Category Archives: 2020 election

post-election resolution #3: don’t argue on the basis of election counterfactuals

On Tuesday night, when things looked dimmest for Biden-Harris, I saw plenty of “Bernie would have won” tweets, and also some arguments that the Democrats would gave done better if they had been more moderate on immigration. On the GOP side, too, there will be arguments about whether the party would have performed better if all their candidates were more Trumpian or whether Trump cost them the White House despite strong economic fundamentals.

It is important to argue about: (a) what is valuable and (b) what is politically feasible. The union of those two circles is what we ought to support. People disagree about both questions, and discussion can be helpful.

Our values can influence our estimates of what would win. If you are further left, you may be biased to believe that leftist policies would win–and the same for people all across the spectrum.

However, disciplined political thinkers try to separate the two matters in case what they want may mislead them about what they can win.

It almost never helps to argue with counterfactuals about past elections. The problem with “Bernie would have won” is not that it’s false. It’s unfalsifiable, untestable. It provides no analytic clarity.

I acknowledge that social science often seeks to estimate counterfactuals. For instance, a regression model estimates what would happen to the dependent variable if the independent variable changed from what it actually is. (If we spent more on preventive healthcare, the data may suggest, fewer people would get sick.) This kind of reasoning is essential for thoughtful planning. However, to make counterfactual inferences, we need the right conditions: usually lots of cases, each with many variables, from which we can infer trends. Natural experiments also work nicely. A given election is one ambiguous datapoint that can fit countless theories.

So my resolution is: I will argue about what I value and whether it can win the next election, but not what would have won in 2020. I think that’s a recipe for confusion. In any case, I am not looking forward to that particular form of debate.

post-election resolutions #1 and #2 (less forecasting and less hobbyism)

Today seems an auspicious occasion to begin posting resolutions for becoming a better person and citizen after the 2020 election.

The first and second resolutions are simple:

  1. Less forecasting, more living in the present. Surveys are valid research tools, but they are particularly hard-pressed to predict future behavior with precision. The polling error that shows up in forecasting sites like FiveThirtyEight reflects the complexity of screening for likely voters–it doesn’t invalidate survey research. But why are we checking FiveThirtyEight in the first place? All the sages teach that we should live in the present or work to change the future. Forecasting violates that advice, but it is almost literally addictive: you get a little dose of pleasure every time a prediction is favorable, and when it isn’t, you can go back for another fix. I hope I can use the methodological limitations of electoral forecasting as a reason to pry myself away from the habit of forecasting everything (COVID-19, the stock market, my own life expectancy).
  2. Less “political hobbyism.People gave $100 million to Amy McGrath. In many cases, they were doing something to harm Mitch McConnell after he did or said something that made them mad. It didn’t work; he won by 20 points. One hundred million dollars is a lot of money. You could start a new college for that. Part of the problem is a profession–political consultancy–whose interests align poorly with the public interest. Someone made a fortune by fundraising for McGrath. (I wrote an article about this in 1994.) Expressing anger by giving money also has an allure; it’s an easy thing to do after observing something that makes you angry. I actually didn’t donate to Amy McGrath (ironically, I was too aware of the skeptical forecasts for her), but I exhibited other symptoms of political hobbyism in this cycle.

youth voting 2020: Tisch College analysis so far

This is your regular reminder to follow Tisch College’s CIRCLE (@civicyouth) for the best data and analysis on youth voting. A list of their recent releases follows. They will have lots of timely data as the actual election unfolds.

Brian Schaffner is also part of Tisch College. He co-leads the Cooperative Election Study (previously the CCES), which surveyed 71,789 people between Sept. 29th and Oct. 27th. (That is an enormous sample). His analysis of the likely voters in the CES shows why the youth vote is pivotal.

2020 CES Presidential vote preferences (likely voters)

Meanwhile, follow Tisch’s Institute for Democracy and Higher Education for detailed information on college students, our Metric Geometry and Gerrymandering Group for research on districting, and our Center for State Policy and Analysis for Massachusetts-related information, including work on the ranked-choice voting ballot initiative here.

how Trump and Biden supporters think about their identities

We surveyed a nationally representative sample of 1,267 Americans from May 29-June 10, 2020. Biden led by 50.5%-37%. That result is now far out of date. However, respondents answered more than 350 questions either in this survey or in surveys that they had taken previously. As a result, we are able to look at many more attributes of the electorate than in a typical political poll: matters like which chronic diseases Biden and Trump supporters have, whether they donate money to colleges, how they define their identity, and whether they have pets. (In case you’re wondering, Trump voters are 8 points more likely to have pets.) Given the remarkable stability of the race, I am confident these patterns still apply.

This new article on the Equity Research website provides a lot of detail. Here I’ll just share the portion about identity. All the statistics reported here are for likely voters: those who rated their own chances of voting in November 2020 as seven or higher on a 10-point scale.

Only 10% of whites said that race was important to their own identity (or “salient”). Among that group, Trump led by 61.5%-31%, whereas Trump’s lead among other whites was just 5 points (47%-42%–less than a majority).

More than half of African Americans (57%) said that their race was salient, and those African Americans supported Biden over Trump by 96%-3%. The margin was somewhat closer (80%-10%) among African Americans who did not report that race was salient

About one third (31%) of Latinos said that race was salient, and they favored Biden by 67%-23%. Among the majority of Latinos who said that race was not salient, Biden’s margin was somewhat narrower (60%-19%).

Biden led narrowly (48%-42%) among women who see gender as salient, but by a wide margin (52%-30%) among those who do not. Women who see gender as salient were also more likely to identify as conservative compared to women who do not see gender as important to their own identity (26% versus 18%). It appears that women who see gender as salient are more often expressing traditionalist views of gender rather than feminist views. Biden also performed better among men if they did not see gender as important to their identities.

Twenty-eight percent of voters reported that religion was important to their identity. Trump led 58%-31% among those voters, whereas Biden led by a similar margin, 58%-29.5%, among voters who did not see religion as salient.

As shown in Fig. 1, Biden led among all categories of people who felt that class was not important to their own identities, and especially among those in the lowest and highest income categories who felt that way. Trump led by substantial majorities among people who felt that class was important to their identities and were in either the low- or high-income brackets. In other words, Trump led among working-class people who identified with their class and among wealthy people who identified with their class.

More on the Equity Research website.

where youth will make the most difference

CIRCLE‘s Youth Electoral Significance Index (YESI) shows where young people are likely to have the most influence on the outcome of the election. The goal is to encourage investment in the youth vote, at least in those places.

CIRCLE has crunched the latest data to produce the final, revised YESI, with these changes since the last time:

  • Georgia: In our presidential rankings, several states moved up or down one or two spots and, notably, Georgia replaced Maine on the #10 slot. Georgia’s recent emergence as a battleground state also informed our Senate rankings. The state has two Senate races in 2020 and the one for the seat held by Kelly Loeffler was #8 in our earlier rankings; the updated version sees that race move up to #7 and the race for David Perdue’s seat enter the top-10 at #6. 
  • Alaska: The Alaska Senate race also makes it into our updated rankings (at #10), while Kansas and Alabama (#9 and #10, respectively, in the previous ranking), drop out of the top 10 entirely—though just barely, as they’re now ranked 11th and 12th.
  • House of Reprensetatives: In our ranking for U.S. House races, the Georgia 7th climbed from #5 to #3 and the Georgia 6th, absent from our previous ranking, is now the #10 race. The Utah 4th and the New Jersey 3rd drop out of the top 10, while the Virginia 7th enters at #8.

This map shows the Senate YESI, but click through for much more detail on the House, the Electoral College, and specific races.