Does Primary Parity Lead to the Presidency?

Does Primary Parity Lead to the Presidency? By Kevin Carter Professor James Lengle 12/4/2009

An Unprecedented Primary On June 2, 2008, then-senator Barack Obama pushed past the minimum 2,118 delegates needed for the Democratic Party s presidential nomination and claimed victory before a crowd of supporters in St. Paul, MN. Meanwhile, in New York, presidential contender Hillary Clinton was giving a speech of her own. In it, she recognized Obama s accomplishment but neither conceded nor dropped out of the race. The New York Times called her speech more defiant than conciliatory as she reiterated her belief that she was not only the stronger candidate but also the popular choice of voters (Zeleny 2008). Her speech should not have come as a surprise to anyone following what had become a bitter and tight primary race. Earlier that day, two states held primaries; Clinton won South Dakota and Obama won Montana. The outcomes were representative of the entire primary race, which lasted five months and consisted of an unprecedented 57 contests held across every state and US territory. The Democratic Primary seemed especially protracted when compared to the Republican s. John McCain had passed the threshold of 1,191 delegates needed to clinch his party s nomination on March 3, 2008 (three months before Obama claimed victory for the Democrats) (Jackson 2008). However, unlike Obama s nomination, McCain s was an almost foregone conclusion after winning New Hampshire and South Carolina then parlaying those victories into a successful showing on Super Tuesday. Further, unlike Clinton, McCain s primary opponent, Mike Huckabee (former Governor of Arkansas) bowed out of the race and encouraged his supporters to unite behind McCain

(Bush 2008). Huckabee would not win more than 12 percent of the vote in any primary after his withdrawal. Political pundits declared March 3 rd a double victory for McCain: He had won both his party s nomination and a head start on whomever would be his general election challenger. "Tomorrow, he can get started," Republican strategist Alex Castellanos told CNN. "He'll have the [Republican National Committee] behind him. He'll have a broad base of financial support. It's a big step. Meanwhile, it looks like the Democrats are engaged in the land war across Russia, so he's got a big advantage now." (Bush 2008) The Big Advantage Castellanos words echo a sentiment shared by many politicians, pundits, and academics. Prior to the 2000 Democratic Primary, the Democratic National Convention s (DNC) Rules and Bylaws Committee chose not to reform the primary system because it produced clear winners early in the process. The Committee s report to DNC chair Ed Rendell stated, Criticism about a front-loaded schedule has to be weighed against a process that seems to be working well. In the last few cycles, the current system has allowed the Democratic Party to identify its presumptive nominee early. As a result, the process has helped the Party unify behind its nominee and focus its resources on the general election Therefore, at this point, there is little incentive to change the system. (2000) Indeed, Al Gore started out as the heir apparent and easily captured the 2000 nomination. (Gangale 2007) Gore s challengers failed to win a single primary while George W. Bush and McCain fought it out for the Republican nomination. Although the

2000 Republican Primary was effectively over by March, McCain still defeated Bush in seven of the eighteen primaries he entered (Stanley and Richard 2007). The general election would prove more competitive than either candidate s primary with Gore winning 48.4 percent of the popular vote to Bush s 47.9. (However, Bush would win the majority of Electoral College votes and therefore, the presidency.) Building off of Castellanos comments, the Committee s report, and Gore s popular vote victory, one starts to see the genesis of a theory: The candidate with a less competitive primary is more competitive in the general election. The Primary System Since 1912, the parties have held presidential primaries to award delegates. Primaries originally started in twelve states and represented only a third of total delegates. As primaries grew in importance and size of delegate selection, it became increasingly difficult to win the party s nomination without winning a plurality of the primary vote. 1972 would be the last year a candidate would have less than a plurality of primary votes yet still win the nomination (Scammon and McGillivray 1988). Even today, unpledged, super delegates (party leaders and elected officials that retain autonomy in their voting on nominees) have been recast as representatives of popular will. Barack Obama s campaign manager, David Plouffe admits to feeding the notion that super delegates should vote the same as their districts voted (Kamarck 2009). Primaries are also more likely to represent the popular will of voters. Since the Reform Movement that established the McGovern-Fraser reforms, primaries have become increasingly binding in their selection of delegates (Kamarck 2009).

Table 1. Number of Binding versus Beauty Contest Primaries: Pre-Reform and Post- Reform Era. Year Presidential Preference Presidential preference poll poll is binding is advisory only (beauty contest) 1952 3 7 1956 3 7 1960 3 8 1964 3 9 1968 3 6 Reform Movement Reform Movement Reform Movement 1972 12 6 1976 17 8 1980 33 2 1984 19 6 1988 34 1 1992 35 0 1996 32 1 2000 30 5 2004 31 1 2008 40 2a a. In 2008, the Michigan and Florida primaries on the Democratic side were initially beauty contests only; eventually those delegates were seated with half votes. Source: Kamarck, Elaine C. 2009 Today, we can safely say that primaries are the most important formal part to the nominating process. Even in the early years of the post-reform era, political scientist James Lengle observed, Presidential primaries have become, without a doubt, the only Road to the White House. (1981) We know that winning the primary is all-important, but the question that follows is: Does it matter how you win it? Essentially, are Castellanos and the DNC Committee correct in their assertion that a non-competitive or uncontested primary increases general election viability? Alternatively, do the rigors of a protracted and intense primary race create a stronger nominee?

Revisionist History Four days after being mathematically eliminated from earning the Democratic nomination, Clinton again stood before her supporters. This would be her last speech as a candidate. In it she encouraged her followers to do all we can to help elect Barack Obama. (2008) While she preached party unity, her surrogates did their best to spin the heated rhetoric of the primary as a chance to battle-test Obama. Mr. Obama is a stronger candidate today thanks to Mrs. Clinton's pursuit of the Democratic nomination. Lanny Davis, former special counsel to President Bill Clinton, wrote in a Wall Street Journal opinion piece. And, far from being polarizing, she has helped bring the party together more than at any time in recent memory. (2008) Null Hypothesis The purpose of this paper is to examine the claim - made by both Clinton s and Obama s campaigns - that a close election creates a stronger candidate. Davis listed the following as potential reasons why a close primary would benefit the eventual nominee: 1) Grassroots support and official campaign organization is set up in nearly every state. 2) Increased media exposure. 3) Increased networks for fundraising. 4) Prepared nominee for general election debates. 5) Exposed potential flaws and weaknesses and allowed campaign to develop effective responses.

Alternative Hypothesis The alternative hypothesis is that the candidate with a less competitive primary is more competitive in the general election. The reasons for the alternative hypothesis are: 1) Concentrated financial support. 2) Support of the party s networks and organizations. 3) Focus issues and attacks on the opposition. 4) Competitive primaries demonstrate an inability to secure the base. 5) Avoids infighting and divisions that may last through general election. Support The literature on competitive primaries answers only part of my question. For example, a table of primary vote percentage and general election results among incumbent presidents showed a clear trend towards the alternative hypothesis that the less competitive the primary the more likely the candidate is to win.

Table 2. Incumbent s Percentage of: Primary Vote General Election Results Clear Sailing Woodrow Wilson (1916) 99% Won Calvin Coolidge (1924) 68% Won Franklin D. Roosevelt (1936) 93% Won Franklin D. Roosevelt (1940) 72% Won Franklin D. Roosevelt (1944) 71% Won Harry S. Truman (1948) 64% Won Dwight D. Eisenhower (1956) 86% Won Lyndon B. Johnson (1964)* 88% Won Richard Nixon (1972) 87% Won Ronald Reagan (1984) 99% Won Bill Clinton (1996) 88% Won George W. Bush (2004)*** 99% Won Tough Sledding William H. Taft (1912) 34% Lost Herbert Hoover (1932) 33% Lost Gerald R. Ford (1976) 53% Lost Jimmy Carter (1980) 55% Lost George Bush (1992)** 72% Lost *The Johnson primary vote includes votes for favorite sons and uncommitted delegates. ** It would seem that George Bush (1992) is the one anomaly in the table. He had a primary vote percentage greater than or equal to three of those listed in Clear Sailing but the author chose to list him in Tough Sledding. I do not know the author s reasoning; therefore, I have left him as originally placed in the table. *** I added George W. Bush to the Clear Sailing so the information would be up-to-date. Source: Levin, Jerome D. 2002, pp.105 Although Table 2 is not clear on where it draws the line between Clear Sailing and Tough Sledding, it does appear that, all things being equal, it is preferable to enter the general election with the support of your party at least for incumbents. Primary competitiveness is further examined in Table 3 in Thomas Gangale s work on the competitiveness of nomination campaigns in terms of number of candidates.

Table 3: Competitiveness of Presidential Nomination Campaigns, 1972-2000 Number of Candidates Democratic Party Republican Party Year Start End Start End General Election Winner 1972 13 9 uncontested Republican 1976 18 10 2 2 Democrat 1980 4 3 13 6 Republican 1984 13 8 uncontested Republican 1988 12 7 11 6 Republican 1992 8 5 2 2 Democrat 1996 uncontested 10 6 Democrat 2000 3 1 10 6 Republican 2004 10 1 uncontested Republican A. End = Candidates remaining in the race at the time of the national convention. Source: Source: Parshall and Mattei 2002, Parties and the Presidential Nomination Process, 41 It is clear that in the four uncontested contests, the unchallenged nominee went on to win the general election. However, Gangale points out that in four of the five elections without an uncontested nominee, the party with more candidates won. Gangale argues, Thus the historical data suggests the next best thing to an uncontested nomination, which can only occur in the case of a popular incumbent is a highly contested nomination. (2007) He suggests parties should promote highly contested elections as they generate more media attention, more interest in the party s candidates, more diversity of issues, and hopefully in the end, a more fully vetted political platform. (2007) However, his logic is extremely flawed. First, he does not consider that Gore was the general election s popular vote winner in 20o0. If he had listed the Democratic Party as the 2000 winner, his theory in favor of more competitive primaries would have held up only three out of five elections. Second, the remaining three uncontested races (in

which the general election winner had a more competitive primary) all involved incumbents. For example, in 1992, the Republican Party started and ended with two candidates for the nomination, and the Democratic Party started with eight and ended with five. The Democratic Party won the general election. Gangale s theory suggests that the Republicans would have fared better had they run more candidates in the primary. Gangale ignores the fact that the Republican nominee was a sitting president. 1976 and 1980 also saw sitting presidents who were challenged in the primary and lost the general. I doubt there are many people (besides Gangale) who believe that these incumbents would have benefited from more challengers. Although we can throw out Gangale s theory, we should not be as quick to throw out his data. Pulling from Table 3, we find that any year an incumbent is challenged in a primary, he loses. Any year the incumbent s primary is uncontested, he wins. Additionally, any year that there is no incumbent from either party (1988 and 2000) the less competitive party wins the general election. 1 When we account for incumbents and Gore winning the 2000 general election s popular vote, we find the opposite of Gangale s theory. We find that the less competitive primary race produces the more competitive nominee. Both tables provide us a good base from which to examine our hypothesis, but Table 2 tells us nothing about open races or comparative competitiveness (i.e. the primary percentage of one party compared to the other s) and Table 3 tells us nothing about the degree of competitiveness (i.e. it tells us how many candidates were in a primary race but not the percentage of the vote received). A study on the effects of 1 This assumes Gore as the 2000 election winner.

competitiveness will need to compare one nominee s primary percentage to the other s and account for incumbency advantage in primaries. Data and Methodology: My dataset consists of the 25 presidential elections since the parties introduced primaries (1912-2008). Table 4 lists the percentage of primary vote out of all primary voters. So, in 2008, McCain won 9.9 million out of 20.93 million votes in the Republican primary (or 47.3%) and Obama 17.86 million votes (or 48.2%) (Real Clear Politics, 2008). Originally, I was concerned that this would give me an inaccurate measure of competitiveness. The percentage of popular vote tells us nothing about the length of a primary race. McCain might have 47.3% of the popular vote, but is it accurate to call his race more competitive than Obama s? Yes, I believe that the percentage of primary vote is an accurate measure of competitiveness. The Republican and Democratic parties hold an almost identical number of primaries each election cycle. Further, although McCain essentially clinched the nomination in March, his early results, which included several losses of less than 20%, are reflected in his overall primary percentage. However, the popular vote percentage still may be skewed by several large states positioned early in the primary. These states would influence the primary percentage because a) they have large populations and b) their position at the beginning of the race would increase turnout. For example, Florida held its Republican primary in January and voted only 36% for McCain. That percentage could arguably skew the results since Florida represented 10 percent of total Republican primary voters. Essentially, McCain s

final primary percentage will look more like the elections at the beginning of the race (i.e. Florida) than the inconsequential ones at the end of the race (such as Oregon, which voted 81% for McCain but only represented 1.5% of Republican primary voters). This would not be true of Obama s primary percentage because his race was competitive until the last primaries. One way to test the skew is to average the nominee s vote percentage in each state. This would weigh each state equally regardless of size, turnout, or position in the primary. In making all states equal, this number is extremely flawed, but it provides a rough idea of a primary percentage skew. McCain s primary state average is 49.3% compared to a primary popular vote of 47.3%. Obama s primary state average is 53% to his primary popular vote of 48.2%. I predicted this average would be slightly higher than the popular vote percentage because there are fewer candidates to split the vote in the later states, which are now weighted equally with the early states. The result from averaging each primary is useless except in supporting the primary percentage as a measure of competitiveness. I am now comfortable saying McCain faced a more competitive primary than Obama despite conventional wisdom suggesting otherwise. Returning to Table 4, we find that in 21 out of 25 elections, the candidate that had a higher primary percentage and therefore, a less competitive primary race, won the general election.

Table 4: Comparative Primary Competitiveness and General Election Outcomes Primary % General % Primary Spread Republican Outcome 2008 McCain 47.3 45.7-1.0 Loss Obama 48.2 52.9 2004 Bush 99.0 50.7 38.0 Win Kerry 61.0 48.3 2000 Bush 62.0 47.9-13.8 Loss* Gore 75.8 48.4 1996 Dole 58.8 40.7-30.2 Loss Clinton 89.0 49.2 1992 Bush 72.0 37.4 21.9 Loss Clinton 50.1 40.0 1988 Bush 68.0 53.4 25.5 Win Dukakis 42.5 45.6 1984 Reagan 99.0 58.8 60.7 Win Mondale 38.3 40.6 1980 Reagan 61.0 50.7 9.9 Win Carter 51.1 41.0 1976 Ford 53.0 48.0 12.8 Loss Carter 40.2 50.1 1972 Nixon 87.0 60.7 61.7 Win McGovern 25.3 37.5 1968 Nixon 37.5 43.4 13.5 Win Humphrey 24.0 42.7 1964 Goldwater** 38.2 38.5-22.8 Loss Johnson 61.0 61.1 1960 Nixon 93.8 49.5 5.1 Loss Kennedy 88.7 49.7 1956 Eisenhower 85.9 57.4 33.6 Win Stevenson 52.3 42.0 1952 Eisenhower 27.1 55.1 25.5 Win

Stevenson 1.6 44.4 1948 Dewey 11.5 45.1-52.4 Loss Truman 63.9 49.6 1944 Dewey 11.6 45.9-59.4 Loss Roosevelt 71.0 53.4 1940 Willkie 0.7 44.8-71.3 Loss Roosevelt 72.0 54.7 1936 Landon 22.0 36.5-71.0 Loss Roosevelt 93.0 60.8 1932 Hoover 33.0 39.6-11.5 Loss Roosevelt 44.5 57.4 1928 Hoover 49.2 58.2 9.7 Win Smith 39.5 40.8 1924 Coolidge 68.0 54.0 65.8 Win Davis 2.2 28.8 1920 Harding 4.5 60.3-10.5 Win Cox 15.0 34.1 1916 Hughes 4.2 46.1-94.8 Loss Wilson 99.0 49.2 1912 Taft 34.0 23.3-10.6 Loss Wilson 44.6 41.8 * I list Gore as the winner of the popular vote in 2000. **Johnson s 1964 primary percentage includes votes for favorite sons and uncommitted delegates. At first glance, these results are encouraging. When the primary spread (the Republican primary percentage minus the Democrat s) is a positive number, it signifies the Republican candidate had the less competitive primary election. A negative spread signifies the Republican candidate had the more competitive primary election. Further, I find that the primary score corresponds to election outcomes 84% of the time. The two instances since 1972 (the year primary reforms took effect) that the primary spread did not predict the election outcome involved incumbents. To account

for a sitting president, I need to assign some point value for incumbency. For purposes of this test, I am assigning incumbency as worth 22 points in a primary election. The number is somewhat arbitrary but it is the value I can subtract from George Bush s (1992) primary percentage to get a negative spread (indicating a Republican loss) while still maintaining a negative spread (and Johnson victory) in 1964. The value 22 comes from those two spreads (-22.8 and 21.9) and appears to be an accurate reflection of how much incumbency is worth in a primary. Accounting for an incumbency advantage is also necessary to perform a regression with primary election and general election percentages. Table 5 shows elections with sitting presidents when we account for incumbency: Table 5: Re-Election Results with Incumbency Valued at 22 Primary Percentage Points Primary % General % -22 Primary Scores Primary Spread Outcome 2004 Bush 99.0 50.7 77.0 16.0 Win Kerry 61.0 48.3 61.0 1996 Dole 58.8 40.7 58.8-8.2 Loss Clinton 89.0 49.2 67.0 1992 Bush 72.0 37.4 50.0-0.1 Loss Clinton 50.1 40.0 50.1 1984 Reagan 99.0 58.8 77.0 38.7 Win Mondale 38.3 40.6 38.3 1980 Reagan 61.0 50.7 61.0 32.0 Win Carter 51.1 41.0 29.0 1976 Ford 53.0 48.0 31.0-9.2 Loss Carter 40.2 50.1 40.2 1972 Nixon 87.0 60.7 65.0 39.7 Win McGovern 25.3 37.5 25.3 1964 Goldwater 38.2 38.5 38.2-0.8 Loss

Johnson 61.0 61.1 39.0 1956 Eisenhower 85.9 57.4 63.9 11.6 Win Stevenson 52.3 42.0 52.3 1948 Dewey 11.5 45.1 11.5-30.4 Loss Truman 63.9 49.6 41.9 1940 Willkie 0.7 44.8 0.7-49.3 Loss Roosevelt 72.0 54.7 50.0 1936 Landon 22.0 36.5 22.0-49.0 Loss Roosevelt 93.0 60.8 71.0 1932 Hoover 33.0 39.6 11.0-33.5 Loss Roosevelt 44.5 57.4 44.5 1924 Coolidge 68.0 54.0 46.0 43.8 Win Davis 2.2 28.8 2.2 1916 Hughes 4.2 46.1 4.2-72.8 Loss Wilson 99.0 49.2 77.0 Adjusted for incumbency advantage, the primary spread correctly predicts the election 23 out of 25 times (92%) and in all 10 elections since primary reforms took effect. Regression Next, I wanted to see how well primary elections correspond to general election percentages. The data in Tables 4 and 5 suggest that having a less competitive primary is more likely to result in a general election victory, but I am still not certain whether the degree of competitiveness matters. For example, Obama ran in a less competitive primary than McCain and we can assume, using the spread analysis from earlier, that he would win the general election. We cannot; however, assume that had Clinton dropped out of the race and Obama pulled in a higher percentage of primary voters, he would

have won the general election by a larger margin. To make that assertion, we need to find a correlation between primary and general election percentages. I enlisted the help of fellow graduate student and friend Colin Shelby to assist me with the data analysis and statistical software STATA. Our first test used the original data for all 25 elections. The results were discouraging: R-squared equaled 0.2248, suggesting only 22% of the variance in the general election percentages is explained by the percentage of primary vote earned. Our second test used only data from the ten most recent elections. These results were more accurate (R-squared equaled 0.351) - most likely because they all occurred after the reform movement. However, these results still did not account for incumbency advantage. Our third and final test used data from the ten most recent elections while subtracting 22 points from any incumbent s primary percentage (see Table 5). These results did not fare much better. We achieved an R-squared equal to 0.3606. The results suggest that the general election percentage vote a candidate receives only corresponds to a small part of his or her showing in a primary. There are many other factors which may influence the election results during the period after the primaries up until the first Tuesday in November. My model does not capture these developments. However, from this final test, I can build a model for predicting the general election percentage of each candidate. The model is: General Election % = 34.2983+0.2434(Primary %) For the past ten elections, with incumbency valued at 22 points, this model is an average of 4 points off predicting the general election. In some cases, it came very close

(it predicted John McCain s score within 0.09% of the actual result) others, not so much (it was 6.87% off of Obama s). Conclusion I cannot say that Clinton endangered Obama s presidential hopes by extending the primary nor can I say that McCain would have won if he polled just a percentage point higher among Republicans. What I can say is that the trend over the past nearly hundred years points decidedly against claims that a more competitive primary creates a more competitive candidate. The fact is: In 21 of the 25 elections, the candidate with the less competitive primary race as measured by percentage of popular vote won the general election. When accounting an expected 22 point advantage for incumbents, the trend holds in 23 out of 25 elections and more important, in the ten most recent elections since the parties instituted primary reforms. Unfortunately, using primary percentages to predict the general election percentages does not hold up nearly as well. This correlation should not be mistaken for causation. We have no way of knowing if Edward Kennedy s run in 1980 weakened Carter s chances or whether it was the chinks in the presidents armor that encouraged an intraparty opponent in the first place. I tend to think it is a little of both. A candidate s inability to secure his or her base (represented by the primary voters which later become general election voters) is both a reflection of a candidate s weakness and a reason for independent and swing voters to question the candidate s electability. Future presidential hopefuls would be wise to win big and win early during the primaries, and if they can t, get out of the way.

References: "2008 Republican Popular Vote." Real Clear Politics. Web. 28 Nov. 2009. <http://www.realclearpolitics.com/epolls/2008/president/republican_vote_cou nt.html>. Bush, Dana. "McCain wins GOP nomination; Huckabee bows out." CNN.com. Cable News Network, 5 Mar. 2008. Web. 20 Nov. 2009. <http://www.cnn.com/2008/politics/03/04/march.4.gop/index.html>. Davis, Lanny J. "Hillary Made Obama a Better Candidate." WSJ.com. Wall Street Journal, 17 Oct. 2008. Web. 28 Nov. 2009. <http://online.wsj.com/article/sb122420333668243091.html>. Democratic National Committee. Beyond 2000: The Scheduling of Future Democratic Presidential Primaries and Caucuses, 11, 16. Washington, DC: Democratic National Committee, 2000. Gangale, Thomas. From the Primaries to the Polls How to Repair America's Broken Presidential Nomination Process. New York: Praeger, 2007, pp. 52, 54 "Hillary Clinton Endorses Barack Obama." NewYorkTimes.com. New York Times, 7 June 2008. Web. 28 Nov. 2009. <http://www.nytimes.com/2008/06/07/us/politics/07text-clinton.html>. Jackson, David. "McCain clinches nomination, turns attention to November." USAToday.com. USA Today, 5 Mar. 2008. Web. 20 Nov. 2009. <http://www.usatoday.com/news/politics/election2008/2008-03-04-goptuesday_n.htm>. Kamarck, Elaine C. Primary Politics: How Presidential Candidates Have Shaped the Modern Nominating System. Washington: Brookings Institution, 2009, pp. 12-19 Lengle, James. Representation and Presidential Primaries: The Democratic Party in the Post-Reform Era. Greenwood, 1981, pp. 6 Levin, Jerome D. Presidential Elections 1789-2000 (Presidential Elections Since 1789). New York: CQ, 2002. Print. Scammon, Richard M., and Alice V. McGillivray, comps. America at the Polls 2. Washington: Congressional Quarterly Inc., 1988. Print. Stanley, Harold W., and Richard G. Niemi. Vital Statistics on American Politics 2007-2008 (Vital Statistics on American Politics). New York: CQ, 2007. Print. Zeleny, Jeff. "Obama Clinches Nomination; First Black Candidate to Lead a Major Party Ticket." NewYorkTimes.com. New York Times, 4 June 2008. Web. 20 Nov. 2009.