chance of being admitted. In order for this strategy to work, Republicans had to abandon norms regarding population size and economic resources required to support a functioning state government.
Consider the unflattering description used by English historian James Bryce (as cited in Stewart and Weingast 1992, p. 231) to describe Nevada’s 1864 admission. “[Nevada’s] population is obviously unworthy of the privilege of sending two men to the Senate, and has in fact allowed itself to sink, for all practical purposes, into sort of a rotten borough…” In fact, Nevada would not have a population large enough to justify one seat in the House until 1970 (McCarty et al. 2002, p. 497–498). Nonetheless, antebellum statehood norms gave way to the need to ease President Lincoln’s reelection bid and the ratification path for the Thirteenth Amendment.
In addition to Nevada, Republicans made similar decisions to admit Kansas and West Virginia during the Civil War, and Nebraska and Colorado during Reconstruction. With unified control following the 1888 election, Republicans admitted the Democratic-leaning Montana and Idaho, a toss-up state that was trending Republican. To offset any potential losses in the Senate (and the Electoral College), they admitted solidly Republican states Wyoming and Washington. The Dakota Territory, also a GOP stalwart, became two states so Republicans could benefit from twice the number of representatives, senators, and electors. The controversy surrounding the admission of these states may seem partisan in nature, but racial and sectional dimensions exist just below the surface. To justify splitting Dakota Territory into two states, Senator John Spooner (R-WI) observed (as cited in Stewart and Weingast 1992, p. 238), “The first gun fired by South Carolina upon the flag of the US gave slavery its death wound, but the relative position of the political parties to-day in this country is not changed with the disappearance of this institution.”
Where the Lodge Bill and the Penalty Clause failed to prevent the disenfranchisement of African Americans, the 1965 Voting Rights Act succeeded. The “Southern Bonus” in the House and the Electoral College described in the previous section has disappeared. In this section, I consider whether the westward settlement and statehood policies of the latter 19th centuries, though non-racial in their origin, created racially-disproportionate effects in the operation of the Electoral College today. Specifically, I measure whether states with Whiter populations or more racially resentful citizens possess more electoral votes per capita than more racially diverse and tolerant states.
Table 1 shed some light on the data generating process behind the modern Electoral College’s inefficient allocation of electoral votes. The table lists the 15 states with the highest electoral votes per capita in order, and one trend becomes immediately apparent – these states all have very small populations, and many of them are quite rural. Another interesting pattern appears regarding which political party created these states. Seven of these states were all admitted to the Union by a unified Republican government between the Civil War and the end of the 19th century.
|State||Year admitted||Partisan control|
|District of Columbia||1961||Democrat|
In breaking norms about when a territory could become a state, the Republican Party’s latter 19th-century strategy created a substantially higher population ratio between the largest and smallest states than what the Founders could have imagined. Using the apportionment formula that includes the three-fifths compromise, Virginia, the largest state in the 1790 Census, was 10.7 times larger than Rhode Island, the smallest state. By contrast, Census Bureau estimates of state population in 2018 indicate that California’s population is 68.5 times larger than that of Wyoming. Of course, it is theoretically possible population changes could reduce this trend. For example, Wyoming could go from being a small, heavily white state, to a large, racially diverse state, but such a development seems unlikely.
This growing disparity in population creates a growing small state bias in the Electoral College. According to one study comparing the electoral effect of the first and most recent Census, Ohio received 90% more power in the Electoral College in the election of 1804 than it would have if electoral votes were perfectly reflective of state population. Using 2015 Census estimates of state population, the same study found Wyoming enjoyed a 306% advantage in the Electoral College (Darling-Hammond 2017). Scholars, however, have not spent as much time trying to quantify the racial effect of Electoral College apportionment inefficiencies across states.
The state serves as the unit of analysis for this study. This conceptualization differs from most of the existing literature, which analyzes the effect of the Electoral College on voters (e.g. Gelman, Katz, and King 2002; Grofman and Feld 2005; Miller 2013; Gelman and Kemp 2015; Templon 2016) or parties (e.g. Nelson 1974; Shaw 2008; Thomas et al. 2013). State-level analysis, however, reflects the Founders decision to allocate votes to states, not people (see Rabinowitz and Macdonald 1986; Wright 2009). I also include the District of Columbia in the sample.
The dependent variable is each state’s electoral votes per million adult citizens. This approach differs from the Constitution’s requirement to apportion House seats based on the total population of a state, not its voting-age population. However, the voting-age population reflects the constitutional norm that states should cast their Electoral College votes based on the results of a popular election. Scholars of political economy use a similar construction of state-level electoral power when analyzing the distributive political preferences of presidents (e.g. Anderson and Tollison 1991). Population scholars also measure the political impact of migration patterns in terms of a state’s electoral votes per million voting age population residents (Frey 2000).
The independent variable of interest is defined as the white, non-Latino share of a state’s voting age population. The rural control variable reflects the share of state population living in a county of 50,000 or fewer residents. This reflects how ruralness is defined in recent scholarship (Cramer 2016, p. 56) and by the Office of Management and Budget. The data for these variables come from Census Bureau counts or estimates of state population aged 18 and above from each presidential election year beginning in 2000. The analysis also projects forward to the 2020 election, using the average annual change in state population from 2010 to 2017 to estimate the future population of each state.
The 2016 Cooperative Congressional Election Study (CCES) provides data to construct a racial attitudes control variable. Because CCES is large enough for its results to be representative at the state level, one can measure the average level of racial resentment among the people of each state. The value of this variable is the product of a factor analysis (Cronbach α=0.634) of the weighted, average response to four questions about race. Responses to each question are measured on a five-point scale with higher values indicating more racially resentful attitudes. The Appendix contains more information about the index.
Traditionally, scholars analyze continuous dependent variables using ordinary least squares regression, but OLS may not be an appropriate modeling technique for these data. The presence of outliers can lead to biased estimates because OLS places great emphasis on observations with very large residuals. The data in this analysis include several high leverage observations – those with extreme values on a predictor variable (Wyoming’s whiteness, ruralness, and racial attitudes, for example). The data also include several outliers – observations whose dependent variable values are unusual given the values of the predictor variables (the District of Columbia’s high electoral votes per capita given its very low White population). Outlier observations can bias OLS estimates by introducing heteroskedasticity. To mitigate the potential for influential observations distorting model parameters, the analysis employs MM-regression, a high-breakdown point estimator that is resistant to samples with a high proportion of outliers and high leverage observations (Verardi and Croux 2009).
Table 2 presents the results of several regression models analyzing the relationship between state racial and rural demographics and electoral power. For each presidential election, Table 2 presents models calculated using traditional linear regression and MM-regression. To identify which modeling technique is more appropriate, one must perform several tests. Cook’s distance is a regression diagnostic tool that can detect the presence of influential outlier data points. An analysis using Cook’s distance reveals several observations in each OLS model are highly influential. Scatterplots of Cook’s distance for all OLS models are available in the Appendix.
*p<0.10, **p<0.05, ***p<0.01.
The Breusch-Pagan χ2 statistic measures the presence of non-constant variance across observations. The results indicate five of the six OLS models suffer from heteroskedasticity, making OLS an inappropriate technique. Finally, the F statistic tests whether the independent variables fit the data better than an intercept-only model. Every MM-estimator model has a higher F statistic than its corresponding OLS model, which suggests an improvement in fit. Because the OLS models do not fare well under these model tests, the analysis proceeds using the MM-regression results.
The effect of race on state electoral power varies across recent elections, according to the MM-regression models. In 2000, a one percent increase in White population was associated with a 0.019 increase in electoral votes per million adults (p=0.002), but this coefficient dipped to 0.010 by 2016 (p=0.063). Figure 1 visualizes the partial effect of race on electoral power in 2016, holding the effect of ruralness at its mean value. Projecting forward to 2020, the MM-regression model predicts whiter states will continue to enjoy significantly more electoral power per capita (p=0.050). In each of the MM-estimator models, the rural share of state population is strongly associated with higher levels of electoral power per capita (p<0.01).
Table 3 attempts to contextualize the effect size of the race variable in each of the MM-estimator models on the median state, holding ruralness at its mean. As the data indicate, the median state’s voting age population ranges between 3 and 3.5 million people in the years covered in this study. Thus, for each election, the table displays the effect of a one-standard-deviation change in the share of a state’s White population on the dependent variable, multiplied by the median state’s voting age population.
|Year||Median State||+1 S.D. Δ Whiteness|
|V.A.P.||E.V.||Extra E.V.||Δ E.V.|
Consider what these results mean in 2016, for example. In that year, the average White, adult population share among the states was 74.5% with a standard deviation of 16.4%. To calculate the effect size, the state’s total population would remain constant, but one would hypothetically “swap out” enough non-Whites citizens to make the state 90.9% White (mean+one standard deviation). Multiplying the standard deviation by the model coefficient indicates this change would increase the electoral votes per million adults by 0.163. When multiplied against the median state adult population (about 3.4 million people), the effect is an extra 0.553 electoral votes. When compared to the median state’s existing eight electoral votes, racial bias represents a 6.9% increase in electoral power.
States also vary in terms of the levels of racial resentment among their people. Table 4 displays the result of an OLS and MM regression model of state-level racial attitudes on electoral votes per million adults. Once again, the presence of influential observations suggests MM-regression represents the more appropriate modeling choice. The racial index variable achieves statistical significance in the MM model (p<0.001), indicating that states with more racially resentful attitudes tend to have more electoral votes per capita. Applying the same procedures described in the previous paragraph, a one-standard-deviation change in conservative racial attitudes from the mean is associated with a 0.753 increase in electoral votes per million adult residents in 2016. This effect represents a 9.4% increase above its existing eight electoral votes. Figure 2 displays the results of the MM model.
*p<0.10, **p<0.05, ***p<0.01.
While these results are consistent and robust, I do not wish to overstate their substantive import. For example, it would be incorrect to conclude that Donald Trump’s 2016 victory was due to the racial bias in the Electoral College. Although Trump’s base includes Whites – especially those with racially resentful attitudes, Wisconsin and Pennsylvania served as the tipping point states in assembling Donald Trump’s Electoral College majority (Silver 2017). Both of these states, though Whiter than the average state, also receive fewer electoral votes per capita than the average state.
Long before Donald Trump arrived on the national scene, the Electoral College and race have been intimately connected. A persistent and evolving sectional struggle has promoted different forms of racial inequality within the Electoral College over the last two centuries. Sometimes, as in the adoption of the Three-Fifths Clause and the Jim Crow Era “Southern Bonus” in the Electoral College, sectional battles were fought on clearly racial terms. Other strategies, such as the Homestead Act’s racial settlement patterns and the Republican Party’s subsequent statehood admissions decisions, had racially-benign motives, yet created racially-disparate power structures.
While the racially unequal operation of the Electoral College undermines America’s core values, the question remains as to whether it has also altered electoral and policy outcomes. The sad answer is that a few presidents owe their elections to the Three-Fifths Clause or the Jim Crow “Southern Bonus.” Many more presidents had to adopt electoral and legislative strategies to account for the disproportionate influence of Southern states. In turn, the Slavery Bonus and postbellum Southern Bonus were determinative of several civil rights policy debates.
According to Humes et al. (2002, p. 456) the slavery bonus created by the Three-Fifths Clause constituted, on average, 6% of the size of the Electoral College between 1790 and 1860. This created a dominant Electoral College strategy: before the ratification of the 14th Amendment, every victorious presidential ticket, save for one, had a Southerner on it. Fortunately, the exception was Lincoln’s 1860 victory. On five occasions, Southerners were at the top of every presidential ticket on the ballot.
Several historians have observed that the slavery component of the Electoral College diluted Northern votes to such an extent that it cost John Adams re-election in 1800 (Amar 2017, p. 65). The racial implications of this election were important. While President Adams was on the verge of extending diplomatic relations to Haiti after its revolution, Jefferson’s administration attempted to undermine the black Haitian regime (Finkelman 2002, p. 1146).
The disenfranchisement of African-American voters following the Civil War intensified the effect of race in shaping elections and policy. For example, in the 1901 reapportionment debate, Representative Marlin Olmsted (R-PA) feared Congress’ failure to invoke the Penalty Clause could affect the outcome of a presidential election (Rosin 2013, p. 61). His concerns were well-founded. According to one study (Valelly 2004, p. 146), had the Penalty Clause been invoked following the 1910 Census, Woodrow Wilson, the last president to grow up in a slaveholding family, would not have won the 1912 election. Considering that President Wilson segregated the federal civil service in his first term, the racial implications of this alternative version of history are also quite salient.
Scholars have also estimated that the nonenforcement of the Penalty Clause gave the South (and therefore Democrats opposed to civil rights) 25 extra House seats per decade in the first half of the 20th century. The “Southern Bonus” in the House altered the outcome of approximately 15 percent of roll call votes during this era (as cited in Valelly 2004, p. 147). This finding is especially significant in light of how many New Deal programs were designed in a way to favor the interests of Southern Democrats. While there are numerous examples, I will focus on Social Security retirement benefits.
Until 1954, when Republicans regained unified control, a majority of African Americans were ineligible for Social Security retirement benefits. The law excluded farm and domestic workers, two industries with disproportionate numbers of African Americans. President Roosevelt’s original plan did not include this policy. According to the report of the president’s Committee on Economic Security (as cited in Lieberman 2001, p. 31), “We are opposed to exclusions of any specific industries within the Federal act.” To become law, however, the bill had to win approval from the House Ways and Means and Senate Finance Committees. Both committees had a Southern Democrat as chairman. The only time these chairmen asked questions during the Social Security hearings, they asked about whether to extend retirement benefits to farm workers and domestics (Lieberman 2001, p. 43).
The “Southern Bonus” in the House and Electoral College are gone, but other barriers prevent making federal policy more racially egalitarian. Ironically, the Republican Party’s strategic statehood admissions, which was done in part to give the party enough power to bolster the cause of civil rights, created states that are currently small and more racially resentful. Thus, these electorates exert an outsized influence in the Senate and the Electoral College. As displayed in Figure 2, six of the 10 states with the highest average levels of racial resentment were admitted by a unified Republican government during or after the Civil War. Many of these states are not usually considered swing states. However, emerging scholarship suggests non-competitive states, play a more important role in determining Electoral College victories than previously thought (Brams and Kilgour 2017; Cervas and Grofman 2017).
Scholars have argued for years about the Electoral College’s role in promoting democracy versus federalism. I hope to recast the normative debate on the Electoral College to consider how race is intertwined with these other concerns. At the Founding, this institution limited democracy in presidential selection and promoted a pro-slavery conception of federalism. Following the Civil War, the South undermined democracy through disenfranchisement, creating a Jim Crow-oriented federalism. The small state bias in the Electoral College today violates the democratic norm of “one person, one vote” to a historically high degree and creates a version of federalism in which race remains associated with unequal allocations of electoral power.
The questions used to construct the racial attitudes index (422c–422f, taken from version 4.0 of the 2016 CCES survey) measured whether respondents were “angry that racism exists,” whether they agreed “White people in the US have certain advantages because of the color of their skin,” whether they often find themselves “fearful of people of other races,” and whether “racial problems in the US are rare, isolated situations.” The publicly available sections of the 2012 and 2014 versions of the CCES included different racial resentment questions. The first asked respondents to compare Blacks to Irish, Italian and other immigrants, and the second asks respondents whether slavery and discrimination have made it difficult for Blacks to work out of the lower classes. As a robustness check, I developed weighted, state-level averages of the responses to these racial resentment questions (422a and 422b) in 2012 and 2014 and found a high correlation to the 2016 racial attitudes index (r=0.858 and 0.899, respectively).
The racial attitudes index is highly correlated with Trump vote share (r=0.907), which raises the possibility that the racial questions on the 2016 CCES simply activate respondent’s partisanship. However, state-level averages of the 2012 CCES racial resentment questions also highly correlate to the vote share of Mitt Romney (r=0.882), who did not appeal to racially resentful voters in his campaign. As indicated in Table A1, an OLS and MM-regression of state-level electoral power in 2012 using the 2012 CCES racial resentment data yields substantially similar results to those reported in Table 3 using 2016 data.
*p<0.10, **p<0.05, ***p<0.01.
*p<0.10, **p<0.05, ***p<0.01.
Adkins, Randall E., and Kent A. Kirwan. 2002. “What Role Does the ‘Federalism Bonus’ Play in Presidential Selection?” Publius 32 (4): 71–90. Search in Google Scholar
Amar, Akhil Reed. 2006. America’s Constitution: A Biography. New York: Random House. Search in Google Scholar
Amar, Akhil Reed. 2017. “The Inaugural Abraham Lincoln Lecture on Constitutional Law: Electoral College Reform, Lincoln-Style.” Northwestern University Law Review 112 (1): 63–81. Search in Google Scholar
Anderson, Gary M., and Robert D. Tollison. 1991. “Congressional Influence and Patterns of New Deal Spending, 1933–1939.” Journal of Law and Economics 34 (1): 161–175. Search in Google Scholar
Banzhaf, John F., III. 1968. “One Man, 3.312 Votes: A Mathematical Analysis of the Electoral College.” Villanova Law Review 13 (2): 304–332. Search in Google Scholar
Berlin, Ira. 2010. The Making of African America: The Four Great Migrations. New York: Penguin. Search in Google Scholar
Best, Judith. 1996. The Choice of the People?: Debating the Electoral College. Lanham, MD: Rowman & Littlefield. Search in Google Scholar
Biber, Eric. 2004. “The Price of Admission: Causes, Effects, and Patterns of Conditions Imposed on States Entering the Union.” American Journal of Legal History 46 (2): 119–208. Search in Google Scholar
Bickel, Alexander M. 1971. Reform and Continuity: The Electoral College, the Convention, and the Party System. New York: Harper & Row. Search in Google Scholar
Boylan, Timothy S. 2008. “A Constitutional Defense of the Electoral College and the Election of the American President.” The Open Political Science Journal 1 (1): 50–58. Search in Google Scholar
Brams, Steven J., and D. Marc Kilgour. 2017. “Paths to Victory in Presidential Elections: The Setup Power of Noncompetitive States.” Public Choice 170 (1): 99–113. Search in Google Scholar
Cervas, Jonathan R., and Bernard Grofman. 2017. “Why Noncompetitive States Are so Important for Understanding the Outcomes of Competitive Elections: The Electoral College 1868–2016.” Public Choice 173 (3): 251–265. Search in Google Scholar
Cramer, Katherine J. 2016. The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker. Chicago: University of Chicago Press. Search in Google Scholar
Darling-Hammond, Sean. 2017. “The Electoral College Is Even More Biased Than You Think. But Democrats Can Beat It.” The Nation, January 19, 2017. https://www.thenation.com/article/the-electoral-college-is-even-more-biased-than-you-think-heres-how-democrats-can-beat-it/. Search in Google Scholar
DeWitt, Larry. 2010. “The Decision to Exclude Agricultural and Domestic Workers from the 1935 Social Security Act.” Social Security Bulletin 70 (4): 49–68. Search in Google Scholar
Diamond, Martin. 1992. “The Electoral College and the American Idea of Democracy.” In After the People Vote: A Guide to the Electoral College, edited by Walter Berns, 44–68. Washington, DC: AEI Press. Search in Google Scholar
Dougherty, Keith L., and Jac C. Heckelman. 2008. “Voting on Slavery at the Constitutional Convention.” Public Choice 136 (3): 293–313. Search in Google Scholar
Edwards, George C. 2011. Why the Electoral College Is Bad for America. 2nd ed. New Haven, CT: Yale University Press. Search in Google Scholar
Farrand, Max. 1911. The Records of the Federal Convention of 1787. New Haven, CT: Yale University Press. Search in Google Scholar
Finkelman, Paul. 2002. “The Proslavery Origins of the Electoral College.” Cardozo Law Review 23 (4): 1145–1157. Search in Google Scholar
Frey, William H. 2000. “Regional Shifts in America’s Voting-Aged Population: What Do They Mean for National Politics?” 00–459. Population Studies Center. https://www.psc.isr.umich.edu/pubs/pdf/rr00-459.pdf. Search in Google Scholar
Friedman, Dan. 2011. The Maryland State Constitution. New York: Oxford University Press. Search in Google Scholar
Gelman, Andrew, and Pierre-Antoine Kemp. 2015. “The Effect of the Electoral College on Voting Power.” Vox.Com, July 19, 2015. https://www.vox.com/the-big-idea/2016/11/22/13713148/electoral-college-democracy-race-white-voters. Search in Google Scholar
Gelman, Andrew, Jonathan N. Katz, and Gary King. 2002. “Empirically Evaluating the Electoral College.” In Rethinking the Vote: The Politics and Prospects of American Election Reform, edited by Anne N. Crigler, Marion R. Just, and Edward J. McCaffery. New York: Oxford University Press. Search in Google Scholar
Grofman, Bernard, and Scott L. Feld. 2005. “Thinking About the Political Impacts of the Electoral College.” Public Choice 123 (1): 1–18. Search in Google Scholar
Hardaway, Robert M. 1994. The Electoral College and the Constitution: The Case for Preserving Federalism. Westport, CT: Praeger. Search in Google Scholar
Herz, Michael. 2002. “How the Electoral College Imitates the World Series.” Cardozo Law Review 23 (4): 1191–1218. Search in Google Scholar
Humes, Brian D., Elaine K. Swift, Richard M. Valelly, Kenneth Finegold, and Evelyn C. Fink. 2002. “Representation of the Antebellum South in the House of Representatives: Measuring the Impact of the Three-Fifths Clause.” In Party, Process, and Political Change in Congress, edited by David W. Brady and Mathew D. McCubbins, 1: 452–69. Palo Alto, CA: Stanford University Press. Search in Google Scholar
Jenkins, Jeffery A., Justin Peck, and Vesla M. Weaver. 2010. “Between Reconstructions: Congressional Action on Civil Rights, 1891–1940.” Studies in American Political Development 24 (1): 57–89. Search in Google Scholar
Katznelson, Ira. 2006. When Affirmative Action Was White: An Untold History of Racial Inequality in Twentieth-Century America. Reprint edition. New York London: W. W. Norton & Company. Search in Google Scholar
Klinkner, Philip. 2016. “The Easiest Way to Guess If Someone Supports Trump? Ask If Obama Is a Muslim.” Vox, June 2, 2016. https://www.vox.com/2016/6/2/11833548/donald-trump-support-race-religion-economy. Search in Google Scholar
Lieberman, Robert C. 2001. Shifting the Color Line: Race and the American Welfare State. Cambridge, MA: Harvard University Press. Search in Google Scholar
Levinson, Sanford. 2006. Our Undemocratic Constitution: Where the Constitution Goes Wrong (And How We the People Can Correct It). New York: Oxford University Press. Search in Google Scholar
Lineberry, Robert E., Darren Davis, Robert Erikson, Richard Herrera, and Patricia Southwell. 2002. “The Electoral College and Social Cleavages: Ethnicity, Class, and Geography.” In Choosing a President: The Electoral College and Beyond, edited by Paul D. Schumaker and Burdette A. Loomis, 161–175. New York: Seventh Bridges Press. Search in Google Scholar
Longley, Lawrence D., and Neal R. Peirce. 1996. The Electoral College Primer. New Haven, CT: Yale University Press. Search in Google Scholar
Madison, James. 1893. Journal of the Federal Convention. Edited by E.H. Scott. Chicago: Albert, Scott, & Co. Search in Google Scholar
McCarty, Brian D., Keith T. Poole, and Howard Rosenthal. 2002. “Congress and the Territorial Expansion of the United States.” In Party, Process, and Political Change in Congress, edited by David W. Brady and Mathew D. McCubbins, 1: 392–451. Palo Alto, CA: Stanford University Press. Search in Google Scholar
Miller, Nicholas R. 2013. “A Priori Voting Power and the US Electoral College.” In Power, Voting, and Voting Power: 30 Years After, edited by Manfred J. Holler and Hannu Nurmi, 411–442. Berlin: Springer. Search in Google Scholar
Nelson, Michael C. 1974. “Partisan Bias in the Electoral College.” Journal of Politics 36 (4): 1033–1048. Search in Google Scholar
Rabinowitz, George, and Stuart Elaine Macdonald. 1986. “The Power of the States in U.S. Presidential Elections.” American Political Science Review 80 (1): 65–87. Search in Google Scholar
Rakove, Jack N. 1997. Original Meanings: Politics and Ideas in the Making of the Constitution. New York: Vintage Books. Search in Google Scholar
Roche, John P. 1961. “The Founding Fathers: A Reform Caucus in Action.” American Political Science Review 55 (4): 799–816. Search in Google Scholar
Rosin, Michael L. 2013. “The Five-Fifths Rule and the Unconstitutional Presidential Election of 1916.” Historical Methods: A Journal of Quantitative and Interdisciplinary History 46 (2): 57–66. Search in Google Scholar
Ross, Tara. 2012. Enlightened Democracy: The Case for the Electoral College. 2nd ed. Dallas: Colonial Press L.P. Search in Google Scholar
Rothstein, Richard. 2017. The Color of Law: A Forgotten History of How Our Government Segregated America. New York: W.W. Norton & Co. Search in Google Scholar
Shanks, Trina Williams. 2005. “The Homestead Act: A Major Asset-Building Policy in American History.” In Inclusion in the American Dream: Assets, Poverty, and Public Policy, edited by Michael Sherraden, 20–41. New York: Oxford University Press. Search in Google Scholar
Shaw, Daron R. 2008. The Race to 270: The Electoral College and the Campaign Strategies of 2000 and 2004. Chicago: University of Chicago Press. Search in Google Scholar
Shugerman, Jed Handelsman. 2002. “The Louisiana Purchase and South Carolina’s Reopening of the Slave Trade in 1803.” Journal of the Early Republic 22 (2): 263–290. Search in Google Scholar
Silver, Nate. 2017. “Donald Trump Had A Superior Electoral College Strategy.” FiveThirtyEight, February 6, 2017. https://fivethirtyeight.com/features/donald-trump-had-a-superior-electoral-college-strategy/. Search in Google Scholar
Slonim, Shlomo. 1986. “The Electoral College at Philadelphia: The Evolution of an Ad Hoc Congress for the Selection of a President.” Journal of American History 73 (1): 35–58. Search in Google Scholar
Smith, George P. 1970. “Republican Reconstruction and Section Two of the Fourteenth Amendment.” Western Political Quarterly 23 (4): 829–853. Search in Google Scholar
Stewart, Charles, and Barry R. Weingast. 1992. “Stacking the Senate, Changing the Nation: Republican Rotten Boroughs, Statehood Politics, and American Political Development.” Studies in American Political Development 6 (2): 223–271. Search in Google Scholar
Taylor, Quintard. 1999. In Search of the Racial Frontier: African Americans in the American West 1528–1990. New York: W. W. Norton & Company. Search in Google Scholar
Templon, John. 2016. “How The Electoral College Favors White Voters.” Buzzfead, November 7, 2016. https://www.buzzfeed.com/johntemplon/how-the-electoral-college-screws-hispanic-and-asian-voters. Search in Google Scholar
Thomas, A. C., Andrew Gelman, Gary King, and Jonathan N. Katz. 2013. “Estimating Partisan Bias of the Electoral College Under Proposed Changes in Elector Apportionment.” Statistics, Politics and Policy 4 (1): 1–13. Search in Google Scholar
Upchurch, Thomas Adams. 2015. Legislating Racism: The Billion Dollar Congress and the Birth of Jim Crow. Lexington, KY: University Press of Kentucky. Search in Google Scholar
Valelly, Richard M. 2004. The Two Reconstructions: The Struggle for Black Enfranchisement. Chicago: University of Chicago Press. Search in Google Scholar
Verardi, Vincenzo, and Christophe Croux. 2009. “Robust Regression in Stata.” The Stata Journal 9 (3): 439–453. Search in Google Scholar
Welch, Richard E. 1965. “The Federal Elections Bill of 1890: Postscripts and Prelude.” The Journal of American History 52 (3): 511–526. Search in Google Scholar
Wright, John R. 2009. “Pivotal States in the Electoral College, 1880–2004.” Public Choice 139 (1): 21–37. Search in Google Scholar
©2019 Walter de Gruyter GmbH, Berlin/Boston