The Art of Polling: When Good Polls Go Bad

There are indeed those who view polling as a fairly reliable way to take the pulse of public opinion, yet even the best-conducted polls can sometimes go awry. The art of polling requires more than asking appropriate questions: timing, methodology, and interpretation come into play. When these factors mesh, polls can be right on target. However, even good polls can go bad under certain circumstances.

John Zogby, founder of the market research company and political polling service John Zogby Strategies, observes, “One of the reasons polls go wrong is through incorrect sampling. If the sample isn’t representative of the greater population, or significant subpopulations, then the results may not be valid. Examples include the 2016 U.S. presidential election, in which many polls did not detect late changes among pivotal groups of voters. In this case, it would be voters in the swing states.”

Another common trap is the phrasing of the questions: Even slight leading or ambiguous phrasing distorts the responses. For example, sensitive topics in immigration or healthcare would show different answers, depending on how the question was set.

Timing is also a critical factor. Polls taken too far in advance commonly miss the shifts in opinion that occur, especially among undecided voters who most frequently make their decisions late in races.

Even the best-designed polls can go wrong in how they are portrayed. Headlines might focus on topline results, with critical nuances in subgroup data ignored. A good poll can go “bad” simply because analysts or the media missed that complexity.

Polling is a science of balance, timing, and interpretation; even the best of polls can go haywire if all these elements fall out of step.

You may also want to read the following article What Does Robert F. Kennedy, Jr., Actually Want?