POLI 325                                                                                                                                 9/7/04

FORECASTING PRESIDENTIAL ELECTION OUTCOMES

            Let us consider how — and how accurately and how far in advance — the outcome of Presidential elections can be forecast. We need to distinguish between short-term forecasts (made shortly before the Presidential election) and long-term forecasts (made at the outset of the general election campaign or even earlier).

            We can consider three types of forecasts: (i) pre-election polls and surveys; (ii) predictive models based on long-term patterns in aggregate data; and (iii) a more qualitative scoring method.

            In the relatively distant past, pre-election polls taken even shortly before election day have sometimes been disastrously and famously wrong — most notably, a Literary Digest poll predicted the defeat of President Roosevelt in 1936 and the Gallup Poll (and almost all other polls) predicted the defeat of President Truman in 1948. However, as a result of refinements in the art and science of survey research and polling, election-eve polls now are extremely accurate. On the other hand, polls taken far in advance of the election (in particular, “trial heats” in the preceding primary season) bear almost no predictable relationship with actual election outcomes.

            However, a number of political scientists (e.g., Michael Lewis-Beck, Alan Abramowitz, James E. Campbell) and economists (e.g., Ray Fair) have devised predictive models that are based on aggregate data available well in advance of election day and that are (perhaps) surprisingly accurate in predicting Presidential election outcomes. You will find an attached news story that appeared in the Washington Post at the outset of the 2000 general election campaign and a summary handout that was presented at an American Political Science Association roundtable on “Forecasting the 2004 Presidential Election.”

            Finally, historian Allan Lichtman has proposed a more complex and somewhat qualitative method for predicting winners in Presidential election in a book called The Keys to the White House.

QUESTIONS TO CONSIDER IN CLASS ON SEPTEMBER 10

1.         A month or two prior to the Presidential elections in the 1920s and 1930s, the Literary Digest magazine would collect from telephone companies and state motor vehicle departments lists with the names and addresses of telephone subscribers and registered automobile owners. The magazine would then mail out many millions of “sample ballots” to these names and addresses. Many people did not respond, of course, but the Digest still received millions of ballots, on the basis of which it forecast the outcome of the election. This procedure was quite successful in predicting the Republican presidential victories in the 1920s and also Franklin Roosevelt’s victory in 1932 but it predicted that President Roosevelt would be badly defeated for pre-election in 1936, when in fact he won in one of the greatest Presidential landslides of all time. (The Literary Digest went out of business soon thereafter.) Can you account for why the Literary Digest forecasts that had previously been successful failed so badly in 1936? 

2.         Attached you will find Presidential trial heats, a collection of charts from an article in the British Journal of Political Science (which, perhaps surprisingly, publishes some of the best scholarly work on American politics) that present information concerning pre-election polls from 1952 (after the lessons of 1948 and earlier had been learned by Gallup and other polling organizations) through 1992. Study the diagrams and the notes at the bottom of the page carefully. What generalizations can you make concerning the accuracy of pre-election polls? (A compilation of Bush-Kerry trial-heat polls from May 1 through August 31 can be found at PollyVote: http://morris.wharton.upenn.edu/forecast/Political/index.html .

3.         The predictive models constructed by Fair, Lewis-Beck, and most others forecasters use information concerning (i) the performance of the economy and (ii) the popularity of the incumbent President that is available long before the election to predict the proportion of the popular votes received by the Presidential candidate of the party the controls the White House. (The models differ with respect to the exact measures of economic performance and Presidential popularity that are used and with respect to what other variables [if any] are also used.) Attached you will find data for economic performance and Presidential popularity, as well as the percent of the popular vote received by the incumbent party’s Presidential candidate, for each Presidential election from 1948 through 2000. Can you devise some kind of predictive formula based on this data? According to your formula, what combination of economic growth and approval does President Bush need to be fairly confident of re-election?

4.         Another attached page presents Lichtman’s “thirteen keys” to winning the Presidency and his evaluations of the keys for each Presidential election from 1860 through 1992. Some of the keys have clear true/false values (e.g., Key 1) while others (e.g., Key 2) are fuzzier. How would you evaluate the keys for 1996 and 2000? Given these evaluations, did Lichtman’s method correctly predict the 1996 and 2000 outcomes? What keys for 2004 have already “turned” (or not)? What does Lichtman’s method (begin to) predict for the 2004 election?

5.         As you may know, incumbency advantage appears to play a decisive role in Congressional (especially House) elections (when typically the overwhelming majority of incumbents who seek re-election are successful). Does the record indicate that there is a similar incumbency advantage in Presidential elections? Using the any record of Presidential elections (such as is found in the appendix of almost every introductory American Politics textbook), we can classify all Presidential election into two types: (1) an election with an incumbent seeking re-election and (2) a “open seat” election. Then, within each type, classify each election into a “hold” for the incumbent party or “turnover” to the challenging party. (To illustrate example, the last four elections illustrate all four possible types: 1988: open-hold, 1992: incumbent-turnover, 1996: incumbent-hold, and 2000: open-turnover.) Does the resulting cross-classification support the incumbency advantage thesis? Does it provide an unbiased test of the hypothesis?