Failed Us Presidential Election Forecasting

1) The 1936 ¨Literary Digest Poll¨ of the US Presidential election between Republican candidate Alf Landon and incumbent candidate Franklin Roosevelt. The magazine used availability sampling method to forecast that there would be a landslide victory for Alf Landon (55% Landon, 41% Roosevelt & 4% Lemke) but the election was won with a large margin by Franklin Roosevelt (61% Roosevelt, 37% Landon). Although this poll has been regarded has the most inaccurate election poll in the US presidential race, yet there has not yet been any proof to what went wrong with the predictions. Some of the suggestions by poll experts and researchers has been:

a. The sample was biased: The poll conducted polls only to people owning a car or a telephone or the both. This sample does not take into consideration poor people that do not hold any of these two properties who are the core supporters of Roosevelt.

b. The response small rate of the sample produced an error: It was recorded that the company conducting the poll sent out more than 10 million straw vote ballots to their sample but only received over 2.3 million responses in return. Out of the ones returned, Landon had more votes than Roosevelt which generated a huge error in the prediction of the poll.

2) The 1948 Crossley, Roper and Gallup presidential election poll used the quota sampling method to predict that Thomas Dewey would win the election over Harry S. Truman but the election result went the other. This failure was due to these reasons:

a. Basing projections on outdated data: The projection was based on data collected in 1940 after the 1936 presidential election which did not really reflect the population of the 1948 voting population.

b. Stopping data collection too soon: The polling companies stopped collecting data long before the election was conducted there not taking into consideration undecided voters and those that changed their mind towards the time of election.

c. Biased selection contributor by the interviewer: The sampling method gave the interviewer the power to decide on who should be interviewed so far they satisfy the quota control requirement of the procedure. as a result of this, working class voters were mostly ignored and they turned out to be deciding factor of the election.

3) The 1980 presidential election poll between Ronald Regan, Jimmy Carter and John Anderson. The pre-election poll predicted that Carter would win with 44% over Reagan - 41%. The election result turned as 50.7% Reagan and 41.0 Carter. The reason for this failed prediction by the press are:

a. Turnout: The poll presumed that there would be huge turnout of eligible voters which most of them were democrats.

b. Problem with response: Some experts reasoned that the turnaround in the election result could have been because some voters were not sincerely telling who they would support. Most democrats that

...