Tracking Polls

One of the first successful polls came in the form of Literary Digest’s presidential polling in 1916 where, they created a straw poll.  A straw poll is extremely unscientific and based solely on luck, and therefore when they predicted Woodrow Wilson’s victory, the public was amazed, describing it as “uncannily accurate.”  However, twenty years later, it predicted that Alfred M. Landon would defeat Franklin D. Roosevelt, with the opposite actually occurring – by a landslide as well.

This sparked the need for more scientific approaches to polling.  Political analysts and statisticians began to track the errors of Literary Digest.  The first was fairly clear; it sampled from largely the middle class by gauging opinion from telephone directories and automobile owners.  Wealthy Republicans were therefore overrepresented in the sample.

The second error was timing in which questionnaires were mailed to sample subjects in September, neglecting the two years after in which the results would change.

The third and final error occurred due to the method of sampling; only those individuals who cared about the election greatly would respond while the majority (78%) simply did not respond to the surveys.

The Gallup Organization responded by making its own poll which was much more scientific than the techniques used by Literary Digest.  Gallup was much more successful than the Literary Digest, predicting almost every election until 1948 when he miscalculated the very close competition between Thomas E. Dewey and Harry S Truman.

Polling has nonetheless become a  lucrative industry of its own.