Thinking Like A Strategist: What's Popular?

Immigration, executive actions, and more, from 272 issue questions

I spend a lot of time arguing that politicians should take popular positions. I’ve also written about how polling can give you the wrong idea about how popular something is. Generally, I think you need to look at a lot of high-quality polling to get a strong idea of how popular an issue is, and I realize it’s probably frustrating to see me gesture at “popularity” without linking the relevant polling.

However! This week, I’ve been given a gift. I was extremely delighted to see a round-up of issue polls on Trump’s agenda published by 538. There’s 272 separate questions in this dataset, across 22 categories. This is basically Christmas. All these polls are recent, from after Trump took office, and from trusted pollsters.

The toplines by issue presented in the article are (imo) slightly deceptive- they’ve categorized polls by issue type, and then run averages of net support across each category. This is probably useful for general comparisons to Trump’s approval (they note that support for his agenda is lower than support for him). However, there’s a huge amount of variation within each category, some more than others, so I don’t want people getting too hung up on one category being a few points higher than another. That’s useful, but it’s not complete.

Since we have this resource, I want to spend some time looking at the different types of questions in each category, and what they can tell us about overall voter preferences. I also want to do something of a worked example of how I personally go from “a pile of issue questions” to “a sense of how popular an issue is”. The dataset of questions is here if you want to dig in.

A note on coding

This dataset is explicitly “polls about Trump’s proposed policies”, which means that support/agreement/yes across all questions is going to be the Trump-aligned position. This is nice for comparison purposes, since respondents tend to bias towards agreement/yes answers, and having to reverse code a bunch of questions can make your results look weird. The questions also seem to be mostly grounded in things that have been actually proposed, rather than in hypotheticals. There’s a couple of “how would you feel if...” in there, but they’re not out of the realm of possibility. Very speculative questions can give weird results, since the respondent is first processing “do I think this would happen” before even getting to their opinion if it did.

All of that is to say, this is a less complicated set of questions to compare than a more wide ranging set of questions would be. I would love to see polling aggregators do more on tracking policy/issue questions, but I get that it’s hard and annoying. Hacking something together to do this is on my personal projects wishlist but so are about 100 other easier things, so that’s not happening anytime soon. Also, RIP Huffpost Pollster, gone but never forgotten.

Immigration

Immigration has the most polls of any category, at 63 total polls. This is the spread of net support across all those polls. While this does average out to a +2 net on average, there’s a lot happening under the hood!

Let’s start with that highest dot- a staggering +81 net. It is “Deport undocumented immigrants who have been accused of committing violent crime”, a question which is not duplicated elsewhere in the dataset. Deporting those accused of non-violent crimes like shoplifting is also pretty popular, at +28, but nowhere close to the violent crimes question.

The next couple high performers are more border questions than immigration questions- things like labeling drug cartels as “terrorist organizations” (rip to that news cycle, glad we’ve moved on from it), reinstating Remain in Mexico, and sending troops to the border. The troop questions specifically vary from asking about declaring a national emergency, sending troops to “secure” the border, using troops to stop migrants from crossing, etc, and vary from +11 to +28.

Deportations

What’s most unpopular? “Deport undocumented immigrants who arrived in the U.S. as children”, at -44. This is followed by similarly negative support for deporting undocumented people who are 10+ year residents of the US, and those who are the parents of citizen children.

General support for deportations varies a lot based on the details- In the *same* Ipsos poll, “deport all undocumented immigrants” is more popular (but not a lot, +6) than the questions about long term residents, children, and parents of citizen children, but less popular than deporting those accused of any type of crime, even nonviolent.

If people were answering surveys completely rationally, this doesn’t make sense. Those more sympathetic groups are a subset of “all undocumented immigrants”, but support for deporting them is far lower. Perhaps respondents can be forgiven for not being perfectly consistent, since the general question about “deport all” is in a different section than the series of questions explicitly asking about deporting different groups. Plus, if you look at the question text itself, it reads more as asking about support for the “effort” than support for actually accomplishing complete deportations. I think this is how respondents interpret most broadly stated questions, with people who feel positively towards the party pursuing the policy believing that any objections or issues will get worked out before implementation. Those opposing the policy are less charitable.

One other lesson from this is that Democrats probably shouldn’t fight on the grounds of not deporting undocumented people accused of violent crimes. Deporting those people is a very popular position, and not the angle to tackle the issue from. Admittedly, I think folks are focused on those who committed nonviolent crimes in their messaging, but even this is pretty difficult grounds. Deputizing local law enforcement to help with deportations is unpopular, so potentially there is some wedge space on the issue of local law enforcement cooperating with ICE.

Birthright Citizenship

A cleaner category is questions about the proposition to end Birthright Citizenship. Slightly cleaner- there’s still a lot going on. Is the proposal to remove birthright citizenship an executive order (between -7 and -1) or a constitutional amendment (-18))? Is it only ending birthright for the children of undocumented immigrants ( -17 or -7), or the children of illegal immigrants¹ (+6)? Is it targeting the children of foreign tourists (-6) or children of foreign workers and students (-21)?

I would read the compilation of evidence here as showing that ending birthright citizenship isn’t popular, and that it gets less popular when people start to think about the broad reaching effects. Similar to the results of the deportation questions above, this reflects voter preferences for dramatic-seeming statements, and distaste for actual action or consequences.

Do these theories around vague dramatic action polling better than specifics hold up in other categories?

Executive

This category is preeeeetty mushy- it seems to be “stuff that can be done by the executive not otherwise specified”. It ends up including a lot of federal agency workforce stuff and similar initiatives, while most of the Executive Orders are bucketed into other categories.

The biggest topic here is broadly “downsizing the government”, including specifics about firing federal workers, lowering standards for firing, etc. You could include the three questions about Return to Office policies for federal workers here, if you wanted.

Sidenote, on that weirdly high 3rd RTO number, that’s by TIPP insights. They once again have a slightly different wording of a question that gives them a much more R leaning result. In this case its “Require federal workers to return to the office full-time” at +33, in dramatic comparisons to an Ipsos question “Requiring all federal employees to work from an office rather than allowing remote work” at +2 and a similar DFP question that tacks on “or else be fired” at +1. They don’t seem to have included a generic Trump job approval question (or at least didn’t publish it). This isn’t necessarily wrong (maybe the Republican coding of their surveys is producing a more accurate result?) but it’s a little weird.

The high point in this set is “The federal government creating a council to investigate political bias at the Federal Emergency Management Agency (FEMA)” at +33, with the stronger “Trump has said he wants to dismantle FEMA, or the Federal Emergency Management Agency” at -10. I can’t really blame this on a preference for gestures over actions- probably people just think we need some form of FEMA.

Downsizing

Okay! Is downsizing the government popular?

Kinda? The straight ask of “downsizing the government” polls at +26, which is pretty impressive. The specific components of what it would mean to downsize do far less well. There are actually 3 questions specifically about a hiring freeze for the federal government/federal agencies. These come in at +0, +1, and +13 (from 3 different pollsters).

Buyout packages have mixed support, and straight up firing is generally unpopular (-17 per the Economist poll). There’s a little weirdness where “making it easier” to fire federal workers seems to be unpopular, but then TIPP has a +12 with “Make it easier to fire federal employees removing by removing job protections” that I don’t quite know what to make of. An Ipsos question of “Making it easier for presidents to fire longtime federal workers” hits -21. It’s possible that job protections are nebulous enough that people are more supportive? There might be a perception of waste/slacking/etc that isn’t captured by “longtime federal workers”. Or it might be TIPP being weird again, either way.

So....the government becoming downsized is popular, but the steps to get there aren’t. My theory about the American people wanting change but also nothing to happen stands.

LGBT Issues

This is, unsurprisingly, almost entirely trans issues, and one odd question about “Removing government websites that use the terms gender, racism, diversity, or bias” that I would probably bucket in with DEI instead (which I have dropped from the graph here)./

The topline given by 538 is +16 on Trump’s LGBT policies in general. This actually masks just how bad the numbers here are for Democrats- spoiler, they’re very bad.

I’ve once again added some subcategories. The only 3 polls where Trump’s position is underwater are all about banning trans people from serving in the US military. Banning transgender people from competing in women’s sports is popular, and has a broad range depending on question wording, from +27 to +38.

The category I’ve labeled “Biological sex” contains things like declaring there are only two genders, requiring the government to recognize only biological sex, etc. For a broad spread of question topics, I’m a little surprised the results are as narrow as they are (one Q on declaring there to be 2 genders is at +9, everything else is between +18 and +21).

I’ll be honest, I find this pretty depressing. Banning trans people from women’s sports is *profoundly* popular among Americans. There’s more wiggle room on the other questions (since I’m using net, you could point to varying don’t know/no opinion numbers as evidence of softer support, since a lot of these are more like 50/30 with a high DK), but it’s still not great. If I squint, I would probably say that the more intrusive into people’s lives a policy is, the less popular it is, but that’s not a very firm conclusion. This does not have the nature of the Immigration category (above), where when you start asking about specifics, support craters.

Of course, this situation is a bit different among Democrats. I couldn’t find crosstabs for all polls, but for that YouGov poll with the +38 number on sports, it’s 34-51 (-17) among Democrats, 64-22 (+42) among Independents, and 93-6 (+87) among Republicans. I wouldn't be surprised if this was the biggest D/R divergence in opinion on a question in this dataset.

Issues like this are really tricky for a party to handle. When something is popular among your base, and profoundly unpopular with the other party, it makes sense for advocates and electeds to consolidate around a position. However, the sports issue in particular isn’t staggeringly popular with Democrats, and it doesn’t help with reaching out to undecideds and independents. This is, frankly, rough.

I hate this! It sucks that public opinion is deeply against allowing what I would describe as a small number of mostly children to do sports. I realize that I am a huge liberal and not a sports person, which is why I have this opinion. However, I realize I am in the minority here. At this point, I don’t think Democrats making an issue of trans people in sports is helping anyone. If Dems are choosing what issues to fight Trump on (to the extent they can do that), or to tweet about, or give speeches about, I would recommend not picking this. I’m not saying people need to come out swinging for bans, but I would love to see a refocusing on helping those trans children as they get older and need things like jobs and healthcare. The ID questions in this polling are mushy and mixed with other issues, but I believe there’s probably popular support for allowing people to have IDs that match their gender presentation, and so on. There’s other ways to help people that aren’t so underwater, and I think that putting down the least popular aspect of the issue opens us up to do that.

This is probably not the most eloquent way of writing about confronting the unpopularity of issues you care about. I guess what I want to model is that 1. It sucks when public opinion is against you and 2. It’s still useful to be realistic about that.

Conclusions

The overall conclusion that 538 came to is that Trump’s agenda is mostly underwater, and I agree with that. On the immigration stuff, we can see evidence that implementation is unpopular and likely to get worse as specific stories come out. This is despite the fact that his bombastic general statements about “fixing immigration” seem to be popular. People also don’t seem to be on board with downsizing the government, or at least not with actually firing people. It’s early in the term, so we’re going to have to see how this all plays out, but I would read this polling as a generally good sign for Democrats that Trump isn’t achieving rampant popularity for all his various nonsense.

The warning I take from this is that there are aspects of Trump’s agenda and messaging that *are* popular, and that Democrats should try to avoid fighting on those. This is boring campaign advice, which most people already know, but talking about him cutting Social Security, Medicaid, Medicare, benefits for families, etc, is far more effective than talking about cuts to DEI programs. To restate my earlier point, the state of public opinion on trans issues is pretty grim for Democrats (and you know, for actual living trans people), and I think it might be a kindness to take the spotlight off them for a while. Especially while Democrats are the opposition, and don’t control any of the branches of government, our job is to oppose Trump where we can and take back power ASAP, so we can get back to helping people.

1. If you went “hm, is the poll that says ‘illegal’ perhaps run by a conservative pollster?”, you are right and should pat yourself on the back. It even feeds into a whole article touting the popularity of Trump’s EOs.