What Can Pollsters Learn From Polling Errors: Challenges & Solutions

To summarise, we can say the most significant challenges to effective boil down to:

Inaccurate results

Citizens use information shortcuts when making political decisions- with new and personal information driving out the old and impersonal. This can often lead to inaccurate predictions.

For example, most pre-election polls during the 2020 U.S. presidential elections overstated Joe Biden’s lead over Donald Trump in the national vote for president and, in some states, incorrectly indicated that Biden would likely win or that the race would be close when it was not. 

These problems led some commentators to argue that “polling is irrevocably broken,” ” pollsters should be ignored, ” or that “the polling industry is a wreck and should be blown up.” 

With the public lacking fixed preferences on many issues, political actors have ample incentive to supply those shortcuts in ways that might broaden support for themselves and the policies they champion. In a democracy, citizens are typically more concerned with some matters than others, and most citizens are not continuously engaged in public affairs.  

Opinion bias

The public does not have fully formed and unambiguous views on many issues. That does not mean there is anything wrong with the public.

People sometimes answer survey questions just to be polite—because they probably ought to have an opinion. That gives pollsters a lot of running room to “manufacture” opinion, especially on issues of narrow rather than wide concern. Even with strong views, a single polling question rarely captures those views well. 

Opinions that pitch one candidate against another might capture which candidate people prefer but fail to capture the respondent’s willingness to go out and vote, thereby failing to predict whether the opinion will have any real impact.

Take the 2016 U.S. elections, for example; it was a classic case of biased opinion. Pre-election polls and forecasts consistently predicted that Democrat Hillary Clinton would win a resounding victory over Republican Donald Trump. Still, he surprised the world by winning 56.8% of the electoral college vote to capture the presidency. 

Clinton supporters perceived a poll as more credible when it showed her with a comfortable lead than when it showed a neck-in-neck race; the opposite was true for Trump supporters. 

Misleading information

Coupled with inaccuracy and biased opinions, polls often put out misleading information. When interest groups commission pollsters to ask leading questions to gather “scientific” proof that the public agrees with whatever demand they make on the government, they demean polling and mislead the public. 

When analysts, sometimes innocently, use poll numbers as a definitive guide to public opinion, even on issues to which most people have given little thought, they are writing fiction more than citing fact. When political consultants use information gathered through polling and focus groups to camouflage their clients’ controversial policies with soothing, symbol-laden, and misleading rhetoric, they frustrate democratic deliberation.