When Evaluating Public Support for Immigration Ban, Look Beyond National Polling

Over at fivethirtyeight.com this weekend, Harry Enten examined public opinion relevant to President Trump's executive order banning immigration from a number of predominantly Muslim countries. In short, Enten finds national opinion closely divided on a range of items that might inform how we handicap and understand the public's response to Trump's order, but also urges caution, both rightly and wrongly concluding that we don't really know much yet. Enten's error of interpretation can be placed squarely under the heading of lessons still not learned (both from the 2016 campaign, and in interpreting Trump's actions). In short, if one examines the attitudes held among the voters that Trump's executive action is intended to excite, either regionally and/or within the broader national polling data hidden within the crosstabs, the results are far from ambiguous.

But first, let's highlight where Enten is correct:

Slight differences in framing and question wording can also have big effects on how well immigration, refugee and terrorism policies poll. Whether Trump’s executive order is viewed in humanitarian terms or (as the Trump administration has tried to frame it) in the context of counterterrorism could go a long way toward determining how much the public supports it.

He's absolutely right about this (and "this" could include questions about anything, by the way). Question wording will influence question results. Further, how Trump's actions are framed and continue to be framed will matter, especially for those among the public who don't yet have strong pre-existing attitudes about these issues (by framing, I'm referring to the way an issue is talked about so as to help people to place it in a broader context, an example follows). If one is primed to think about Trump's ban in humanitarian terms (by putting his order in the context of Syrian refugees, for example), then one is also likely to consider questions of morality and responsibility to individuals fleeing war torn countries who are seeking refuge when asked about policies that might affect them. If one is primed to think about the order in terms of terrorism and as a way to bolster national security, then one might be more likely to view the order as appropriate, if potentially impractical.

But here's where Enten may be overstating the case: we know a lot more about relevant public attitudes if we dig beyond national polling results. And looking beyond national polling results makes a lot of sense if we recall the seeming narrow focus of Trump's campaign, his subsequent "thank you" tour to rural areas in specific regions after his election, and, if one is willing to extrapolate, his first actions while in office. The shock among political analysts and commentators on election night was partially a reflection of elite bubbles and some poor statewide polling, but also in response to expectations set up by national polling numbers that showed Hillary Clinton holding onto a healthy lead (better than the one Barack Obama held over Mitt Romney in 2012), and one that appears to have been accurately reflected in the final popular vote tally (the grumbling of sore winners notwithstanding).

This last point raises the question of how and whether we should interpret national polling data in the age of Trump as bearing on his actions in an immediate sense (this is not to say that national polling won't be interpreted and reacted to by other actors in the system – namely in Congress – in important ways, but we're talking about the public in this discussion). National polling reflects a – usually – regionally stratified sampling of individuals. So let's do a simple thought experiment about who these individuals are with respect to the 2016 election and any recent national polling. Among the 50 states and the District of Columbia, 30 gave their electoral votes to Trump, 21 gave their electoral votes to Clinton. The total population, according to 2016 estimates from the U.S. Census Bureau in states where Trump was the victor was greater than Clinton's  by approximately 42 million people. But if one accounts for the share of the vote that each received in those states and divides their populations accordingly, the weighted population of Clinton voters to Trump voters is greater by approximately 6 million people. Translated into percentage terms, this would amount to a 2 point Clinton advantage, the same as her final lead in the popular vote. So is it shocking to show narrowly divided opinion on an issue that was, if not front and center, rather prevalent throughout the 2016 campaign, and that starkly divided the two candidates? Clearly not.

So how relevant a universe is national polling if one is trying to understand Trump's decisions and a public response that he might react to? As many have noted, Trump's transition was notable for his seeming disinterest in reaching out to those who didn't vote for him, with the final punctuation delivered in his inauguration speech. Given this, it seems as though there is little that we an learn about Trump's actions from national polling numbers that reflect a populace among whom a majority voted for his rival – and that he's already sought to discredit when those results go against his own actions or preferences.

Let's dig deeper by taking Texas as an example of a state that Trump won. Texas is a good comparison point in that its voters as a whole, and its Republicans in particular, probably look a lot more like Trump's supporters in those states key to his victory, as well as other states in the U.S. that Trump won than do topline results from national polls. This is both a conservative and a liberal example (in non-ideological terms). Conservative in that Trump did not win the primary here, and some share of Texas Republicans are likely more attuned to the traditional Republican message than Trump's brand of Republicanism. More liberal in that Texas is a Republican state, and likely deviates from the states he won by slim majorities in the rust belt that proved so critical. So what do we know about how Texans might react to Trump's most recent actions?

Here, there is little ambiguity. A majority of Texas voters overall, and more than three-quarters of Republicans didn't think that Texas should accept Syrian refugees in October 2016. Given how much that country has been in the news of late, and in particular with that coverage focusing on the atrocities meted out on civilians, it's hard to imagine that attitudes towards Iraq, Iran, Libya, Sudan, Yemen, and Somalia are likely to be significantly different.

Loading chart...
Should accept36%
Should not accept50%
Don't know/No opinion14%

Loading chart...
Should accept63%28%12%
Should not accept20%57%77%
Don't know/No opinion17%15%10%

But let's say that the question, with its limitation to Syria provides one with a reasonable degree of skepticism. That argument would go something like this: with six more countries added to the ban, and further, a ban that is national in focus as opposed to applying solely to Texas, maybe Texans would express more concern. This is certainly plausible. But we've also asked other, more severe questions based on candidate Trump's proposals. With respect to religiously targeted bans, Enten notes:

Just 41 percent of Americans supported a temporary ban on Muslims entering the country who are not U.S. citizens, according to an August 2016 ABC News/Washington Post poll. A slight majority (52 percent) were opposed. A July CBS News/New York Times survey, which asked a similar question, found only 35 percent of voters thought the U.S. should temporarily ban Muslim immigration.

Not so in Texas, where in June 2016, a majority of voters expressed support for a ban on Muslims entering the United States, including 76 percent of Republicans. And this was not a temporary ban either, the question asked, "Would you support or oppose banning Muslims who are not U.S. citizens from entering the United States?"

Loading chart...
Strongly support31%
Somewhat support22%
Somewhat oppose15%
Strongly oppose22%
Don't know10%

Loading chart...
Strongly support13%31%48%
Somewhat support13%21%28%
Somewhat oppose23%15%10%
Strongly oppose44%19%6%
Don't know8%14%8%

While it's abundantly clear that the media and the public is still in the earliest phases of understanding the Trump presidency, one hypothesis might help in shedding light on Trump's words and his actions: though he may like to lay claim to a popular mandate and popular support, even in the face of contrary evidence, when he speaks (and acts), what matters is those who have and continue to agree with him. Take this illustrative comment from an interview last week as the President was discussing his allegations of mass voter fraud:

“Well, let me just tell you, you know what's important? Millions of people agree with me when I say that,” Trump said. “If you would have looked on one of the other networks and all of the people that were calling in, they're saying, 'We agree with Mr. Trump. We agree.' They're very smart people.”

So while we may not know much about the broad public reaction to Trump's actions on immigration, nor on other high profile matters likely coming in the days and weeks ahead, we should keep in mind that Trump’s target audience is much narrower than the public writ large, and if we want to understand the public Trump is responding to, we better get a handle on who they are, and what they think, because that may be the best road map available for what's to come.