As anyone who reads this blog knows, I am very skeptical of political polls. All of them.
But I am particularly skeptical of generic polls, which measure how many people say they will vote for each party’s candidate.
My reason? The disparity between how many “safe” (strongly partisan) congressional districts there are for each party.
Why does this matter? Here’s an example that should give you the answer:
Suppose there are four districts roughly similar in voting population. Party A wins three of them by, say, 15% each – landslide victories. And Party B wins one of them by 50% – a huge victory, with a higher victory margin than the other three districts combined.
Because the one district Party B won in outdistances the other three combined, if a poll were conducted among all four, Party B would come out ahead. But in the actual races, Party B would have won three of the four districts.
Want another example? Okay. Suppose there are two districts, and one of them has twice as many voters as the other. If Party A wins the smaller district by the same percentage that Party B wins the larger district, generic polling for those districts would show Party B ahead (twice as many voters preferring it), but in the actual races it would be one district for each.
With the above in mind, I did a quick review of all the congressional races in 2022. To isolate districts that were obviously, decisively safe, I only looked at the ones with victory margins of at least 40%.
In my quick count (I might be off by one or two, but no more than that), Democrats won by 40% or more in 72 districts. Republican won by 40% or more in 41 districts.
A 72-41 “safe” district margin – almost two to one in favor of Democrats – ain’t what you’d call close.
From this I conclude that, with almost twice as many “safe” districts for Democrats as there are for Republicans, if the parties tied in a generic vote Republicans would probably win a large majority of the house.
It seems to me that for the Democrat Party to retake control of the house, it would have to win the generic vote by at least 3-4%, maybe even more.
And that, folks, is why generic poll results don’t mean much to me.
End of blog.
