Let’s do away with pre-election candidate polling

The 2022 elections saw a record barrage of political spending and advertising. It also saw unprecedented polling and surveys, especially by the corporate media. While polling may be informative, it is also manipulative, and potentially destructive to American politics. For many reasons, the media should stop polling and also stop reporting on polls.

Pollsters will tell you that public opinion polling has become more complicated in the last decade. Fewer and fewer individuals answer their phone or are willing to participate in surveys. The demise of land lines and prevalence of cell phones with area codes not easily tied to identifiable political boundaries means that the costs of doing good polls has gone up because more people have to be called to get statistically meaningful results.

Moreover, since at least the 2016 presidential election the public has become increasingly skeptical regarding the accuracy of the polls. Polls up to the closing days predicted Hillary Clinton would win the presidency but she lost to Donald Trump in the electoral vote. In 2020 polls suggested a Joe Biden landslide yet the Electoral College vote was close, and had 43,000 votes swung the other way in Arizona, Georgia, and Wisconsin Trump would have been reelected. Now in 2022, Democrats defied the polls and had a stronger election than expected. As a result, many voters seem reluctant to answer the pollster, thereby feeding into the skepticism over polls.

Yet many of the problems with polls lie with the pollsters, or at least the media which reports them. For one, it makes no sense to report on national polls for presidential elections because we do not elect the president by popular vote. It is only the Electoral College that matters, and the reality is that the race for the presidency is across 50 states and the District of Columbia. When the state level data is examined for 2016, for example, one could see Clinton’s lead evaporate in Michigan, Pennsylvania, and Michigan in the last week as undecideds made up their minds and voted for Trump in these three states that gave him his victory.

Moreover, the media fails to really explain that polls are snapshots in time and not predictors of the future. Even the best polls done have a one in twenty chance of being statistically wrong. Most polls have margins of error or three to four points. Thus in close races a poll can be statistically accurate yet pick the wrong winner. Polls can frame the questions wrong or fail to predict who is really going to vote. Overall, there are lots of inherent limits to polls in forecasting election results.

But the real problem is that there are too many entities out there doing polling. Some are good, many are bad. Bad polls get reported alongside good polls in places such as Real Clear Politics or Rasmussen Reports.

For the bad polls, to save money they cut down on sample sizes or make other methodological short cuts, again resulting in inaccurate results. There are also partisan polls meant to impact voter behavior. Moreover, sites such as FiveThirtyEight rely on polls to make their election forecasts. I remember on Election Day 2016 FiveThirtyEight still said Clinton had nearly an 85% chance of winning. With mistakes like this no wonder the public believes there is voter fraud or manipulation.

But beyond bad polling, incessant polling is bad for democracy. For the corporate media, commissioning their own polls is a way for them to manufacture their own news. Do a poll, report the results. The poll becomes the news. News becomes the horse race about who is ahead or behind and not about where candidates stand on the issues.  For the corporate media, its poll also gets reported by other media sources, thereby making it the story and not the candidates.

But there is also evidence that polling impacts voter and candidate behavior.  Thirty years ago the “Made for TV Election” documented how commercial new stations manipulated coverage of candidates and races to enhance voter interested and maximize their ratings. There was evidence that reports of polls and early results on Election Day affected voter choices, creating a band wagon effect. By that, people changed their votes to go with a winner or simply chose not to vote because the polls suggested their candidate was going to lose.

Or consider how a poll can impact voters’ choices and candidate behavior while enhancing the bottom line for a media outlet.  Consider a poll that reports a couple of months before the election that a candidate has a substantial lead. One wonders if such a poll might influence decisions not to debate or otherwise impact campaign or voter strategy. The poll may or may not have been accurate, but subsequent polls showing a closer race created the impression of a horse race or a narrowing of the lead. Had the initial poll been accurate, who would have paid attention to the race after it? Political ad revenues might have dried up. It makes bottom line sense for a media outlet to report a narrowing of polls to encourage viewers to watch the horse race. More viewers mean more ratings, and more advertising dollars. There is a powerful incentive to do polling and report their results in order to enhance profits.

Incessant polling is bad for American democracy and elections. We would be better off if the media did less polling, covered more substantive policy and issues, and let voters and candidates decide elections.

David Schultz is a distinguished professor at Hamline University in the departments of Political Science, Legal Studies and Environmental Studies.