I could write a long, involved piece on this issue. But Andrew Prokop, writing for vox.com, has saved me the trouble.
Here is his chart of how polls have misestimated senate candidates over the past 4 election cycles…and which party most often gets the short end of the stick:
![]()
See a trend there? In case you don’t, let me point out that there are 48 senate races listed, and 40 of the 48 – that’s 83% of them – underestimated the Republican vote.
Moreover, the 8 races in which Democrats were underestimated averaged a difference of just 1.8%. By contrast, the 40 races in which Republicans were underestimated averaged a difference of 5.0%. So not only were 83% of these races misestimated in favor of Democrats, but the extent of misestimation – 1.8% versus 5.0% – was almost three times as much. That is huge.
You might want to keep this in mind when you read some of the poll data being reported for the senate (and, presumably, the house and governor races as well) between now and election day.


