Opinion: Voters shouldn’t base how they vote on polls

In recent years, U.S. elections have featured a preponderance of political polls, which influence how many citizens vote. Are these polls usually reliable guides? Should voters depend on them? Should there be government regulations governing these polls to ensure they are not fraudulent?

Polls can provide useful information about public opinion, but they are not election results or guarantees of what will happen on Election Day. They can also influence the very voters whose opinions they are supposed to measure.

The reliability of political polls depends on how a poll is conducted, whom it surveys and how its results are presented. A properly conducted poll can estimate public opinion reasonably well. However, even a professionally conducted poll can be wrong.

Suppose a poll has a margin of sampling error of approximately plus or minus 3 percentage points. That means the two candidates’ support levels could plausibly be much closer than the headline suggests or even reversed, considering sampling error alone.

The margin of error does not capture every possible source of error. A poll can have a small margin of error and still be inaccurate because it surveyed the wrong mix of people or incorrectly estimated who would actually vote.

Some polls can be wrong because those surveyed may not represent the electorate. Also, people who respond to surveys may differ from actual voters. Pollsters must estimate who will actually vote, not merely who says they intend to vote. Voters can change their minds after a poll is conducted. Importantly, the order and phrasing of questions can influence respondents’ answers.

Voters are advised to treat polls as information about the electorate, not instructions on how to vote. Polls measure opinion at a particular time. They do not guarantee what those polled or the wider electorate will do on Election Day.

But polls can definitely influence voters’ behavior. Some voters may support a candidate who appears to be winning. Others may rally behind a candidate who appears to be losing. Some may decide that their vote is unnecessary if their preferred candidate seems certain to win. These, however, are possible effects, not proof that every poll changes voting behavior.

Voters should be cautious about polls highlighting a candidate’s popularity. Popularity is not the same as the quality of a candidate’s policies. A poll cannot tell voters which candidate will best serve their interests, whether a promise is realistic or whether a candidate’s record meets expectations.

In the U.S., political polling is generally a form of speech and research. Requiring government approval of polling results could create its own dangers, particularly if the government could decide which results are acceptable.

A better approach would be to strengthen transparency and accountability without giving politicians control over the findings.

Pollsters should be required to identify their sponsors, survey dates, target populations, sample sizes, sampling methods, weighting procedures and full question wording. Any claimed margin of error should be appropriate to the sampling method. Steps should be taken to address fraudulent polling practices, including knowingly fabricating data, making deceptive claims about how a poll was conducted or misrepresenting sponsorship.

There should also be safeguards against government interference. Officials should not be able to suppress a poll simply because they dislike its findings, and independent pollsters should not need political permission to publish legitimate research.

The objective should be to regulate dishonesty and deception, not unpopular findings or legitimate differences in methodology.

A poll can be biased without being a scam. For example, a pollster may select an unrepresentative sample or frame a question in a leading way. Some of these problems arise from poor methodology rather than deliberate fraud.

Political polls have a legitimate role in a democracy. They can help the public understand what fellow citizens think and hold political leaders accountable when those leaders claim to speak for the public.

But polls also have limitations. Their findings can be selectively presented or misunderstood. The experiences of the 2016 and 2020 U.S. elections illustrated how polling errors can materially distort perceptions of a close race. Polling performed better in the 2022 midterm elections, demonstrating that accuracy varies by election and methodology.

To reiterate, there should be greater transparency in polling, independent scrutiny and meaningful penalties for deliberate fraud, while preserving the freedom to conduct and publish legitimate research.

Above all, voters should not let a poll decide their vote for them. A poll indicates something about what other people say they intend to do. It should not indicate what voters ought to do.

A more fundamental question is whether voters are choosing candidates based on their own examination of the issues, records and promises, or allowing the results of one or more polls to influence their judgment.

In a democracy, the purpose of polling should be to inform voters, not replace their independent judgment. When voters go to cast their ballots, they should have a sound understanding of the candidates they plan to vote for and against, rather than base their decisions on a poll indicating that one candidate is leading another by a few percentage points or by a substantial margin.