The 2016 US Presidential Election showed that polling can be an ineffective tool for measuring outcomes. Former Secretary of State Hillary Clinton’s historic loss to Donald Trump was a long shot, yet, it happened. Polling data often doesn’t take into account gaps or the over or under representation of a particular population group.
Why Do Polls Exhibit Inaccuracies?
Public polling is a robust industry in the US and that’s one sign of a functioning and free society. However, after the 2016 election, many are skeptical about polling data.
It’s important to understand why the disparity between reality and polling results exists in the first place.
Polling Organizations Conduct Polls Differently
Take, for example, CNN and Fox. They conduct polls by telephone and using live interviewers. CBS and Politico field polls using opt-in panels online. The Associated Press and the Pew Research Center, however, use a panel of respondents who are recruited offline. A fourth way consists of robocalls and opt-in samples. These different approaches can yield different results and affect data quality. For instance, robocalls are causes of annoyance to many people and so are phone calls. At times, respondents may answer just to get off the phone. These problems with different types of polling can, at times, contaminate the data.
Barriers to Polling Have Disappeared
Today, the only barrier to someone wanting to conduct a poll is a mere few thousand dollars. This democratization has its benefits and drawbacks. The methods used to poll people will obviously vary in quality from a professional news organization to an independent agent. In 2016, fast and cheap polls overran the polling landscape. They failed to correct for over-representation of college-educated voters leaning towards Hillary Clinton.
With the danger of over representation of such polls still looming, it’s important to focus on marketing to narrow any gaps.
The Margin of Error May be Double that Reported
The margin of error is an oft-misunderstood term. It only addresses one source of potential error. Random samples are usually corrected for the population. However, non response, coverage error and mis-measurement are three possible further sources of error. Not including these sources implies to the public that they don’t exist.
Polls Can Dissuade People from Voting
If a candidate is leading by large margins in polls, there is evidence that it dissuades people from voting. Voters that lean towards the winning candidate may assume their vote won’t make a difference and stay home. To keep themselves from being surprised on election day, candidates should persuade as many people to get out there and vote. The ability to motivate and connect with voters on a personal level can bridge the gap between inconsistent margins of polling error. This is where an experienced voiceover can make or break a campaign. The potential returns are too crucial to pass up.
