Can I Trust Polls?

Answers to your questions about opinion polls from the world's leading polling experts and tips to help you judge their quality. Can you trust them, and how do they actually work?

Explaining the world of opinion polls

Thousands of polls are published in newspapers, reports, and other publications every day. They help us measure the feelings of our neighbours or the wider public sentiment. In that sense, they are snapshots taken at a specific moment in time. The purpose of polls or their ability to forecast the future can be misunderstood, especially during election times. Not all polls are created equal, but how can you tell the difference between them?

We've asked our Professional Standards Committee expert on polls to answer the questions often posed to ESOMAR by those curious to learn more about the polls they see in their newspapers and elsewhere.

About Kathy Frankovic

Kathleen Frankovic
ESOMAR Committee Member, Consultant at YouGov

The Basics

What are polls?
What are key questions I should ask when evaluating a poll?
How are polls conducted?
Are they done differently across the world?

Don't polls get the results wrong a lot?

How accurate are election polls?
“The polls got it wrong!” This headline might lead people to think that opinion polls are not accurate but is this true?

In a paper summarising an analysis of a database compiled by Kantar of 31,310 polls from 473 elections and voting events across 40 countries from 1936 to 2017, Jon Puleston concluded that when examined at a global level, polls are generally very accurate. 

The average error of polls conducted within seven days before an election was shown to be +/-2.5%. (Are we getting worse at political polling, Jon Puleston, ESOMAR, September 2017).

The analysis also showed that the accuracy of polls would steadily increase the closer you get to the election event. Polls conducted before an election will have an average error roughly twice the size of a poll conducted in the last week of the election.

This can be because many voters make up their minds over the course of an election campaign, and opinions can change based upon campaign activities but also because of external events (strikes, scandals etc.), the number of undecided voters and even how many people actually turn out to vote.

A follow-up study in 2023 looking at 29 elections and 1,400 individual polls conducted between 2017 and 2023 shows very little change in the overall findings compared to what we learnt from the original analysis.

Most polls conducted within a week before an election are remarkably accurate average +/-2.5% variance with election results and 80% are accurate to within +/-3% and 98% correctly predict the outcome. The further out the polls is conducted before an election the bigger the variance gets and the more miss-reads on outcome and there is a simply linear relationship.

So why do election polls get it wrong sometimes?
There is so much coverage of election polls that when one is wrong those mistakes dominate the news and discussions about polling.
So, what went wrong in the UK 2015 general election?
What about the 2016 U.S. election mistakes?
Are enough people being asked?
Don't the errors point to a sampling issue?

Why have them at all?

Why do we conduct polls?
Your opinion actually matters to a lot of people
What do we learn from polls that we wouldn’t know otherwise?
Does having lots of polls make up for poll errors?

How do I make out the good from the bad?

How can I tell if a poll is any good?
Don't get duped by poor polls.
What does representative sample mean?
What's better? Random or representative?
Making sense of different types of sample techniques
Should I care about response rates?
Is declining participation an issue for polls?

Should there be more checks and controls?

Shouldn’t we ban pre-election polling?
Don't polls surely inevitably bias the results?
How independent are polling companies anyway?
Surely the reason why the 2015 U.K. polls were all wrong is that they colluded with each other – or reported results that would match everyone else’s.
Should all polls be vetted and verified by the government?
After the polling “errors” in 2015 and 2016, shouldn’t we just give up on polls?

Grab your copy of the questions and answers

We have created a handy document for you featuring the questions and answers from the world-renowned expert on polling. Your document to better evaluate polls you encounter.