CNN made changes to their polling methodology after their 2020 polls “missed the mark,” the network announced Friday.
The network said that they made the changes to account for political differences between groups who do and do not respond to polls, which they say was likely the cause of polling errors in 2020. The 2020 polls failed to accurately assess support for Republican candidates across the board, CNN reported.
“Assessments of what went wrong suggest that the problem derived from differential non-response – that is, that the people who took surveys were not the same politically as those who did not,” CNN Polling Director Jennifer Agiesta wrote.
Friday’s poll assessing President Joe Biden’s approval rating was the first poll released that was generated using the new methodology. It found that 52% of Americans approved of Biden’s performance and 48% did not – an approval rating that is much higher than other recent polls have found. (RELATED: President Biden’s Approval Rating Sinks To New Low: Poll)
new @CNN poll on Biden’s job performance:
— John Harwood (@JohnJHarwood) September 10, 2021
CNN and the survey research firm SSRS will now conduct polls over a longer period of time using a larger sample than is usually used for news polling. They will contact a group of respondents randomly selected by mailing address and use that group’s responses to “measure long-established trends” like presidential approval ratings. The survey will also measure Americans’ party identification and other political traits.
Instead of selecting a random sample of respondents based on geographic location, CNN and SSRS will randomly select from an established sample of addresses that includes characteristics about where the respondents live. CNN will “increase outreach to underperforming segments of the sample,” the outlet said.