Our old friend Ed Kilgore attempts to solve the puzzle presented by the difference between the special election results, which show a sharp move in the Democrats’ favor, and the comparatively weak advantage the Democrats enjoy in the generic congressional preference polls. While he doesn’t come down with a conclusive answer, he wisely suggests that the Democrats should be cautious about being overly optimistic about their chances in November.
One thing I think is not getting enough consideration in this debate, however, is the evidence that a lot of Trump voters are voting for Democrats in these special elections. We saw in Pennsylvania’s special election and again in Arizona’s. What we might expect to see is a big difference in turnout where Democrats are turning out in higher than expected numbers and Republicans are staying home. This would explain how districts that voted for Trump in huge numbers are now nail-biters or even Democratic pickups. But, at most, this only explains a fraction of the deterioration in the Republicans’ level of support in these recent special elections. A larger factor appears to be that a significant number of registered Republicans are turning out but not voting for the Republican candidate.
This contrasts with consistent polling that shows Trump retaining a high level of support from Republicans, so it needs some explaining. It could be an artifact of the difference between looking at how people are registered and how they are currently self-identifying. If you no longer consider yourself a Republican but are still registered as one, then there can be a disconnect between the poll results and how actual ballots are cast. But, in this case, these would be registered Republicans who voted Republican as recently as 2016. As they fall off as self-reporting Republicans, Trump is still able to retain a high level of support for those who still self-identify as Republican.
I suspect this explains some of the discrepancy we’re seeing. Actual elections are an infinitely better measure than opinion surveys of what the voting public actual thinks. And by that measure, the Republicans are losing support from within.
This matters because Ed is correct when he notes that turnout in November is likely to be older and whiter than in presidential years, as is always the case, and so we should expect some of the GOP slippage to be corrected for that reason alone. It will help the Republicans counteract at least some of their enthusiasm deficit. But if the Republicans who turn out don’t vote for the Republican candidates at their normal rate, that’s how a tsunami can occur.