The Truth About Political Opinion Polls


We all have heard the numbers lately.  Week after week, almost daily, the media outlets report the percentage of Americans who favor Clinton versus Trump. Today, Trump is leading in the polls, tomorrow Clinton is in the lead, and by the next day they are both running neck and neck. The numbers fluctuate constantly, but this century old method (polling) has been used since the early 1900’s to predict the presidential elections. So are the preliminary reports really accurate?  Since the numbers change so much, are political polls even worth paying attention to?

For starters, polling is just one form of methodology used to record individuals’ opinions concerning a specific question or set of questions. The idea of polling is to sample enough people in the population so that the opinions represented can be generalized to the entire population.  This means there has to be enough people representing the diversity of all demographic groups (e.g., economic class, race and ethnicity, education level, gender or preferred gender, marital status, religious preference, political party, and geographic location) in order to accurately generalize the results.

Consequently, one issue with polling is that respondents may not always represent a reasonable sample of the population.  Particularly now that many polls are conducted online versus the traditional landline calls that were done not many decades ago.  A whole segment of a population may easily be missed for two reasons: 1) respondents have to be accessible to/for the survey, and 2) be willing to complete it.

The second issue with political polls is that opinions can change. How respondents vote during preliminaries is subject to change by Election Day in November.  Not to mention the unreliableness in the actual numbers of voter turnout.  It is no mystery that saying what you intend to do and what you actually do differs all the time. In the preliminary polls, 70 percent of respondents may say they indeed plan to vote, but then only 35 percent actually do.  There is just no guarantee.

The cost of conducting political polls to accurately predict the election outcome has become costly and time consuming as more people have cell phones over landlines.  This is because the Telephone Consumer Protection Act prohibits “automatic dialers” from calling cellular phones.  Automatic dialers were a low cost and time efficient method pollsters used to contact people via landlines.  Since more consumers have cellular phones and automatic dialers are prohibited, research companies would have to spend a sustainable amount of time and money for a live person to call one number at a time and they would need to make enough calls to get enough people to answer their survey.  Response rates for surveys are reportedly as low as 8 percent, meaning for every 100 calls made only 8 are likely to respond.

According to notable pollsters like Gallup and Pew Research Center, who have conducted opinion polls and presidential predictions for years, are now saying all the efforts in getting it right is just not worth it.  So, why does the media continue to report on preliminary poll numbers if the electoral outcome is not predictable and is unreliable?  It’s simple, all the talk about who is leading makes for good media attraction and ratings!