While these figures are informative, they still leave us with two core questions: 1) How would American Muslims choose between candidates when the crisis in Gaza is just one of several considerations?; and 2) To what extent do a candidate’s policy stances towards Israel factor into these voters’ calculations?
To partially plug these gaps in our knowledge, we embedded an experimental component within the MAAS. Specifically, we ran a conjoint experiment (n=1183), commonly used in marketing to assess the relative importance of a particular component (say, price or some technical feature) for a consumer’s willingness to purchase an item. In this instance, we assessed the voters’ willingness to cast a ballot between two hypothetical congressional candidates while varying the following components:
Candidate race/ethnicity: White, Middle Eastern, Black
Candidate political party: Democrat, Republican
Candidate religion: Atheist, Christian, Jewish, Muslim
Policy: Ban LGBTQ material in K-12 education: Opposes ban, Supports ban
Policy: Level of climate action needed: No action, Moderate action, Aggressive action
Policy: US aid to Israel: Maintain aid, Condition aid, Freeze aid
Figure 3 is an example of how these factors were presented to the survey’s respondents:
Figure 3: Example of conjoint experiment as presented to respondents
The goal with this approach is to mimic the decisions that voters could potentially face and isolate the effect of any one trait or component in driving their ultimate choice. This experimental assessment also gets us closer to revealed preferences (what people would actually do in a real world scenario) rather than stated preferences (what people merely say they would do). Another notable advantage of conjoint analysis is that it allows researchers to test the impact of multiple variables by having each respondent make several choices between two options (in our case, each respondent chose between two candidates six times).
Figure 4 presents the results of the conjoint experiment and highlights one more benefit to this method: clear interpretability. The point estimates denote the change in probability of a given respondent choosing a candidate with that component (whether a personal trait or policy position) relative to a reference component (which are zeroed out in the plot for clarity). The reference for demographics is White for race, Democrat for political party, and Atheist for religion. The reference for policies is that the candidate opposes banning LGBTQ content in schools, takes no action regarding climate change, and maintains aid to Israel.
Figure 4: Results of the Conjoint Analysis
Our first key finding is that demographic traits, by and large, do not significantly move the needle when it comes to American Muslim votes. Respondents were only slightly more likely to pick a candidate who was Middle Eastern (2.5%) or Black (3.7%) relative to one that was White. Similarly, while Atheists had a slight disadvantage relative to Christians (4.5%) and Jews (3.9%), it was only with Muslim candidates that respondents evidenced a meaningfully higher likelihood to choose them (13.5%), all else being equal. Surprisingly, this trend of near parity held even when it came to party affiliation. Republican candidates were only at a slight disadvantage (3.2%) relative to Democrats.
However, on the policy front, there was considerably more variation. For example, the issue of curricular content clearly still matters for Muslim voters. Respondents were 12% more likely to vote for someone who supports banning LGBTQ material in K-12 education relative to someone who opposes such a ban. The issue of climate change had a somewhat more muted, though still noticeable, impact on vote choice. Relative to a candidate that does not believe any action needs to be taken to address climate change, a candidate who believes that moderate action is required is 6.9% more likely to be chosen and one who advocates for aggressive action has a nearly 10% edge (9.4%, to be precise).
Then, there is the issue of aid to Israel. The results in Figure 4 offer the clearest indication that the genocide in Gaza is the primary consideration when American Muslims cast a ballot. But, just how much does this issue matter? A candidate who wishes to maintain the current aid regime suffers a severe penalty, all else being equal. Relative to someone who supports the status quo, American Muslim voters are 30% more likely to cast a ballot for candidates who wish to condition aid to Israel, while those pushing to freeze aid altogether are a whopping 55% more likely to earn the community’s vote.
In one respect, the evidence presented in this post confirms the obvious—Gaza matters a lot to American Muslims and will likely drive their vote choice to a considerable degree this November. What these data highlight is just how much foreign policy in the Middle East, especially Gaza, matters above and beyond all other issues. Moreover, given the magnitude of our results, coupled with the findings of our earlier survey highlighting voter intentions, it is clear that a candidate’s foreign policy will weigh heavily on the voting behavior of the American Muslim community—not just this cycle, but likely for the foreseeable future. When push comes to shove, American Muslims vote with the ummah in mind.