The Hard Calculations
There's little simple about campaign strategy
I’ve been explaining American democracy to folks a lot recently. U.S. elections are global stories, both because they’re dramatic events (if the United States remains tops at anything, it’s storytelling) and because everyone else has to live with the choices the American system makes.
One of my mantras about the relationship between U.S. elections and U.S. foreign policy has long been that foreign policy doesn’t matter to U.S. elections but U.S. elections matter for U.S. foreign policy. Despite what you hear about bipartisanship in foreign policy, make no mistake: on issues big and small, there is a difference between the two parties, even if from time to time the differences within the parties matter a lot too. (And even traditional redoubts of bipartisan legislative cooperation, for instance, are hardly immune from the calculus of partisan contention.) Yet when Americans decide how to vote, they mostly end up voting on domestic political issues and performance. That’s life in a hegemon: the feedback loops from foreign policy to domestic politics are weak enough that even though U.S. decisions have systemic effects, they’re resolved through processes driven by other factors.
I’ve laid that mantra aside for the last few weeks of this election, however. It turns out that the election is so close that, even though traditional foreign policy topics hardly rate compared to the economy or health care, specific issues might really matter electorally. The role of the Arab American vote in Michigan, brought to the fore by Israel’s conflicts after 10/7 and, specifically, by U.S. assistance to Israel in those conflicts, is almost certainly the clearest example.
(Let me caveat that discussion. One big foreign policy issue has clearly mattered: immigration and “the border”. I’m insistent in every other context that foreign policy has to include how the U.S. government treats all sorts of foreign actors, not just governments but corporations, terrorist groups, and, yes, individual migrants. But polling typically divides immigration from “foreign policy”, and so I think that survey data should be interpreted with that in mind. It’s also telling that you get radically different impressions of how much foreign policy matters when you ask whether several topics are important or not versus when you ask what is the most important issue. Pew, for instance, finds huge numbers of voters saying foreign policy—and immigration—matters, but Gallup finds that foreign policy net of immigration and terrorism ranks in the back half of issues by importance. When respondents are pressed to name their most important problem, the economy is the runaway favorite, followed by immigration and then in a tie for like ninth place comes foreign policy. So I feel pretty confident in saying that voters aren’t basing their choice on the Ukraine war or relations with Taiwan.)
What I want to do in this post is to think through the differences between how campaign stances tailored to attract a particular group, which cares dramatically, might influence other groups, which may not care but which are very large. I find that it’s very easy to assume that because a group is at the tipping point given all other factors being constant, it is therefore also easy to assume that all you have to do is change your policies to attract those voters. But this is too simple. Any stance a campaign adopts on a high-profile issue runs the risk of changing how other groups might vote, too, and thus of nullifying or even negating those benefits. To put it another way: the identity of pivotal voters isn’t fixed, but it’s determined by shifting factors. If it were easy to satisfy pivotal voters, then everyone would do it—but when you try, you’ll wind up creating some new group of pivotal voters instead.
My goal here, to be extremely clear, is not to make campaign strategy recommendations. I don’t have the skills to do that. Rather, my goal is to help people think through—systematically—why even a well-run campaign that is focused on electing its candidate might not take the obvious route. It’s not because they’re stupid, or blind, or craven. Rather, it could well be because they are aware that electoral politics is not as simple as hot takes and passionate advocacy want it to be.
As I mentioned above, one specific group that’s attracted a lot of attention in this election is Arab Americans, and foreign policy happens to matter a lot for them right now. It’s tough to poll small populations, especially when response rates for polls generally are headed toward zero, but the consistent finding from surveys and journalism is that Arab American voters are deeply angry at the Biden-Harris administration (although much more so at Biden than Harris). Further, this anger seems to be translating into a loss of support for Democrats among this group. Like, a big one—from 70 percent of this group usually voting Democrat to (let’s say) 40 percent voting for Harris.
In raw electoral terms, the effect isn’t huge on a national scale. Grossly, Arab Americans comprise about one percent of the electorate, so we are talking about approximately a quarter to a third of a percentage point in the national popular polls . You immediately want to respond, I’m sure, that every vote counts in an election this close, and yes, every vote absolutely does. But we all know that every vote doesn’t count equally because not every state counts equally. However, it turns out, there’s a large concentration of Arab American voters in Michigan—which you may have heard is a swing state.
The electoral importance of this group becomes apparent if you imagine them as potentially tipping the balance of Michigan, without which Harris appears to have very few paths to victory. If we take the high end estimates from the Arab American Institute, about 5 percent of Michigan’s population (and, let’s assume, of its electorate) is Arab American. Throughout this post, I’m going to use a lot of numbers, but I want to be very very clear that these are approximations—not least because, as I will stress, the population we want to know (the voters who will turn out in this election) is something that is unknowable and shaped by the campaigns and election, rather than being a target that’s relatively stable.
In my anecdotal experience, folks in places like Dearborn have strong family connections to Lebanon and Palestine/Gaza/the West Bank; certainly areas in suburban Detroit have also been quick to pass ceasefire resolutions in city and village councils. What is happening in the Middle East will often directly affect folks through family and other ties. It’s not just a CNN effect, that is, it’s something that’s really personal and really hard to address with platitudes. That is: let’s assume lip service won’t cut it.
(This observation of anger is not something limited to my experience. After I’d hit “schedule send” on this post, news came that the Arab American PAC, traditionally a backer of Democrats, has rejected endorsing both Trump and Harris on the grounds that “both candidates have endorsed genocide”.)
So let’s math out exactly how important this group could be to the Michigan, and thus the national, election. Roughly a thirty percentage point slide in Arab American votes for Michigan would translate to roughly a 1.5 percentage point decrease in Democratic vote share in Michigan. With the polls in Michigan essentially tied, it is not implausible to imagine that this could be the margin of victory.
You might wonder whether the argument for a full-on push to persuade those voters back wouldn’t be the obvious move for the Harris campaign. Here’s where things get complicated. First, the electorate isn’t static. Turnout of voting-aged population in Michigan varies dramatically, from a 20-year low of 58.2 percent in 2000 to a high of 70.5 percent in 2020. (That is, by the way, not far off from the French turnout in the 2022 elections, suggesting that “low American turnout” isn’t a national trait but rather a reflection of whether you live in a competitive state.)
Yet there’s variation within turnout as well. For instance, in recent midterm elections, Ohio has seen nonwhite voters turn out at very different rates—sometimes higher, sometimes lower—while Michigan has seen steady increases, with nonwhite voters even turning out more highly than white noncollege voters in 2022. Nationally, turnout for Black voters in presidential elections peaked in 2012 (62%) before falling to the mid to high 50s for the 2016 and 2020 elections.
If we stylize this as a five-percentage point shift in turnout for having a Black presidential candidate at the topic of the ticket, then it seems reasonable to assume that, with about 12 percent of the Michigan electorate being African-American, having Harris instead of Biden run could generate something like a half-point boost for the ticket relative to the counterfactual. (Things are more complicated than this, not least because the Harris campaign has sky-high support from Black women and well-reported troubles with Black men, but just stick with me for a moment.)
What does that mean for strategy? A small shift in turnout for a large group—much smaller than the thirty-point slide in Arab American support we are assuming, but with an electoral group that’s about two and a half times as large—has an effect size somewhere around a third of the impact of the change in Arab voting intentions. So shifts in turnout can counteract the electoral power of a group to some extent—crudely, you can make up yardage with a different play.
Now let’s extend that kind of analysis to another group: non-college whites. This analysis will be way more speculative, but again I am just trying to drive home the point that what might seem like small effects to one group could be, in principle, large enough in absolute terms to offset big effects for a smaller group.
NCWs remain the largest single voting group, even if they are diverse within themselves. Let’s stipulate that about half of Michigan’s voters are non-college whites—we are interested in gross sizes here, and also it makes the math really convenient since that’s roundly ten times as many voters as Arab American voters.
If the Harris campaign went all-in on protecting Palestinian (and Lebanese!) rights, we might well assume that other groups would notice and adjust their views accordingly. Now, we’ve already established that most groups don’t care that much about foreign policy issues. When a group does care they can become something close to single-issue voters on this. (A ten-point change in a group’s voting behavior is massive; a twenty or thirty point change is tectonic.) But that doesn’t mean that other groups won’t notice changing messages.
Israel has something like a ten-percentage point net positive approval with non-college Americans (smaller than its net positives with college and advanced degree holders!), and approval of Israel’s current wars is something like 50-50 (depending strikingly on partisanship). All of this together says to me that a reasonable estimate is that not many non-college whites would change their votes based on a changing stance on Israel …
… but the group is so large and sufficiently positively disposed to Israel that even a small change (let’s say three percent) would end up having an effect on the total vote outcomes that it would equal that of the change in the Arab American vote.
Let me reiterate here that a) I don’t have any special polling data here and b) I am only making a point about how small shifts in big group behaviors could nullify big shifts in small group behaviors. Would a full-on shift regarding Palestinian policy be that important to another group? I don’t know, although I do have to think that it’s within the realm of possibility—especially because there’d be other campaigns and groups amplifying any such shift.
My point is more general: it could be that your favorite political strategem (of which this just happens to be a prominent example) isn’t being rejected because it’s stupid but because the situation actually could be complicated. As a corollary: the intuitive idea that big groups matter more in elections does, in fact, sometimes happen to be true—there’s as much at stake in a one-percentage point change in how NCWs vote than a five-percentage point change in how Black voters do. And at a still more meta level: a lot of the popular discussion about elections takes place in big, blunt categorical terms, but a lot of what matters to campaigns takes place at the margins—rates of change, turnout rates, and so on that might not look like much but which consume the efforts of campaign professionals.
There’s lots of caveats you could make to this back-of-a-small-envelope calculus. If the population of Michigan’s Arab American voters is smaller than 5 percent (and those numbers run well ahead of Census figures, although Census estimates for MENA populations are notoriously skewed), then the electoral weight of that group would decline similarly.
Contrariwise, other groups could also welcome a shift toward a less pro-Israel stance, amplifying the effect. Younger voters do not share older generations’ support for Israel, so there might be ground to be gained there. On the other hand, younger folks turn out less than older folks, who tend to be more supportive of Israel, so the net assessment there has to seem like something like a wash.
Further, I’ve caricatured these choices as being implicitly something like “all for Israel” or “from the river to the sea”—but obviously any politician worth their salt could find an intermediate solution (like playing a diplomatic role) that would be more popular—and, incidentally, speak to Trump’s weaknesses (not a diplomat, that guy). But would that really be enough to satisfy people whose social media feeds are full of their family members being killed? (Note that this description also applies to not a few folks in Michigan with connections to Israel! And even people without such direct connections may be feeling like they’re being targeted through different channels.)
Yet on still another hand (I think we’re up to something like the fifth or sixth hand here), the Harris campaign isn’t running just in Michigan. Any message they put out there will also be heard—one way or another—in Pennsylvania, Arizona, Wisconsin, and elsewhere. Even if Harris can’t win without Michigan, the campaign might well calculate that the effects of messaging on the other six swing states’ electorates might be too risky, or too certain, to go big on the issue.
It’s deeply frustrating to hear that things are complicated and change is hard. (It’s also not the path to profitable Substacking.) Nevertheless, I think this exercise is valuable because anyone opining about campaign strategy should remember that campaigns exist to win elections. Activists always, always lobby people to believe that the only road to winning is to adopt their favored stance, but going all-in on any group’s favored policy ultimately means giving others less of what they want—and that has effects of its own.
If the solution isn’t obvious, the problem remains so. This issue is a tough one—in many dimensions—for Harris. She’s not in charge of the foreign policy decisions that are driving dissatisfaction with the ticket. The campaign faces lots of countervailing pressures that might not be obvious to people who don’t experience contact with strongly, or even weakly, pro-Israeli opinions. That, in turn, drives frustration among voters who do think the issues are simple—why won’t the candidate do this one obvious thing? And in a competitive communications environment, it’s not like changing a campaign’s strategy ends the debate. The enemy gets a vote; softening support for Israel might invite vicious attacks.
Given all of this, what should you expect to see the Harris campaign do? Well, if you can’t clearly win on an issue, work to win on other issues. Work to drive higher turnout from other communities. Emphasize abortion rights, the economy, and broadly appealing issue. (Hey, look what she’s talking about in rallies in Flint!) At the same time, reach out to the affected community directly, and quietly, to discuss the issue and express concerns. (Folks being outreached aren’t always thrilled by the attempt, however.) Strictly as a campaign calculation, even if the campaign can’t win back all the votes, there’s a big difference between a 30-point change and a 15-point one. (That Trump has been somewhat calibrated in his public statements about the wars means it’s a little harder to have an outside group run ads blasting him as opposed to peace, even if I’m fairly sure that his vision for the region runs closer to Likud’s.1)
So where does all this leave us? It’s still plausible to me that Michigan’s election might turn on the electoral consequences of U.S. support for Israel. It’s not clear to me that the seemingly most obvious response to this would be the right call for the Harris campaign to make. But it is clear that the other behaviors and strategies we see are consistent with a campaign trying to use other levers to win.
Although this is something where a second Trump term could produce outcomes from full-on hardliner to something a little more unpredictable. He does actually talk to other voices from around the region, and he does have something more than just a kneejerk reaction to foreign policy. In this case, it’s the unpredictability that makes it harder to discuss, but the range of unpredictable outcomes involves some really potentially gruesome and dangerous stuff.