2012
New Hampshire Primary Forecasts Obama Reelection
Lack of Primary Challenge is Key to November Victory
by Helmut Norpoth
January 11, 2012
The outcome of the New Hampshire Primary predicts that President Barack Obama will win a second term in the November election, defeating Mitt Romney or any other Republican challenger by a comfortable margin.
This forecast comes from a statistical model of presidential elections(The PRIMARY MODEL) that uses primary performance as a major predictor of the presidential vote in the general election. In addition, the model relies on a cycle in presidential elections and adjusts for the partisan shift during the New Deal era. The model covers elections as far back as 1912, when presidential primaries were first used in large numbers. Since 1952, however, only the New Hampshire Primary is included.
Overshadowed by the intense Republican battle, Obama won the primary contest of his party in New Hampshire in commanding fashion. For the record, Obama captured 82 percent of the votes in the Democratic primary of that state, against token opponents. Any time a candidate of the party that controls the White House has gone unchallenged for renomination, the odds are overwhelmingly in favor of that candidate’s victory in the November election. This goes back as far as 1912, when the incumbent president William Howard Taft first lost the primary battle and then the general election. Primary challenges to Truman (1952), Johnson (1968), Ford (1976), Carter (1980) and Bush (1992) augured poorly for them or their party in the general election.
The Primary Model allows for forecasts that pit Obama against each of the Republican contenders. The better their showing in the New Hampshire Primary, the stronger their showing against Obama in November. In the event that Mitt Romney, the winner of the Republican Primary in New Hampshire, goes on to be the party’s nominee, Obama would defeat him by 53.2 to 46.8 percent of the two-party vote. In a race against Ron Paul, the second-place Republican finisher in New Hampshire, the Obama vote would go up to 56.9 percent. It would level off at 57.1 percent against Huntsman, Gingrich, Santorum or any other Republican.
While Romney did exceedingly well in New Hampshire for an out-party candidate in primaries, the showing in presidential primaries carries far more weight in the November election for the candidate of the party in the White House than it does for the candidate of the party out of the White House. Voters are known to base their ultimate decisions far more on what they think about incumbents than challengers. Obama’s reelection prospect is also aided by the cycle evident in president elections. As reported earlier, this alone puts Obama at 51.8 percent of the two-party vote.
Link to Huffington Post article
While that may be too close for comfort, primary performance puts him in a comfortable position to secure reelection. The likelihood of an Obama victory against Romney, given the vote forecast of the Primary Model, would be 0.88. And, given the forecasts for races against any other Republican, the likelihood would be near certainty (0.99).
For the record, the Primary Model, with slight modifications, has predicted the winner of the popular vote in all presidential elections since it was introduced in 1996.
HISTORY AND PRIMARY:
THE OBAMA RE-ELECTION
Helmut Norpoth
Stony Brook University
and
Michael Bednarczuk
University of Wisconsin, Milwaukee
ABSTRACT
President Barack Obama is going to defeat Republican challenger Mitt Romney by a comfortable margin. Obama has history on his side as well as the fact that he was unchallenged in the primaries. The PRIMARY MODEL, which formalizes these predictors, forecasts an Obama victory with 53.2 percent of the two-party popular vote. This forecast assures Obama’s re-election with 88-percent certainty. The forecast model relies on primary elections as well as an electoral cycle, using elections as far back as 1912, the first year of presidential primaries. The primary performance of the incumbent-party candidate and that of the opposition-party candidate enter as separate predictors. For elections since 1952, the primary-support measure relies solely on the New Hampshire primary. In the period since then, no other primary beats New Hampshire in predictive power.
Democrat Barack Obama is going to defeat Republican Mitt Romney by a comfortable margin in the 2012 presidential election.
This forecast comes from a statistical model that uses the primary performance of the candidates and a cycle in presidential elections to predict the presidential vote. In plain English, Obama has history on his side as well as the fact that he was unchallenged in the primaries. The model, called The PRIMARY MODEL because of its heavy reliance on primaries, covers elections as far back as 1912, the beginning of presidential primaries.
Since 1952, however, only the New Hampshire Primary is used; we justify the choice of New Hampshire at some length.
Overshadowed by the intense Republican battle in that state this year, Obama won the 2012 primary contest of his party in New Hampshire in commanding fashion. For the record, Obama captured 82 percent of the votes in the Democratic primary of that state, against token opponents. Any time a candidate of the party that controls the White House has gone unchallenged for renomination, the odds are overwhelmingly in favor of that candidate’s victory in the November election. This was evident right in 1912, when the incumbent president William Howard Taft first lost the primary battle and then the general election. Primary challenges to Truman (1952), Johnson (1968), Ford (1976), Carter (1980) and Bush (1992) augured poorly for them or their party in the general election.
The moment the New Hampshire Primary was decided, the model was able to make a forecast for any match-up in November between Democratic and Republican candidates.
(first posted:
1.12.12
http://www.huffingtonpost.com/helmut-norpoth/new-hampshire-primary-for_b_1200199.html)
The forecast gives Barack Obama 53.2 percent of the two-party vote over Mitt Romney with 46.8 percent.
It is an unconditional forecast subject to no updating or other revision.
It assures Obama’s reelection with 88 percent certainty.
THE FORECAST MODEL
The PRIMARY MODEL, with some modifications, has proved itself in the four previous elections (Norpoth 2000, 2001, 2004, and 2008). It relies on the showings of candidates in primaries—hence the name of the model—along with a cyclical dynamic of presidential elections, and a partisan adjustment.[i]
The Primary Vote: Winning or losing the primary battle has proved to be good barometer of the outcome of the general election, going back all the way to 1912, when primaries were introduced. Our measure of primary support makes use of the vote percentage in primaries. For sitting presidents, who count on being renominated without challenge, we simply take the vote share received in primaries. For contests without a sitting president, where competition is natural, we form a measure of relative strength. We do so by expressing a nominee’s primary vote relative to the vote received by that candidate’s chief rival (the one with the next most votes, or the leading vote-getter if the nominee did not win the primary battle). Short of knowing who the nominee will be, this rule can be applied to any candidate in primary contests so we can make forecasts for all possible match-ups until the nominations are decided. [ii]
[i] For the 1912 election, the two-party vote was approximated through a regression of the congressional vote on the presidential vote. The intrusion of Teddy Roosevelt’s third-party campaign was so severe that the Republican candidate ended up in third place with only 23.2% of the total popular vote while Wilson, the Democrat, won with 41.8%. Using a regression of the House vote on the presidential vote in the 10 elections preceding and following the 1912 case (1872-1952), we derived an estimate of the two-party Republican vote in the 1912 presidential election (56.3%) that was used in this analysis. Note that the correlation between the two-party vote for president and House in that period was extremely high (.95).
[ii] The two-candidate measure of primary support was truncated within a range from 35 to 65 percent. The relationship between primary support and the general election vote is linear only within the restricted range of primary support.
From 1912 through 1948, the results of all primaries were included, but from 1952 onward only the New Hampshire primary is used in our model. In 1952, the adoption of a presidential ballot in the first primary dramatically changed to dynamic of the presidential nominating contest (Adams 1987, Buell 2000). How well does New Hampshire’s track record in predicting the November vote compare with that of other states since 1952? Let us first consider primary contests within the incumbent party, the one that controls the White House. Figure 1 includes all the states that held primaries in every election year since 1952. The correlation between the support received by such nominees in the New Hampshire primary and their vote in the general election in November is quite high, but it turns out, only second highest to Massachusetts.[i] So should we replace New Hampshire with Massachusetts?
Figure 2
Before considering that option, let us take a look at the primary contests within the party in opposition to the White House party, again for the states that held primaries in all elections years since 1952. As shown in Figure 2, New Hampshire dominates the rankings here.
[i] The opposition-party nominee’s performance in Massachusetts offers almost no guidance for picking the winner in November. The New Hampshire average of incumbent and opposition primary performance clearly beats the Massachusetts average.
Figure 3
Given the critical role of the order of primaries, we also considered the possibility that whichever state goes next to New Hampshire might offer a better prediction for the general election. We did this test for second, third, fourth, and fifth primaries past New Hampshire. None of them produce averages for incumbent-party and opposition-party primary performance that top New Hampshire. As a final test, we examined whether adding any other state to the New Hampshire equation produces a better prediction for the general election. The answer, according to Figure 3, is a resounding no. New Hampshire alone does better than combinations of New Hampshire and other primaries. [ii]
Primaries aside, at least one caucus state might clamor for attention in the election forecasting business. Iowa has featured the first contest in presidential nomination politics for quite some time, preceding the New Hampshire primary. The Iowa precinct caucuses have been a regular event only since 1972, so the window for testing their effect is smaller than it is for primaries. Nonetheless, could Iowa serve as better barometer of the general election during that time frame? Barack Obama, after all, followed up his victory in the Democratic caucuses in Iowa with a victory in November over McCain. The statistical evidence for the 1972-2008 period does not elevate his case to the norm. The correlations between support in Iowa and general election vote fall short of the ones for New Hampshire. For incumbent-party nominees, the correlation is 0.44 for Iowa compared with 0.69 for New Hampshire. For opposition-party nominees, the correlation for Iowa actually has the wrong sign (-0.33). Obama’s success in going from Iowa to the White House as the nominee of the out-party was truly exceptional. It would be a bad reason to revise the forecasting model.
The inclusion of Iowa would not be an improvement over New Hampshire in the PRIMARY MODEL.
What makes the New Hampshire primary such a superior predictor of the general election vote? For one thing, its first place on the primary calendar guarantees New Hampshire the nearly undivided attention of all presidential candidates and the media. Plus it is easy for candidates to enter the New Hampshire primary; getting on the ballot only requires a $1,000 filing fee from a candidate. Furthermore, New Hampshire allows Independents to vote in a party primary; they account for close to half of New Hampshire primary voters. Hence candidates are not just tested by the party faithful but also by large numbers of Independents, whose vote in November is pivotal for the outcome of the general election. With all the serious presidential candidates campaigning, and Independents eligible, voter turnout in New Hampshire is nearly as high as in the general election. It is hard to think of any other primary that puts candidates to such a rigorous test in every election year than New Hampshire.
The Presidential Cycle: In addition to primaries, the forecast model enlists a cyclical dynamic (Jones 2002, Norpoth 1995). Over the course of two centuries, American presidential elections have exhibited a distinct cycle. This is not the pattern associated with long-term partisan realignments but a more short-term cycle. As an example, take elections since 1960. Whenever the White House party was in its first term, it has won reelection in five of six cases, with an average share of 55.5% of the two-party vote. Compare that to instances in which the White House party went for reelection after two or more terms. Since 1960, it has lost six of seven such elections, averaging a vote share below 50 percent. It appears that the prospect of losing looms large for the presidential party when it has held the White House for two or more terms. But the White House party has little to fear when it has been there just for one term.
That may be so because a president who enters the White House by defeating the candidate of the incumbent party—perhaps the sitting president himself—starts out with a pledge of change: from Kennedy’s “New Frontier” in 1960, through Reaganomics in 1980, to Obama’s“Hope and Change“ in 2008. The electorate has loudly registered the demand, “It’s Time for a Change.” But change will take time to be implemented. At such moments in history the mass public may be willing to show some patience with the efforts of the new administration to work its magic. But when two terms are up the public may no longer be inclined to give the administration the benefit of the doubt. By then, of course, the two-term limit prevents a sitting president from running for another term, making the odds of retaining the White House even longer for the presidential party.
Whatever the explanation, the pattern illustrated for the last 50 years can be spotted in presidential elections all the way back to 1828, when popular voting took hold across the United States (Norpoth 2002). This “cycle” in presidential elections, which occurs in an irregular fashion, is handled by means of a second-order autoregressive process with a positive parameter for the first lag and a negative one for the second (Yule 1971). It produces an estimate for the average tenure of a White House party that comes to two and half terms. So it is not just recent history (since 1960), but electoral history going back to 1828 that augurs well for Obama in 2012. Having ousted the Republicans from the White House in 2008, he has an expected lease on the property for two terms, while leaving his party with a 50-50 chance of retaining it for another term. History alone forecasts an Obama victory with 51.8 percent of the two-party vote, though with only a 66-percent certainty.[i]
Table 1
Model Estimates
The parameter estimates of the forecast model, along with diagnostics, are shown in Table 1. Note that the dependent variable is the Democratic percentage of the major-party vote, regardless of whether that party was in the White House or not. As a result, the primary-support variables had to be inverted for elections with Republicans in control.[ii] With the primary support of the incumbent-party candidate carrying much greater weight than the one of the opposition-party candidate, whatever happened in 2012 on the Democratic side packs a bigger predictive wallop for the vote in November than what happened in the Republican contest. The estimates for the autoregressive vote parameters translate into an expected length of two and half terms of party control of the White House. That is good news for a first-term president like Obama in 2012, seeking reelection. Taken all by itself, the electoral cycle makes Obama the favorite this year, but the predicted vote margin would be too small to offer much comfort). Finally, the adjustment for pre-New Deal partisanship produces a constant estimate that suggests a close balance in presidential elections, notwithstanding the lead that Democrats have enjoyed in party identification for much of the time since the New Deal.
The 2012 Forecast
All of the information required by the Primary Model to make a forecast for the presidential election this November has been known since the New Hampshire Primary was held. Hence we can offer an unconditional forecast for the contest between Democrat Barack Obama and Republican Mitt Romney. This is a final forecast; there is no possibility of revision. The prediction equation for the presidential vote in 2012 (expressed as the Democratic share of the major-party vote) is:
.445 (DPRIM – 56.7) +.138 (RPRIM – 47.7) (-1) +.366 (53.7) -.333 (48.8) + 48.2
where DPRIM represents the primary support of the Democratic (incumbent party) nominee for President and RPRIM that of the Republican (opposition party) nominee, capped within a 35-65 percent range. It may come as a surprise to some observers that Obama was on the ballot in the New Hampshire primary. He did win in commanding fashion against a field of unknown vanity or write-in candidates. No Democrat of any stature challenged Obama for renomination in New Hampshire or any subsequent primary or caucus. New Hampshire thus provided a clear signal that Obama’s renomination would go uncontested. No sitting president who enjoyed this benefit has lost the general election. Meanwhile in the hotly contested Republican primary in New Hampshire, Romney won handily.[iii] For the 2012 general election, the Primary Model forecasts 53.2 percent of the two-party popular vote for Obama. This assures Obama’s reelection with 88-percent certainty.
Conclusion
In the 2012 presidential election, the twin advantages of history and primary make Obama a strong favorite to defeat Mitt Romney. Hardly any president who came to office in an election that ousted the White House party has lost his bid for reelection. This likelihood practically goes to zero for sitting presidents who faced no challenge in primaries. Using primary elections has numerous advantages for election forecasting. It enlarges the set of presidential elections needed for model estimation all the way back to 1912; it features not only the incumbent side, but also opposition candidates; and it provides a real-life test of the candidates’ electoral performance, not proxy variables or trial-heats. For over 50 years by now, New Hampshire has encapsulated the primary message. Considering primary contests in both parties, no other primary (or caucus, to include Iowa) tops New Hampshire in forecasting the outcome in November. Nor does adding any other state to New Hampshire improve the accuracy of the prediction. So as the first primary New Hampshire hands the Primary Model one more advantage—being the first to make an unconditional forecast of the presidential election in November.
References
Adams, William C. 1987. “As New Hampshire Goes . . .” In Media and Momentum, eds. Gary Orren and Nelson Polsby. Chatham: Chatham House.
Buell, Emmett H. 2000. “The Changing Face of the New Hampshire Primary.” In In Pursuit of the White House 2000, ed. William Mayer. New York: Chatham House.
Jones, Randall. 2002. Who Will Be in the White House? Predicting Presidential Elections. New York: Longman.
Norpoth, Helmut. 1995. "Is Clinton Doomed? An Early Forecast for 1996." PS: Political Science & Politics 28: 201-07.
Norpoth, Helmut. 2000. “Of Time and Candidates: A Forecast for 1996.” In Before the Vote:Forecasting American National Elections, eds. James E. Campbell and James C. Garand. Thousands Oaks: Sage Publications, 57-81.
Norpoth, Helmut. 2001. “Primary Colors: A Mixed Blessing for Al Gore.” PS: Political Science & Politics 34: 45-48.
Norpoth, Helmut. 2002. “On a Short Leash: Term Limits and Economic Voting.” In The Context of Economic Voting, eds. Han Dorussen, and Michaell Taylor. London: Routledge, 121-136.
Norpoth, Helmut. 2004. “From Primary to General Election: A Forecast of the Presidential Vote.” PS: Political Science & Politics 37: 737-740.
Norpoth, Helmut. 2008. “On the Razor’s Edge: The Forecast of the Primary Model.” PS: Political Science & Politics 41: 683-686.
Yule, G.U. 1971. “On a Method of Investigating Periodicities in Disturbed Series with Special Reference to Wolfer’s Sunspot Numbers.” In Statistical papers of George Udny Yule, eds. A. Stuart and M. Kendall. New York: Hafner Press, 389-420. (originally published in 1927)
[1] For the 1912 election, the two-party vote was approximated through a regression of the congressional vote on the presidential vote. The intrusion of Teddy Roosevelt’s third-party campaign was so severe that the Republican candidate ended up in third place with only 23.2% of the total popular vote while Wilson, the Democrat, won with 41.8%. Using a regression of the House vote on the presidential vote in the 10 elections preceding and following the 1912 case (1872-1952), we derived an estimate of the two-party Republican vote in the 1912 presidential election (56.3%) that was used in this analysis. Note that the correlation between the two-party vote for president and House in that period was extremely high (.95).
[1] The two-candidate measure of primary support was truncated within a range from 35 to 65 percent. The relationship between primary support and the general election vote is linear only within the restricted range of primary support.
[1] The corresponding correlation for states with primaries in fewer than all 15, but more than 11 elections: Maryland (.72), New Jersey (.59), Nebraska (.55), Florida (.35), South Dakota (.05), Ohio (-.07), and West Virginia (-.26). While Maryland beats New Hampshire it covers only 11 elections.
[1] The corresponding correlation for states with primaries in fewer than all 15, but more than 11 elections: New Jersey (-.45), Ohio (-.36), Maryland (-.30), Nebraska (.-.29), South Dakota (.08), Florida (.01), and West Virginia (.00). None of these states beats New Hampshire in this contest. The average performance for both incumbent-party and opposition-party contests favors New Hampshire over Maryland.
[1] Some might wonder how two primaries could produce a worse forecast than one alone. The reason is, simply, that adding predictors is costly, especially when the number of observations is quite small (15 for the full set of elections from 1952 to 2008). The adjusted R-squared is a measure of fit that takes that cost into account.
[1] This is based on the cyclical equation: VOTE (t) = 49.2 +.52 VOTE(t-1) - .47 VOTE (t-2), where VOTE refers to the Democratic percent of the two-party vote in 46 elections from 1828 to 2008. The cyclical forecast was posted September 22, 2011. (http://www.huffingtonpost.com/helmut-norpoth/comfort-for-obama-history_b_976246.html
[1] The inversion was done around the mean of 56.7 percent for incumbent-party candidates, and 47.7 percent for opposition-party candidates.
[1] Obama got 82 percent of the votes cast in the 2012 Democratic primary in New Hampshire, putting him safely over the 65-percent cap. Romney won the Republican primary in New Hampshire with 97,532 votes to 56,848 votes for second-place finisher Ron Paul, giving Romney a relative share of 63.2 percent among the top two finishers. http://www.cnn.com/election/2012/primaries/state/nh, 1/12/2012.
ENDNOTES
[i] This is based on the cyclical equation: VOTE (t) = 49.2 +.52 VOTE(t-1) - .47 VOTE (t-2), where VOTE refers to the Democratic percent of the two-party vote in 46 elections from 1828 to 2008. The cyclical forecast was posted September 22, 2011. (http://www.huffingtonpost.com/helmut-norpoth/comfort-for-obama-history_b_976246.html
[ii] The inversion was done around the mean of 56.7 percent for incumbent-party candidates, and 47.7 percent for opposition-party candidates.
[iii] Obama got 82 percent of the votes cast in the 2012 Democratic primary in New Hampshire, putting him safely over the 65-percent cap. Romney won the Republican primary in New Hampshire with 97,532 votes to 56,848 votes for second-place finisher Ron Paul, giving Romney a relative share of 63.2 percent among the top two finishers. http://www.cnn.com/election/2012/primaries/state/nh, 1/12/2012.
[i] The corresponding correlation for states with primaries in fewer than all 15, but more than 11 elections: New Jersey (-.45), Ohio (-.36), Maryland (-.30), Nebraska (.-.29), South Dakota (.08), Florida (.01), and West Virginia (.00). None of these states beats New Hampshire in this contest. The average performance for both incumbent-party and opposition-party contests favors New Hampshire over Maryland.
[ii] Some might wonder how two primaries could produce a worse forecast than one alone. The reason is, simply, that adding predictors is costly, especially when the number of observations is quite small (15 for the full set of elections from 1952 to 2008). The adjusted R-squared is a measure of fit that takes that cost into account.
[i] The corresponding correlation for states with primaries in fewer than all 15, but more than 11 elections: Maryland (.72), New Jersey (.59), Nebraska (.55), Florida (.35), South Dakota (.05), Ohio (-.07), and West Virginia (-.26). While Maryland beats New Hampshire it covers only 11 elections.
Helmut Norpoth
Dept. of Political Science, Stony Brook University, Stony Brook, NY 11794-4392
(631) 632-7640 (voice)
(631) 632-4116 (fax)
helmut.norpoth@sunysb.edu
Copyright 2012 primarymodel.com