Skip to content

Update: Party defections are tiny and mostly meaningless

A few days ago I wrote about an AP study showing that over the past year about 1 million voters had switched their party registration to Republican compared to 600,000 who had become Democrats.

This was based on a party switching report from L2, which combined actual data from states that report it with modeling from states that don't. Today, a competing firm, Catalist, claims that L2 is wrong: using its own actual data from all states—no modeling!—party defections have been about equal.

I can't say for sure who's right, but there are about 170 million registered voters in America. The L2 study shows a net difference in defections of about 400,000, which amounts to 0.2% of all registered voters. The Catalist report, depending on how you read it, shows a net difference of 0.1% or less.

In other words, (a) the two reports aren't really all that different, and (b) the numbers are too small to be meaningful anyway—which should have been my conclusion from the start. AP's big problem wasn't so much that they used terrible data from L2, but that they drew sweeping conclusions from tiny net movements.

13 thoughts on “Update: Party defections are tiny and mostly meaningless

  1. Pingback: The Republican Party is picking off Democrats – Kevin Drum

  2. D_Ohrk_E1

    Some states have public data, in aggregate, of voter registrations, while others require you to aggregate yourself after you pay for full voter roll data. I'd give more credence to the firm that paid to access and aggregate all the data over modeling.

  3. arghasnarg

    Another point is to the extent that I'm hearing about defections, they're strategic, attempting to influence primaries.

    I mean, there's chatter about that most elections. But there's been a lot this time, especially in places like Arizona and NC.

  4. bebopman

    “…. they drew sweeping conclusions from tiny net movements.“

    Yeah, I think I’d worked at a place (or 2) where that was above the masthead.

  5. Special Newb

    Also a number of these switchers where already voting for their new party they just switched their official label.

  6. Austin

    400,000 is enough to swing 5 states, if that 400,000 breaks the right way in PA, WI, MI, AZ and GA.

    I agree it’s probably meaningless because the number is aggregated over all 50 states… but it wouldn’t take much to tip those 5 states in particular from one party to the other.

  7. Spadesofgrey

    Last notable surge out of the Democratic party was during 2002-4 period when a bunch of previous democrats left the party(including several famous ones) coalition. It's a large reason despite Bush's general unpopularity, the party struggled both those elections.

  8. name99

    Is the complaint about the precise numbers, or about whether the trend is/could be happening?

    As I've said repeatedly, those who know only the present don't even know the present. There's plenty of interest in all of Rick Perlstein's books, but Reaganland (covering the late 70s) is especially interesting in how much (to me) it feels like our current moment; right up to the pundits insisting on what "Americans" think and how "Americans" feel, even as they are utterly clueless as to what half the country thinks and feels.
    Or to put it differently, if you predicted Trump would lose, and predicted Brexit would lose, then perhaps your radar as to what "the public" are doing all across this great country is not as finely tuned as you imagine? Might I remind you, as one immediate data point, that Chelsea Boudin was in fact recalled? And that all the complaining in the world about how this was led by Republicans, was racist, was blah blah blah, did not in fact change that fact?

    Or to put it differently, widespread agreement that Nixon broke the law did NOT translate into Carterland; that revulsion only lasted one election before larger cultural discontent kicked in (to me it looks like SF today feels like Travis Bickle's New York, and will have the same cultural resonances and consequences).
    Likewise, Trump and cronies can be found guilty of something related to 1/6, hell Trump can even land up in prison, but that won't necessarily translate into a second Biden victory or a wave of Democrats in Congress.
    "The People" are generally better than pundits at distinguishing between
    - someone truly evil,
    - a scoundrel, but our scoundrel, and
    - the fact that he agrees with some of Trump's positions doesn't make him Trump2.

  9. galanx

    Wasn't it Kevin that was panicking over this, and demanding that Democrats jettison all those icky progressive principles?

Comments are closed.