Are Trump's Indictments ACTUALLY Helping Him in the Primaries?
Plus the multiple ways that waving is associated with higher wellbeing.
Hello friends—Happy Monday! It’s August 23, 2023: Welcome to Pulse of the Polis 19.
Let’s get into some social science!
I’ve got 2 projects for you today
Are Trump’s Indictments Rallying His Base? Evidence from the Counterfactual Format | White Paper
Among the many, many, many monikers former-President Donald Trump has picked up is “Teflon Don”1. Trump has been at the center of a stupefying number of political controversies over the last few years. It is literally impossible to remember all of them without the aid of a database but, incredibly, Trump not only remains a viable political candidate but is (at time of writing) the front-runner for the 2024 Republican Presidential nomination. One of the more serious controversies is, you know, the whole “holding onto and flagrantly sharing top secret information” thing, constituting 40 of his 91 (current) criminal indictments. (Though not the charges that resulted in the photograph at the head of this newsletter.) Despite the fact that committing crimes pertaining to top secret information is, ya know, generally frowned upon: recent polling suggest that this has made a substantial number of Republicans more likely to vote for him. This coheres with a broader press narrative that urged substantial caution with pursuing criminal charges for Trump in the first place: prosecution, these folks averred, would only serve to energize his most ardent supporters.
But is this actually the case? A recent preprint from Soubhik Barari, Alexander Coppock, Matthew Graham, and Zoe Padgett claim that Trump’s indictments have hurt his electoral prospects after all. They argue that the finding that some Republicans are more likely to support him is an artifact of the way that the question is posed. When seeing this type of more/less likely to support question, respondents frequently substitute this question to one that signals their support more generally. Meaning people who say “more/less” likely to support are using these responses simply to say that they “do/don’t support” the person/topic overall.
Instead of relying on this question type, the researchers conducted a survey of approximately 5,000 people with SurveyMonkey. Half of the respondents were asked about their support for Trump2 and belief that he improperly disclosed nuclear secrets using the prevailing “more/less” question type. The other half were asked about these topics using “counterfactual” questions. That is, they first informed the respondent about the issue at hand and then asks if, in a counterfactual world, would be more/less likely to, e.g., support Trump had they not heard of the information. They found that, when asked using the “more/less” question type, Republican respondents reported being more likely to support Trump. However, using the counterfactual style of question, they find that support for Trump dampens, albeit marginally (by about 1 percentage point for electoral support). However, these more modest changes are what one would expect given the existing literature on political controversy as well as the fact that opinions about Trump is largely baked-in. The authors conclude that this suggests that response substitution is likely playing a role in folks’ responses on these important topics.
I like this paper a lot. I was initially more skeptical about whether the counterfactual framings were going to be effective because most people’s projections about hypothetical realities are almost always going to be contaminated with the knowledge and emotion they carry with them in this one. But this image shows that a decent proportion of the sample is approaching the hypothetical as one would expect: Learning damaging information should be, well, damaging. This leads me to a wonderful array of follow-up questions that I would love to see future work tackle such as: Are these people less interested in politics, less attached to the GOP and/or Trump, less likely to consume partisan news sources, etc. The paper shows a lot of little details which convey a great deal of thought and care on the topic. I think that this is a very important methodological contribution to this general topic and that survey-writers should take note.
That said, there are a couple comments that I’d still like to make. First, I don’t entirely buy that this type of question is immune to response substitution effects. Don’t get me wrong: judging by its coherence to other literature3, I’m fully on-board with saying that it’s better than what usually gets used. But I think that many people will still answer that the additional information didn’t affect their position very much as a means of signaling their support/disdain more generally. Second, piggy-backing off the first, I’d love to know how much of this comes from removing the substitution effect but, unfortunately, the research wasn’t designed for that particular question4.
If there’s one thing I wish was asked that wasn’t, it’s whether people would be more/less likely to vote overall. Not just vote for Trump in the primary, but actually bother to cast a ballot in the first place. I think that there’s some contingent of possible GOP voters (including leaners and independents) who are largely turned-off by Trump and would simply stay home if the choice is round two of him v. Biden. But I think that there’s also a contingent who are downright pissed that Trump is being charged; and anger, coupled with identity, catalyzes action. In the previous “more/less likely” questions, people answering “more likely” might not be taking “support” to mean their affect towards Trump, the individual, has grown. Rather, “more likely to support” might mean that they feel this development makes them more animated towards acts of support, e.g., voting. So I would personally love to see future research investigate which question better gets at changes in eventual action. Though it may not be possible to test it on this particular topic. Because, after all, the election is still over a year away5.
Saying Hello Linked to Higher Wellbeing, but With Limits| Gallup
This study of approximately 4,500 Americans looked to see how greeting one’s neighbors is associated with Gallup’s Health and Well-Being Index. The index is a 0-100 scale calculated by considering items from five dimensions of wellbeing: Career, Social, Financial, Physical, and Community. Those who regularly greeted zero neighbors on average scored 51.5 on the Well-Being index. Those who scored 6 scored 64.1, with little difference from there onward up through 16 or more. Interestingly though, the increases were not just seen in the social and community dimensions. This pattern of gain-up-through-about-6-people-then-plateau held for all five dimensions of health when stratified by how many people respondents tended to greet. Older adults, parents of children under 18, and those with higher household incomes tended to greet more people than those who were younger (6.5 vs 2.9 for 18-29 year olds), non-parents (6.1 vs 4.9), and lower-income (6.4 vs 4.6 for those making under 24,000 a year). The report argues (in so many words) that few actions linked to wellbeing will only be associated with a single dimension and often will have multiple simultaneous causes at play. Here, for example, wealthy people will tend to not only have more time to meet people (and thus wave to more people) but, in a services oriented society such as the US, being connected to more people likely leads to increased financial opportunities. Waving, then, may itself drive some factors of health, be a consequence of others, and simultaneously both driver and outcome for others still.
One thing that I really respect about this analysis is that the authors are careful to explain that the causal link between waving hello and general health and wellbeing is multifaceted and reciprocal. One might not necessarily expect that waving hello would be correlated to better physical health6 until you remember that those in poor physical health are likely not as able to get out much; consequently, they systematically lack opportunities to wave to others in the first place. General wellness is not so much a single thing but a web of factors feeding back on each other.
It’d be easy to toss-up the associations and let readers and the press go wild with it. (Physical trainers hate this one trick at having better health in SECONDS a day). But they took the route of good, conscientious social scientists and made plain the reasons why the obvious conclusion is likely not the correct one. It’s not that, like, I expected less from Gallup (or anyone else in particular) but good social science practice requires effort and diligence. And I would prefer to not just admonish people for failing to uphold the standards but praise folks for doing so and leading by example.
I’m not all that surprised that parents of children under the age of 18 report higher average numbers of waving than those without kids. For one, kids are often friends with other kids in the neighborhood meaning that the parents probably know each other and are on friendly terms. Second, if you have young kids, there are great odds that someone’s going to stop you to baby-talk your kid. So, congratulations, you’ve scored a new waving buddy. I also don’t find it surprising that age and income are positively associated with waving; those who have more leisure time either due to their life stage (e.g., semi or full-on retirement) or due to their income probably will have more time to bump into people and make casual acquaintances.
Commenting on their categories of general health for a second, there’s one category that I’m surprised is missing: spiritual. I know that the prevailing view is that America is sprinting full-bore towards areligious secularism, but the fact of the matter is that religion and spirituality is at least somewhat important for nearly 3/4s of the country—according to Gallup’s own polling on the issue! The importance of faith can’t just be chalked up to being a vehicle for feelings of community or social wellbeing. Spiritual connection, feeling attached to something that supersedes normal comprehension, is valuable to millions of Americans in its own right! My hunch for why that dimension isn’t included is that one could technically lead a contented life without it, so it would be difficult (though not impossible) to make a scale that doesn’t say that one is deficient in “well-being” simply because they are not religious. I’d love for future work to have that in there though because, for many Americans, faith is probably just as important as any of the other sources of wellbeing.
One demographic category I’m unironically interested in seeing broken-down is dog owners. I mentioned earlier that the fact that parents know more kids makes a lot of intuitive sense. But I not only have a toddler, I also have a dog---and I expect that the latter would have a similar effect on number-of-people-waved-to as the former and for similar reasons. If you go outside because you’ve got a dog to walk, you’re liable to encounter other folks who are out-and-about because they’ve also got a dog to walk. So not only are you seeing the same person a few times a week just by happenstance (and our social wiring makes it feel super awkward to let that go mutually unacknowledged for some reason), but the dogs are likely going to want to make each others’ acquaintance which will come with the bonus of a new acquaintance for the people walking them. Not to mention the fact that many of those who are out walking without dogs still like dogs and will find any excuse in the world to pet one. So, congrats, you’ve got a new soft-tie there too! Which makes me wonder whether or not dog ownership can actually engender more social capital…7
In the general for Democrats and Independents, in the Primaries for Republicans.
G. Elliot Morris of FiveThirtyEight shows that Trump’s approval ratings have taken a hit of late, arguably due to the mass of indictments, so there’s even more coherence there.
Generally, substitution effects are dampened by providing respondents with an outlet for the question that they want to answer: namely a “do you support/oppose XYZ” prior to (or on the same page as) the questions asking about change in their position. If one of their experimental conditions included this sort of structure in addition to the two conditions they tested, they could compare how closely the “dampened” condition performed compared to the counterfactual.
Christ, this is going to be a long 16 months.
It’s not like waving burns all that many calories.
Congratulations, you just watched the inception of one of my several half-baked research ideas.