Generic Ballot

One Democratic talking point has been how Republican enthusiasm had peaked and that Democrats were getting more excited for the election. They pointed to Gallup’s congressional ballot tie last week. What they failed to mention was:

This was registered voters, rather than likely voters. It’s the actual voters that count, not the ones registered that don’t.
Gallup’s poll has varied wildly from week to week. Something atypical in generic ballots. This makes their results suspect.
Gallup is the only pollster showing this. Every other pollster has had an equal or higher generic ballot number.




I split out the generic ballots into three, showing one with likely voters, one with registered voters, and one combining the two. One thing the trend lines for registered and likely voters both have in common was a fairly steady ballot through April, an uptick for Democrats in May and June, a significant change to Republicans in July and August, and then a big increase in September.

While a couple of pollsters who measure registered voters and have been more favorable to Democrats have yet to weigh in this month there are more who have favored Republicans who haven’t either. Every pollster doesn’t poll every month, so they might not. (Note: This graph contains generic ballots for 20 pollsters. On average they’ve done 3.8 generic ballot tests over 9 months.) It seems likely September will end up with the same big jump it’s had so far.