Save up to 98% of your content-consumption time.See how
International edition
Watch the source · CNN

What if the polls are wrong?

17:41946,808 views5 min read · 74% time saved

This source-linked analysis examines What if the polls are wrong? and connects its main claims to supporting evidence.

Explore topics and the full AI summary
  1. The speaker analyzes the accuracy of election polls, focusing on the persistent question, "What if the polls are wrong
  2. Republicans were underestimated by 3 points in 2002 post-9/11, 3 points in 2014, 3 points in 2016 (Trump’s surprising election), 5 points in 2020, and 3 points in 2024.
  3. Democrats were underestimated by 4 points during Clinton’s impeachment year in 1998 and by 3 points for Obama in 2012.
  4. In 2025 gubernatorial races, polls underestimated Democrats by an average of 6 points, including an 8-point underestimation in New Jersey and 5 points in Virginia.
  5. The accuracy of polls depends heavily on voter turnout and social trust factors that affect willingness to participate in polling. Turnout disparities
  6. The central argument is that polling sometimes underestimates one political party in a given election cycle, but these errors vary historically and between parties
  7. Polling errors sometimes balance out, such as in the 2018 Senate races with overestimation and underestimation canceling each other.
  8. A notable example cited is the 2024 Ohio Senate race, where the polls were tied but the Republican candidate won by 4 points, reflecting a 4-point underestimation of Republicans.
  9. Social trust surveys show Democrats are more likely than Republicans to trust others and pollsters, impacting response rates and poll accuracy.
  10. The speaker draws on multiple historical instances of polling errors