Question

Some of the most widely publicized surveys are those related to elections. Sometimes the projections made even a day before an election are grotesquely inaccurate, yet the exit polls on the day of election are almost always accurate. How do you explain this difference in the degree of accuracy

Answer 1

The variation in accuracy of the exit polls and surveys is based on social desirability bias. As per this bias, the voters vote for someone else in the elections but claim to have voted someone else in the exit polls. This is done under the social fear of being publicly criticized for their voting preference. Hence they claim to have voted for the most popular candidate in the exit polls. However when the results come out, there is a considerable difference in the results claimed by exit polls and that claimed by actual counting. This effect is also known as the Bradley effect. Hence, not much importance is given to the results of exit polls and surveys as the accuracy of such results can be quite varying.