For more than a decade, Mathew Creighton has published numerous articles on topics related to xenophobia. Many of these rest on the skillful use of list experiments (Item-Count Technique, ICT) to correct for social desirability bias. Hidden Hate synthesizes these contributions in a book that will be of great interest to both academics and non-academics and that is particularly well suited for the classroom. The synthesis is articulated around a coherent and original narrative that conceives individuals as strategic agents, attentive to context when they answer direct survey questions concerning their attitudes to migration and to the naturalization of migrants. Aware of the stigma or social disapproval, attached to specific opinions, they consider this context before deciding whether to openly express their views or mask them. List experiments in public opinion surveys, conceived to make the respondent feel that his or her views will remain totally anonymous, help uncover these hidden views and measure the degree of social desirability bias influencing estimates of the attitudes toward migration in the population.
The book’s main take-aways are 1) that there is more xenophobia in the population than conveyed by answers to direct questions on migration, 2) that group comparisons based on direct survey questions do not adequately capture group contrasts in levels of xenophobia and trends in levels of xenophobia over time in those groups, and 3) that explanations of levels of xenophobia based on multivariate statistical methods that use answers to direct questions are likely to be wrong because individuals with different characteristics are likely to mask their real views to different degrees. Creighton takes these findings as the point of departure for a reflection on the persistence of discrimination in liberal societies and on ways to address it. He argues that laws and norms penalizing the overt expression of xenophobia and prejudice are insufficient, for there are many areas in life where they can easily be masked. He therefore advocates policy interventions that force transparency in key areas where discrimination has been demonstrated or is suspected. In the area of hiring, for instance, Creighton believes that anonymizing CVs is not enough and that companies should be asked to make explicit and to document their hiring criteria and processes. The goal of these interventions would be to prevent masking and to nudge people and organizations into non-discriminatory behavior.
The book draws on Creighton’s list experiments with US data before and after the great recession, with data from the UK before and after Brexit, and data for the Netherlands, Ireland, and Norway. It is structured around chapters that examine the roles of economic crisis, political debate, frames that connect immigrants and immigration with specific geographic, ethnic, racial, religious, or immigration status, and the in-group, out-group ethnicity of those expressing attitudes toward migrants, in explaining the extent to which survey respondents conceal their xenophobic views.
Key to the study’s relevance is the suitability of list experiments in unveiling masked views on sensitive topics. In its simplest form, the experiment divides a sample into a control and a treatment group. Both groups are provided with a list of two or three statements and respondents are asked to state the number of items with which they agree, without being specific as to which ones. It is important that these statements are selected so that the probability that respondents agree with none or with all of them is minimal. Otherwise, the anonymity of the answers is lost. In addition to this, members of the treatment group receive an additional statement than members of the control group. This extra statement concerns the sensitive topic of interest (in this book, attitudes toward migrants with different characteristics and migration). Finally, members of the control group are asked to openly express agreement or disagreement with the extra statement included in the expanded list provided to the treatment group (overt support).
Based on the answers provided by members of the control and the treatment groups, one can calculate the average number of statements with which members of each group agrees. This number should be greater in the treatment group than in the control group. Since randomization ensures that the control and the treatment groups are comparable in every aspect, the difference between the two averages reveals the proportion of respondents in the treatment group who agree with the sensitive statement of interest (covert support). This proportion can be treated as a more valid measure of agreement with the statement than is direct questioning since the self-image of members of the treatment group is protected by the fact that they only state the number of statements with which they agree without being specific as to which ones. Meanwhile, the difference between the proportion of respondents in the control group who agree with the sensitive statement when asked directly (overt support) and the estimate of the proportion of respondents in the treatment group who agree with the sensitive statement, obtained by subtracting the average for the control group from the average from the treatment group (covert support), provides a measure of the extent of social desirability bias concerning the topic of interest.
List experiments have been around for more than forty years and variants have developed to tackle this or that problem. They compete with other methods developed to address the question of masking on sensitive issues. The author spares technicalities to the reader, technicalities that can be found in his single- and co-authored articles, and simply contrasts list experiments with statistical approaches to correct random measurement error, as in confirmatory factor analysis. As the author stresses, masking is not a random process and, therefore, it cannot be successfully addressed through these other statistical methods. To be sure, the list experiment method has been validated through meta-analysis of published studies [see Ehler, Wolter and Junkermann, 2021]Footnote 1. It is not without problems, however, such as its relatively low statistical power, which demands very large samples, or, as a recent study claims, its underperformance with highly politicized sensitive topics, precisely the ones for which the method was developed [see Rinken, Pasadas del Amo, and Trujillo Carmona, 2024]Footnote 2. In light of this discussion, the book’s findings are generally plausible.
The findings presented in the book are sobering, for they reveal that xenophobia is much more prevalent in Western countries than traditional surveys suggest. In fact, from what one gathers in Rinken et al.’s article cited above, the list experiment method, despite its virtues, may still underestimate how prevalent xenophobia is. One can imagine, for instance, that people practiced in masking their prejudice may have turned this into a habitus and may disagree with xenophobic statements even when, as in the treatment situation, they are protected by total anonymity.
The book also shows that xenophobia is quite ecumenical, for it targets every kind of immigrant group, regardless of ethnicity, race, religion, or even migration status (i.e. refugee vs other types of migrants). Contrasts in expressed views to different groups of immigrants, as revealed through direct questions, mainly reflect different degrees of masking of people’s real views. In the United Kingdom, for instance, overt opposition to immigrants from Muslim countries increased in the pre-post Brexit period, whereas covert opposition decreased. More generally, people seem to feel more at ease expressing opposition to migration from Muslim countries and Muslims in general than in expressing opposition to other groups.
For the academic reader, one of the most striking, if not totally unexpected, revelations through the list experiment method, is the absence of a gradient by education in levels of xenophobia. That is, highly educated individuals are by and large as prejudiced as are less educated ones. It is a pity that the author does not zero-in on this major finding for it calls into question the credibility of most of the published literature on the individual-level determinants of xenophobia, of the theories built on them, and of the validity of deductive theories that seek validation through statistical findings that do not correct for strategic behavior. It also throws light on the intriguing question of the persistence of a positive “direct” effect of education, not only on xenophobia, but on countless other political topics, once one controls for the variables that should mediate its effect. It suggests that it partly captures the fact that highly educated respondents are more likely to hide their real attitudes on these topics than are less educated ones.
It is also a pity, and this is not meant as a critique, that Creighton does not elaborate more on the implications of a lack of an education effect on xenophobia for the interpretation of the rise of anti-immigrant populism and the toughening of migration policy and controls in advanced democracies. Since political and media elites, regardless of political orientation, tend to be highly educated, one can speculate that the closing of borders is not just a pragmatic elite response to bottom-up popular demand or a strategic move by elected governments to stop the rise of populist right parties. Rather, it is a policy that the elites themselves condone although they would never admit to it.
While strong on description, the book is somewhat weaker on explanation. Throughout, the author persuasively emphasizes the role of context and framing in determining the amount of masking of real prejudice and which groups may incur more into this masking behavior. Indeed, the experiments show that masking varies across contexts and depending on how migrants are framed. The experiments do not speak, however, to the specific causes of this variation. Creighton goes into elaborate explanations for which he sometimes lacks empirical or theoretical backing. This weakness is particularly relevant in the chapters devoted to change over time, for the research design on which these are based, the one-group pretest-posttest design, is among the weakest types of quasi-experiments. The author highlights changing economic circumstances and the Brexit political debate (and the 2015 refugee crisis) as factors explaining why US citizens were more likely to express support for a wall between the United States and Mexico and the degree of opposition to migration from different geographical areas. The explanations he provides are persuasive and provocative. For instance, Creighton criticizes materialist explanations of xenophobia that emphasize economic insecurity as a cause of xenophobia. He argues instead that economic insecurity makes it more legitimate to express xenophobia and prejudice. While plausible, however, the interpretations remain speculative to the extent that the one-group pretest-posttest design does not rule out that other uncontrolled factors explain the observed changes [see Shadish, Cook and Campbell, 2001]Footnote 3.
The book also leaves us wanting to know more about the reasons why some groups are more xenophobic than others, as measured indirectly through the list experiment. It is clearly not the author’s intention in this book and therefore this comment should not be taken as criticism. It simply points to where future research should be directed.
Finally, Creighton underplays the political implications of the highly prevalent xenophobia in economically advanced democracies. In various chapters, the author draws our attention to the apparent contradiction, as in the Netherlands, between the high-level of tolerance expressed by respondents in public opinion surveys and rising support for the anti-immigrant far right. He interprets this contradiction through the lens of his analytical frame, which expects people to mask their real views when they anticipate social stigma if they express them openly. While I find this interpretation correct, it distracts us perhaps from the larger problem, which is that the prevalence of xenophobia is much greater than is support for the far right: in other words, the fact that there is still much room for stricter migration policy and discrimination against migrants as well as for more support for the far right if migration remains at the center of the political debate.
On the whole, Hidden Hate is an important work. By highlighting the lack of validity of direct measures of xenophobia, it calls into question much of what we think we knew about the explanation of individual and aggregate levels and trends in xenophobia in economically advanced democracies, provides a useful methodology to address the problem, and proposes useful and original policy ideas to limit discrimination.
