‘Margin of Error’: Can election polls reliably predict races?

In 2016, Parry said people didn’t seem to consider margin of error. It was a huge factor in 2016’s polling and why so many people were left surprised.

LITTLE ROCK, Ark. — In 2016, a lot of people said they were surprised by the outcome of the presidential race as some polls seemed to favor Hillary Clinton taking the presidency. 

Fast forward to 2020 and that obviously was not the case. What happened with the polls? Are polls reliable enough to teach us about outcomes here in Arkansas?

University of Arkansas Political Science Professor, Janine Parry, said polls are incredibly accurate if conducted the right way. In 2016, Parry said people didn’t seem to consider margin of error.

It was a huge factor in 2016’s polling and why so many people were left surprised.

“With respect to 2016, the great majority of polls were not only accurate, but they were nearly perfect,” she said.

She said there were a handful of super competitive states well within the margin of error. Those were what led to some people’s disappointment because the electoral college came into play.

What is the margin of error? Why is this one of the most crucial parts of polling?

“For example, if the Arkansas Poll shows Donald trump is likely to get 63% of the popular vote and we have a margin of plus or minus four, there is 95% confidence that the final result will be within four points of that 63 percent,” she said.

The Arkansas Poll that Parry mentioned is one of the most comprehensive election related polls in Arkansas. Parry oversees it.

It accurately predicted President Trump would win Arkansas in 2016 and Tom Cotton in 2014.

For this year’s 2020 election, it shows 65% of voters say they are likely to vote for President Trump. It shows 75% plan on supporting Tom Cotton.

The likelihood both win in Arkansas is well within the poll’s margin of error of plus or minus 3.4%.

Parry said the Arkansas Poll is extremely reliable because they do things ethically. She gives advice on how to spot good polls and ones that are purposefully misleading. Within a click or two on the poll’s website, you should be able to find the answers to these three things:

  1. You should be able to tell how the sample was collected, whether by computers or live callers. Live callers are generally better.
  2.  You should be able to see how many men and women participated in the survey and how many people participated overall. Larger sample sizes are typically better.
  3. You should be able to see what the questions were, what order they were asked and what the options were.

Parry said polls are crucial. They do more than see what people think, they can influence it.

“Polls can hold politicians accountable for policies they pass and accurately reflect what people want,” she said. “But they can distort and skew who gets elected and who doesn’t as well.”

The Arkansas Poll didn’t look at the Arkansas District 2 race with French Hill and Joyce Elliott, but Parry said many local polls are showing less than a percentage point leaning favorable to Hill right now. She said that race could truly go either way because of the margin of error.

To see the results of the Arkansas Poll visit here