Ballotpedia's approach to covering polls

How does Ballotpedia...
...define candidacy?
...define incumbency?
...order candidate lists?
...choose battlegrounds?
...cover primaries?
...cover endorsements?
...cover campaign finance?
...cover polls?
...cover recall efforts?
...call an election?
...cover recounts?
...handle postponements?

Updated January 2022

Polls are meant to survey a representative sample of a population and provide valuable insight into popular opinion. They can do so accurately or inaccurately. This page outlines Ballotpedia's approach to covering election polls. We include poll reporting in our coverage of marquee races.

In 2022, we began relying primarily on aggregation from FiveThirtyEight and RealClearPolitics for our list of polls for elections involving candidates (not ballot measures). We report polls included in those sites' aggregation, where available. In addition to this criterion, we only report polls for which we can find a margin of error or credibility interval.

For races that have no available aggregation from either site, we rely on the following criteria for including polls:

• We avoid covering polls that do not disclose:
  • the questions asked
  • the results of the survey
  • the method in which it was conducted
  • the margin of error
• We include polls that use opt-in samples only if the results are weighted
• We avoid polls that do not report results in a timely manner.

Polls can also be used to influence opinion. They can be commissioned and released by candidates, campaigns, and satellite groups with a stake in the outcome. In order to provide readers with as much information as possible, we do report on internal polls that meet the criteria described above. For many primary elections, and elections in less populated areas, internal polls are the only polls available. Ballotpedia strives to identify internal polls as such, adding the name of the candidate who commissioned the survey to our tables and updates. We also include the names of other sponsors, such as satellite groups with a stake in the race.

Nathaniel Rakich of FiveThirtyEight wrote, "Usually, the goal in leaking an internal poll is simple: make your candidate look good. In practice, that means that internal polls typically paint a rosier picture for the candidate conducting them than exists in reality." If an internal poll shows the candidate who commissioned it performing worse than expected, the likelihood that the poll will be leaked or released may be impacted. Ballotpedia encourages readers to keep this in mind when viewing internal poll results.

Approach used from 2018-2021

For all elections that included poll coverage on Ballotpedia between 2018 and 2021, we did not use aggregation lists to determine the polls we covered and instead used the above approach outlined for races with no aggregation. We also excluded polls from our coverage that were conducted using primarily interactive voice response (IVR). We discontinued the latter policy in 2022.

Resources

To understand more about benefits and limitations of polling and how to interpret results, see these external resources:

• 2020 Pre-Election Polling: An Evaluation of the 2020 General Election Polls, American Association for Public Opinion Research, July 2021.
• How to Read Polls in 2020, FiveThirtyEight, June 15, 2020.
• A Field Guide to Polling: Election 2020 Edition, Pew Research Center, November 19, 2019.
• AP Style Guide: Polls and Surveys, Politico, accessed August 28, 2018.
• Polls Are Still As Accurate As They Were 75 Years Ago, Smithsonian Magazine, March 14, 2018.
• Maybe Election Polls Aren't Broken After All, Wired, March 12, 2018.

See also

• Scope
• Contact us
• What people are saying