At the American Political Science Association (APSA) annual meeting, which took place in Boston last week, we were given several forecasts for the coming United States presidential election, to be held on November 4th. This kind of guesses is usually presented in the APSA meetings every four years, although with limited success so far.
Referring to controversies in previous occasions, Michael S. Lewis-Beck (University of Iowa) and Charles Tien (Hunter College and Graduate Center, CUNY) stated: “The statistical modelers are back. The presidential election forecasting errors of 2000 [when they predicted a comfortable victory of Al Gore] did not repeat themselves in 2004. On the contrary, the forecasts, from at least seven different teams, were generally quite accurate”.
Their “Jobs model” looks as follows:
Vote = f (Presidential Popularity, Economic Growth, Jobs Creation, Incumbency Advantage).
Alan I. Abramowitz (Emory University) applies the “time-for-change model”, postulating that the popular vote for president can be accurately predicted before the national conventions based on three factors: the incumbent president’s approval rating at mid-year, the growth rate of the economy during the second quarter of the election year, and the length of time that the president’s party has controlled the White House.
Barack Obama is predicted to receive 55.1 % of the major party vote to 44.9 % for John McCain. He bets that a Democratic victory in November is “all but certain”.
Along similar lines, Robert S. Erikson (Columbia University) and Christopher Wlezien (Temple University) project from the index of “leading economic indicators” and opinion polls.
Their forecasts point to a cautious prediction in favor ofBarack Obama with 52.2% and a 72% probabilityof popular vote victory.
Alternatively, Helmut Norpoth (Stony Brook) uses the “primary model” holding that the voting in presidential primaries is a leading indicator of the vote in the general election.
He predicts that Barack Obama would edge John McCain by the slimmest of margins, 50.1 to 49.9% of the major-party vote (which is within one unit of the standard error).
Interestingly, the “primary model” suggests that Hillary Clinton would have done somewhat better against McCain, while, in contrast, Republicans could not have picked a stronger nominee than McCain from the field of candidates in the primaries.
Somewhat more innovatively, Alfred G. Cuzán and Charles M. Bundrick (University of West Florida) use “the fiscal model” of presidential elections. This model rests on the premise that the share of the two-party vote going to the incumbent party is inversely related to changes in federal spending (relatively to Gross Domestic Product), so that the more the federal government “charges” the citizens for its services, the fewer votes the government’s party gets.
They anticipate that the incumbent Republicans will lose the White House in a close election, taking about 48% of the two-party vote.
Andrew H. Sidman (CUNY) and Maxwell H. H. Mak (Stony Brook) adopt a different approach based on multiple state-level forecasts instead of a single forecast at national level. This increases the number of observations and allows for translating popular votes into votes in the electoral college.
They predict that, in an election that may actually be too close to call, Barack Obama will be elected with 295 electoral votes over John McCain with 243 electoral votes.
Carl Klarner (Indiana State University) also analyzes district- and state-level factors for both the congressional and the presidential elections.
Obama is predicted to get 53% of the popular vote and 346 electoral votes and have an 84% chance of winning the presidency. He adds, however, that “racial considerations” regarding the candidate “could result in a Democratic defeat”.
See a summary of the forecasts of these and a few other “quantitative models” here: CLICK
In a previous piece published two years ago, I argued that this kind of empirical exercises would work better if they were supported by a more clearly specified model or hypothesis about the likely relationship between relevant variables. Usually, these forecasts do not discuss explicitly an initial model on the basis of reasonable assumptions regarding voters’ motivations or parties’ or candidates’ platforms. The exercises mentioned combine structural and conjunctural variables and opinion polls in a crude linear additive regression statistical model. Over successive elections, the most veteran forecasters have introduced numerous changes in the set and measurement of variables, but driven only by retrospective corrections, trial and error.
See also Douglas A. Hibbs' "Implications of the ‘bread and peace’ model for the 2008 US presidential election" in the new issue of "Public Choice" (Volume 137, Numbers 1-2 / October, 2008), which is less ad hoc than a lot of these other models.
Copenhagen, Denmark
‘Implications of the ‘Bread and Peace’ Model for the 2008 US Presidential Election’ By Douglas A. Hibbs, Jr. (Gothenburg University, Sweden)
Abstract
Presidential election outcomes are well explained by just two objectively measured fundamental determinants: (1) weighted-average growth of per capita real personal disposable income over the term, and (2) cumulative US military fatalities owing to unprovoked, hostile deployments of American armed forces in foreign conflicts. The US economy weakened at the beginning of 2008 and average per capita real income growth probably will be only around 0.75% at Election Day. Moreover cumulative US military fatalities in Iraq will reach 4,300 or more. Given those fundamental conditions, the Bread and Peace model predicts a Republican two-party vote share centered on 48.2%.
Forthcoming in Public Choice, September 2008
Blog:
James E. Campbell (University at Buffalo, SUNY) is the first to give us a prediction on the basis of variables including a post-convention opinion poll. He is also the first to predict a victory of John McCain:
James E. Campbell said...
Trial-heat poll & second quarter GDP (half for successors) [also convention bump model]
52.7% of popular vote for McCain with 57 83% probability of winning.
"although with limited success so far" I do not believe that this is a fair characterization. "Limited success" suggests that the forecasts have not done that well and this is inaccurate. Some forecasts have done quite well, even in 2000. As for the record last time out, I wrote in PS after the election: "From a collective standpoint, 2004 was a good year for the seven forecasts presented in the October PS. Two forecasts (Wlezien and Erikson, Lewis-Beck and Tien) were in the quite accurate range; two (Abramowitz, Campbell) were in the reasonably accurate range; and one (Norpoth) was in the fairly accurate range. One (Holbrook) was in the inaccurate range and one (Lockerbie) was very inaccurate. This was joined in the stratosphere of errors by the prediction from the leading forecasting model in economics. The venerable economist Ray Fair’s late July forecast of a 57.5% Bush vote was a whopping 6.3% greater than the actual vote (Fair 2004)."
Buffalo, NY
September 8th
Rein Taagepera said...
While very much hoping for Obama to win, I am only cautiously optimistic.
None of these models based on previous US elections could include a correction factor for one of the major candidate being black, because this situation hasn't occurred yet.
Recall Los Angeles Mayor Bradley who ran ahead in all polls but lost the race for governor of California to Deukmejian. He was black. Visibly, there are voters who are ashamed of their motives. So they tell the pollsters one thing and vote different in the end. Many US political scientists wonder about this "Bradley effect".
We had a similar situation in Estonia during the Singing Revolution. The degree of success of ethnically Estonian candidates was accurately forecast by opinion polls. The degree of success of ethnically Russian candidates was misrepresented by opinion polls. The Russian voters polled liberal, then voted reactionary.
Tartu (Estonia), before picking up the Skytte Prize in Uppsala and returning to Irvine (California) on 30 September.
I am a Professor in Political Science, elected member of the Academia Europaea and life-member of the American Political Science Association. I use this blog to interact with colleagues, former students and other people interested in the science of politics and political economy. All are welcome to use this resource.
0 Comments:
Post a Comment
Links to this post:
Create a Link
<< Home