Substack cometh, and lo it is good. (Pricing)

Kansas abortion proposition vs. Trump support 2020

No big surprises it seems. The r-squared of between 2020 election and 2022 ballot measure is ~0.80, so 80% of the variance in the second can be predicted by variance in the first.

9 thoughts on “Kansas abortion proposition vs. Trump support 2020

  1. The correlation/r-squared may be quite high, but what seems to have been surprising is the constant term: considerably higher (i.e., agin’ the constitutional amendment) than was expected before Tuesday night. So the redder that counties were in 2020, the less likely to vote against the amendment, but the 2022 vote was blue shifted relative to the 2020 vote. That is what determined the election outcome, no?

  2. Abortion deeply interests one wing of the Republican party, the Christian conservatives. Other Republicans don’t care about it much. We may be going back to a Republican party that looks more like it did in the 60s-70s on this one issue.

  3. @RazibKhan: yes. but i think a lot of republicans voted against it too

    My point exactly. That’s what I meant by “the constant term.” Eyeballing the graph:

    1) counties that went 80% for Trump in 2020 voted just now (on average) roughly 40% against the amendment.

    2) counties that went 60% for Trump in 2020 voted just now (on average) roughly 55%-60% against the amendment.

    3) a null set but reading off the fitted line*: counties that went 40% for Trump in 2020 voted just now (on average) roughly 70% against the amendment.

    The correlation/r-squared captures the distribution around the line: high values mean that the distribution about the line is quite narrow.

    The constant term captures the (equivalently horizontal or vertical) location of the line. That was what everyone (who found the result surprising) was surprised by. That is what indicates “a lot of republicans voted against it too.”

    *The fitted curve is non-linear (a lowess line?), but it is easier to refer to it as a line.

  4. Forgot to draw the conclusion from my points 1-3 above

    A regression on those 3 observations gives a slope coefficient of -0.75 and a constant term of 1.0. If (voters in) each set of counties had been 20% less likely to vote No on the amendment,

    1) 80% Trump, 20% No
    2) 60% Trump, 35-40% No
    3) 40% Trump, 50% No.

    the slope would have been the same, but the constant term would have been 0.81, and the amendment would have been the knife edge that all expected; IIRC, the result was 59% vs. 41% or thereabouts, so a 20%age point shift would have passed the amendment (if all that is needed is a simple majority).

  5. Correction: No need for a 20% shift, a 10% shift (-10% for No, +10% for yes) would have been sufficient (and the constant term would have moved from 1.0 to ~0.9). But my point about the constant terms stands (even if my seat-of-the-pants arithmetic sucks).

  6. I think a lot of Republicans voted against it but I don’t think they care too much about abortion either way. So in a referendum, they might vote for the pro-choice position. But if the legislature/judiciary were to ban abortion, those Republicans won’t necessarily switch to Democrats.

    The Pro-Life camp should learn from the gay marriage camp of 2000s and early 2010. How the public in state after state voted to ban gay marriage but the legislature and the judiciary conveniently ignored them. In those cases, the blacks were the swing voters. They didn’t like gay marriage but they didn’t hate it enough to switch to the GOP. The Pro-Life camp should learn from the tactics of the gay marriage advocates.

  7. Kansas Republicans didn’t want a constituitional amendment as worded in the ballots. It was that simple.
    This doesn’t mean that they don’t oppose abortion, rather the solution presented by the referendum.

    The explanatory statement on the ballot begins, “The Value Them Both Amendment would affirm there is no Kansas constitutional right” to abortion.

    This sounds like a blanket ban, but only as you read on is it clear that the law empowers the people, “through their elected state legislators, the right to pass laws to regulate abortion or to require the government funding of abortion, and would reserve to the people of Kansas, through their abortion, including, but not limited to, in circumstances of pregnancy resulting from rape or incest, or when necessary to save the life of the mother.”

    I can imagine many voters reading that first sentence and not even considering the rest. Contrary to media invective, most pro-life voters didn’t want a blanket ban… they wanted some exceptions.

  8. The large scatter is mostly in the high trump percentage area and includes counties with similar Trump vote percentages but differences by a factor of two in the abortion vote. It would be helpful to know what differences between those counties with similar Trump vote percentages is driving the scatter, but I don’t know Kansas county level demographics well enough to guess intelligently without a lot more research.

    A null model would be 100% non-Trump supporter against the amendment, 25% Trump voter against the amendment, but the reality is surely not that clean. Some non-Trump voters surely voted for it, and that probably reduced the overall percentage of Trump voters against it.

    In particular, it would be interesting to know if the non-Trump voter differences or the Trump voters differences between counties are driving the scatter and what kinds of Trump voters are more or less pro-choice within the GOP coalition in Kansas.

    I have no doubt that @HarryJecs is correct about the legislative v. ballot issue factors, but it would be interesting to better understand the GOP coalition’s factions in Kansas.

Comments are closed.