Is French Islam an Oxymoron?


Review of Why Muslim Integration Fails in Christian-Heritage Societies, by Claire L. Adida, David D. Laitin, and Marie-Anne Valfort

Cambridge: Harvard University Press, 2016

In 2011, the American anthropologist John Bowen published a book entitled Can Islam be French? Relying on field work and ethnographic data, Bowen argued that France’s five million Muslims were responding to the distinctive French strain of secularism called laïcité by adapting. They were, on Bowen’s account, crafting new species of political associations, religious organizations, and educational venues, while also innovating theologically to reconcile their faith commitments with prevailing social and legal norms. With subtle legerdemain, Bowen thus subverted his titular question: Through mundane and everyday practice, he intimated, France’s five million Muslim citizens were already well on their way to being, well, French.  

In the subsequent five years, how have post-recession economic malaise, repeated terrorist atrocities by European Muslims, and the Syrian migration crisis reshaped intercommunal relations in France? Bowen has reaffirmed his optimism. But other omens bode more darkly. In September 2015, for example, the French newspaper Le Monde tallied “more than fifty” acts of vandalism against mosques in the French mainland.  In the first six months of that same year, a watchdog called the L’Observatoire national contre l’islamophobie identified 274 “anti-Muslim acts,” a 281 percent rise on the year before. Important context for these estimates is supplied by data on bias crimes against other religious minorities. Some 500 anti-Semitic incidents were recorded in the same period in France—an 84 percent hike on the previous year in a nation that has Europe’s largest Jewish community. Whatever the official norm of religious tolerance, practice on the ground in France seems decidedly mixed.

To some observers, those statistics point toward a more sinister subterranean current in the French psyche. In the sort of literary coup de foudre that seems possible only in France, a distinguished historian and sociologist, Emanuel Todd, sparked heated debate in 2015 by diagnosing among the French a resurgent epidemic of “zombie Catholicism.” While no longer practicing Catholics, Todd argued, many of his co-citizens retained that faith tradition’s authoritarian, anti-egalitarian, and xenophobic frames of sentiment. Worse, Todd argued, they also persisted in a nostalgic longing for a transcendental form of political union to replace their lost faith. Todd’s polemical sparked violent objections. Only in France, perhaps, would a sociologist’s book elicit a vigorous response from the nation’s prime minister accusing the academic of a “théorisation … brumeux.”

To judge by its title, a book called Why Muslim Integration Fails in Christian-Heritage Societies (“Why Muslim Integration Fails”) might be thought to contain a comprehensive explanation of the social, historical, and economic forces that Bowen, Todd, and their intellectual antagonists started to map. It should, that is, provide a synoptic account of the complex portmanteau concept of “integration,” and then detail both proximate and long-term determinants of its formation or dissolution over time. But this volume, by political scientists Claire Adida, David Laitin, and Marie-Anne Valfort, frames and answers a question that is narrower across two dimensions.

First, Adida et al. are primarily concerned with labor market participation, rather than the causes of bias attacks or legislation seeking to suppress certain religious practices. Second, their parsimonious explanatory apparatus is derived from economics. It is focused primarily on individual decision-makers, and is not well suited to investigation of structural, ideological, or cultural dynamics of the sort that Todd and Bowen explore. These disciplinary and methodological implications are not flaws per se. But they do belie the ambitious promise of the book’s title.

To understand the contribution of Why Muslim Integration Fails, it is helpful to step back and briefly recapitulate the two standard treatments of racial discrimination in employment contexts that the book exploits. Each is associated with a different Nobel Prize winner. First, Gary Becker explored ways in which individual tastes for discrimination might interact in various settings. His taste-based model of discrimination attracted criticism because it simply posited animus as a given, rather than explaining discrimination as a form of optimizing behavior. Second, Kenneth Arrow proposed an alternative model of statistical discrimination, in which employers used racial categories as imperfect proxies for underlying unobservable characteristics, such as productivity. Hence, the concepts of Beckerian taste-based and Arrovian statistical discrimination alike capture elements of individual behavior. To the extent that the success or failure of minority integration (however defined) depends on institutional forms, historical dynamics, or legal structures, therefore an account focused on these two mechanisms alone may be incomplete.

Of even more importance, the distinction between taste-based discrimination against a religious or a racial minority does not track the distinction between illegitimate and acceptable forms of discrimination. As Adida et al. briefly note, statistical discrimination is rational in the individuated, market-oriented sense commonly used by economists. But it may not be legitimate or normatively desirable. It might be based on incorrect, even culpably erroneous, beliefs about group differences. Even when it is based on correct beliefs, there may be reasons to condemn it. A relatively uncontroversial example today is employer discrimination against pregnant women, or women of childbearing age, which might be a rational form of statistical discrimination if you thought such women were more likely to take time off, disrupting projects and workflow. Even given such costs, there may still be powerful reasons to condemn and prohibit such discrimination by law. What’s economically rational for individuals, that is, isn’t the same as what’s right.  

The core ambition of Why Muslim Integration Fails is to parcel out individual-level reactions to the fact of Muslim affiliation between statistical and taste-based forms of discrimination. Because the authors claim to find both species of discrimination, they allocate blame to both French Muslims and nonMuslims, and suggest reforms on both sides. I have, as will become clear in a moment, hesitations about some of their conclusions. But the authors’ meticulous and extensively documented empirical strategy for eliciting and distinguishing different forms of discrimination deserves considerable praise. Even if one cavils at some of the inferences they draw from those findings, the empirical element of the project is admirable. It is, indeed, a model of the sort of useful and incisive work that empirical social scientists can do.

In order to isolate the discrete effect of religion, and specifically Islam, on others’ behavior, Adida et al. must disentangle it from the effects of possible confounding factors such as class, racial identity, and ethnicity. Their solution is to study a population of Senegalese-French Muslims and Christians, who do not differ systematically from each other except in respect to their religion. To understand the difference faith makes, Adida et al. then deploy a battery of methods, including a large-n survey, a correspondence test (involving the mailing of different CVs to employers), ethnographic interviews, and two rounds of experimental games. As the authors acknowledge, their identification strategy raises a question of external validity. Islam in France is associated with North African migrants, not Senegalese ones. It is possible that interactions between religiosity and ethnicity generate different dynamics and beliefs in respect to that numerically dominant tranche of French Muslims. That Adida et al. replicate elements of their findings with larger European and American data sets, nevertheless, provides some comfort that their conclusions are not limited to a singular and narrow demographic context.

Deservedly, the experimental games form the centerpiece of the book. Adida et al. brought together a group of Senegalese and non-Senegalese French nationals, solicited through careful interviews on the Paris streets, to participate in a sequence of ‘games’ that allow for the identification of statistical or taste-based discrimination. Although Adida et al. offer comprehensive descriptions of each game in the book, I can only offer a taste of their ingenious approach. Hence, in their dictator game, participants were presented with money and an array of names and faces. They were then asked how they would give to each person in the array. As economists such as John List have demonstrated, the dictator game can illuminate varying levels of altruism and winnow out taste-based discrimination.

Adida et al. added a first twist to the game by changing the number of Senegalese Muslims in each game session to test whether the presence of more Muslims altered levels of altruism. In yet another variant on the dictator game, participants were asked to guess what another, non-Senegalese French player would give, which allowed the researchers to gauge participants beliefs about their peers’ preferences. By deploying these and other tactics, Adida et al. are able to draw inferences about levels of both statistical and taste-based discrimination. These inferences were then supplemented with other sources of data (e.g., how employers responded to CVs identical except for religious affiliation) and extant scholarship.

Of the two conclusions that Adida et al. draw, their finding of persistent and strong taste-based discrimination among French non-Muslims against Muslims—even after controlling for ethnicity, race, and other characteristics—is the more compelling, if also less surprising. Adida et al. show that non-Muslim French participants in the dictator and other games evinced strong taste-based discrimination against Muslims qua Muslims. In a twist, they also demonstrate that as the number of Muslims in a game session increased, non-Muslim participants’ antipathy waxed. In the words of former Minister of the Interior Bryce Hortfeux, “When there’s one [Muslim], that’s OK, it’s when there’s a lot of them that there are problems.”

This “Hortfeux effect,” as Adida et al. aptly christen it, tracks a well-known theory of racial threat associated with the American sociologist Hubert Blalock. Writing in the late 1960s, Blalock predicted that (white) racial majorities would exercise increasingly strong racial preferences as the proportion of a (black) racial minority in a jurisdiction increased. Blalock’s theory has found large resonance in the United States. Last year, for example, the political scientist Ryan Enos published a study of white voters’ behavior before and after the eliminated of concentrated public housing largely populated by African-Americans in downtown Chicago. Enos found that voter conservatism and turnout increased linearly with proximity to those projects before their destruction, but that both the right-shift and the turnout fillip disappeared after black residents had been dispersed.

Should the identification of taste-based discrimination surprise? I have no doubt there are those who deny that racial or religious discrimination occurs in France, just as there are in the United States. But they are unlikely to read an academic study of this sort. Even then, systematic repudiation of the claim that religious bias is an aspect of French life should be welcomed. At the same time, I suspect that many empathetic French readers, as well as Muslims who have lived in other European countries, will not find Adida et al.’s finding of taste-based discrimination much of a surprise.

The experimental games deployed by Adida et al. provide direct evidence of taste-based discrimination. But they do not supply direct evidence of statistical discrimination by employers. Participants were not asked whether they would hire Senegalese Muslims; the games tapped into judgments about trust and altruism, not fitness in the workplace. Instead, Adida et al. largely rely on an earlier study entitled Allah a-t-il sa place dans l’entreprise? by Dounia Bouzar and Lylia Bouzar for evidence of statistical discrimination in employment contexts. Drawing on this study, Adida et al. single out three aspects of French Muslim behavior “that can legitimately be considered as problematic for a French firm’s productivity and espirit de corps”: the expression of religious norms in disruptive ways, the greater prevalence of gender bias among Muslims, and lower linguistic abilities. Based on these generalizations, Adida et al. conclude, employers have “a sense that, all else being equal, Muslims will perform less well in their firms.”

Looking closely at the evidence for these three bases for statistical discrimination, however, reveals a more nuanced picture. Of the three grounds for discrimination, Adida et al. find direct evidence only for the second: Muslims in their sample evince more gender bias than non-Muslims. They do not, however, explore whether Muslims engage in biased or harassing behavior at a different rate than non-Muslims. To the contrary, they leave the impression that gender bias is a uniquely Muslim phenomenon in France. This seems unlikely. (Consider a 2005 governmental study that found that one in ten French women have been beaten by an intimate partner, with a death occurring on average every four days). It may alternatively be that both that French Muslim men have higher levels of latent bias, and yet are less likely to engage in harassing or demeaning behavior toward women in the workplace as a consequence of cultural norms that conduce to greater reticence rather than more aggression. I don’t mean here to take sides, but rather mean to stress that it remains an open question whether French Muslims’ behavior in public settings such as the workplace is meaningfully different from non-Muslim behavior so as to warrant employer skepticism at the hiring stage.

The case for statistical discrimination based on poor language skills or disruptive religious behavior is also weak. Adida et al. find evidence of roughly equal linguistic capacity, which would seem to undermine an argument for statistical discrimination. The best they can say is that it “remains a possibility” that French Muslims tend to have weaker mastery of the French language. Further, their evidence of the incidence of disruptive religious behavior, drawn from Bouzar and Bouzar, suggests that it is “rare and extreme.” Instead, what triggers a “torrent of anger” among French non-Muslims is the use of a headscarf by a Muslim woman, which is viewed as being “like imposing your opinions.”  The French, no doubt, have their reasons for directing women en foulard. But do these reasons really warrant a “torrent of anger”? Adida et al. simply  assume so.

As I’ve noted, Adida et al., do fleetingly recognize that not all forms of statistical discrimination are founded on true and morally defensible beliefs. But in discussing their findings on statistical discrimination, they do not adequately consider whether the statistical discrimination they purport to identify should be a valid ground of decision (akin to the quality of an undergraduate degree) or an invalid ground (as pregnancy, in addition to race, sexuality, and faith are under U.S. law). To an American audience, it is hardly obviously that the decision of a Muslim woman to wear a headscarf, any less than a Sikh man’s decision to wear a turban or a Jewish man’s decision to don the yarmulke, should be deemed a legitimate incitement to discrimination.

This may be so even if the observant person’s behavior is somehow disruptive. American law again contains many accommodation mandates for religion, disability, and other minority traits. The scale and justifications for these rules, of course, are contested. But it is not at all obvious that the refusal to accommodate religious headdress necessarily falls within the domain of legitimate employer judgments a liberal democratic state that aims to promote religious inclusion should allow. The question of what can “legitimately” justify statistical discrimination, in other words, raises real normative questions that Adida et al. do not engage with in an adequate fashion.

In offering proposals for reform, Adida et al. are careful and largely sensible. When they propose public campaigns against taste-based discrimination, more robust public enforcement of extant antidiscrimination laws, and enhancing primary education for economically and socially marginalized minorities, their agenda seems both obvious and compelling—although not at odds with the conclusions offered by many commentators on French social relations. But at times the normative commitments that underwrite particular sentiments are neither clearly stated nor obviously compelling. What to make, for instance, of their proposal that Muslims forego Islamic names for their children, like, for example “Aziz”? Although Adida et al. briefly acknowledge that this practice may implicate costs, they do not engage with the powerful arguments that the legal scholar Kenji Yoshino (among others) has lodged against such “covering.”

It is surely right to say, as Adida et al. do, that both French non-Muslims and Muslims have a role to play in mitigating intercommunal tensions, and to attain a new modus operandi. Adida et al. might further be read to suggest that the burden of adaption should be evenly shared. I am not so sure. It is not just that the evidence for statistical discrimination is weaker, and the moral case for some accommodation of religious expression is stronger, than Adida et al. let on. It is also that responsibility for taste-based discrimination can only be allocated to those who hold and act upon those reprehensible tastes.

Whether such change will happen is another matter. In the twilight of dwindling postwar economic success, in the fever-pitch state of post-Paris-attack emergency, and as public support for the National Front waxes, it would take, I think, an exceptional effort. The French, to be sure, have long insisted on their exceptionalism. Whether it extends to this febrile and controversial domain remains an open question.

Posted on 28 March 2016

AZIZ HUQ is Professor of Law at the University of Chicago Law School.