The silent majority problem and why campaign pollsters keep ‘fighting the last war’ on elections
The article argues that election pollsters repeatedly misread voters because they rely on flawed samples, outdated assumptions or turnout models that fail to capture who will actually vote. It says these errors can mislead campaigns and the public, from the Literary Digest’s mistaken 1936 forecast to more recent presidential elections.
The article traces polling’s history from the Digest’s postcard survey of 10 million people, which wrongly predicted Alf Landon would beat Franklin D. Roosevelt, through later polling failures. It cites the 1948 “Dewey Defeats Truman” headline, incorrect 2004 exit polls, Mitt Romney’s overoptimistic 2012 turnout assumptions, and pollsters’ 2016 prediction that Hillary Clinton would win. It also describes how campaign advisers such as Lou Harris, Pat Caddell, Richard Wirthlin and Dick Morris shaped modern political polling.
In short
- Polling errors have repeatedly misled campaigns and voters.
- Bad samples and turnout assumptions can distort election forecasts.
- The article says pollsters keep adapting to past mistakes.
Explained simply
Election polls estimate how people are likely to vote by asking a sample of voters and using it to represent the wider electorate. The result depends on who is included, how answers are adjusted and how pollsters estimate who will turn out.
Polling has a long record of both influencing campaigns and getting elections wrong. In 1936, a survey by the Literary Digest predicted that Republican Alf Landon would beat President Franklin D. Roosevelt, but its respondents were not representative of voters as a whole. Later misses, including the 1948 “Dewey Defeats Truman” headline, prompted pollsters to change their methods.
Campaign advisers such as Lou Harris, Pat Caddell, Richard Wirthlin and Dick Morris helped make polling a routine part of political strategy. Polls can shape what campaigns focus on and what the public expects, so errors in samples or turnout estimates can have effects beyond a mistaken prediction.
Art Culture Elections Politics World