Jewish Perceptions of Antisemitism in the European Union, 2018: A New Structural Look. 3
Sources and Methods
Recent Approaches to the Study of Antisemitism
The study of contemporary antisemitism utilizing the social sciences approach should first of all specify the main investigative strategies pursued, the sources of available data, the respective limitations, and the principal interpretative implications.
The first stage has entailed assembling and collating inventories of events rated as antisemitic. Such events may include violence or harassment against a specific person as a Jew, against other Jewish persons, Jewish properties, or Jewish institutions. Such events may be reported or denounced by Jews and/or by others. Such databases were developed by community and academic-oriented Jewish organizations, as well as by general public organizations (Knobel 2013; OSCE-ODIHR 2016; Kantor 2018; FRA 2018b). The advantage of these types of documentation is that they represent factual events that really happened. Examining the variation of their frequency diachronically and spatially enables the creation of a historical series of data, which, in turn, may constitute the background for local and international comparisons between simultaneous fluctuations in antisemitism and in other social aspects. The limit of this method is primarily its dependency on people’s willingness and ability to report such events. Research evidence (including FRA 2013 and FRA 2018a) indicates that reporting is extremely reticent and imperfect, leading to gross underreporting of such events. In addition, clearly, only a small minority of individuals within a given population are involved in explicit acts of harassment, profanation, and violence. Antisemitism comprises a much wider range of actions and written or verbal expressions apt to touch upon Jewish sensitivities. The possible ratio stands at several thousand documented acts of antisemitic acts worldwide per year, versus tens or hundreds of millions of people holding and/or expressing antisemitic views during the same year.
A second approach explored the frequency and nature of anti-Jewish attitudes among the total population (ADL 2013; ADL 2014). In the largest such study on record, in 2013, the Anti-Defamation League approached 500 individuals in each of 102 countries, for a total of over 50,000 respondents (ADL 2014). Respondents were asked to express agreement or disagreement with a list of anti-Jewish prejudices. Those who agreed with six or more statements were included in an index of antisemitism. The study did not investigate actual discrimination enacted against Jews. This, incidentally, might have been difficult or even impossible, considering that in many countries the Jewish population is very small and many or most people do not know personally any Jew. There may be doubts about the meaning of such a definition and measuring procedure of antisemitism, exclusively circumscribed to the cognitive dimension, but one cannot minimize the usefulness of the ADL study, which for the first time provided some comparable measure of the extent of antisemitism in 102 countries. Another serious caveat concerns the description of society as a rigid dichotomy of antisemites and not-antisemites. It is actually clearly demonstrable that antisemitic prejudices and attitudes are diffused gradually and flexibly across populations (Staetsky 2017; DellaPergola 2015), and this should be reflected in data analysis.
An American organization, the Pew Research Center, repeatedly investigated the attitudes toward Jews of representative samples of the total population in several countries, including Europe (Pew Research Center 2016/16 and 2017), as did, among others, the American Jewish Committee (e.g., Jodice 1991; Karmasin 1992) and the CNN broadcasting network (Greene 2018). A recent and quantitatively more impressive research framework is represented by the several Eurobarometer surveys of perceptions of the existence of antisemitism among general population samples in each of the 28 member countries of the European Union (e.g., European Commission 2019). These surveys also explored the extent of knowledge about Jews and Judaism in the respective countries.
A third type of approach has focused on perceptions of antisemitism by the Jews themselves. Major examples were the two large surveys of Jewish populations undertaken by the FRA in 2012 among Jews in nine EU countries (FRA 2013) and in 2018 among Jews in 13 countries (FRA 2018a). Results of the 2018 study (see more detailed discussion below) constitute the principal basis of the present paper. The basic demographic profiles of those included in the surveys were shown broadly to correspond to data on Jews known from other independent sources, which confirmed the reliability of the samples. The two studies, in addition to providing detailed evidence about Jewish perceptions and experiences with antisemitism, collected data on the sociodemographic and Jewish identification profiles of respondents.
All the different cross-sectional public opinion studies mentioned so far – whether aimed at the total population or at Jews only – attempted to cover a wide range of perceptions and experiences related to antisemitism. In addition to general issues of sampling quality and errors, their validity is limited to the specific points in time when they were carried out. Repeated observations – as in the case of the 2012 and 2018 FRA surveys – allowed the construction of rough time-series and the assessment of ongoing trends. Longitudinal surveys, involving compilation of repeated observations of the same individuals over different points in time, would produce a better sense of changing perceptions among the same individuals – hence a more accurate picture of the evolution of perceptions over time. To the best of my knowledge, however, such a survey design has not yet been tested in the field of antisemitism studies.
A fourth method of investigation has focused on the contents of discourse, mostly about Jews and/or about Israel. This important field, which so far has been explored insufficiently, consists of the analysis of openly or latently antisemitic verbal and textual contents expressed in the conventional printed (e.g., Partington 2012) and electronic media, in the web and in the fast-developing social media. In exploratory work undertaken so far, selected key words related to Jews, Judaism, and other general topics were suggested to samples of respondents and the associations emerging in relation to those words were verified (Petrenko and Mitina 1997; Guetta 2013; Russian Jewish Congress – Levada Center 2018b). More systematically, it is possible to examine various potential channels of diffusion of anti-Jewish prejudices, such as in large bulks of electronic mails, the display of internet sites, exchanges on Facebook networks, and the like (Schwarz-Friesel and Reinharz 2017). Carefully selected semantic associations among words can result in more powerful and lasting effects on larger audiences than mere acts of violence accomplished in specific locations. Whereas the latter are easy to detect and report, the former may be more elusive, as they require devoting substantial time to reading and listening, careful coding of contents, as well as a wide command of historical, philosophical, and literary sources, and meticulous attention to nuances.
Studies incorporating broad sets of external indicators distributed along a defined time span have been infrequent although they could provide the explanatory context to specifically antisemitic attitudes and behaviors that are observed over the same time span – for example, we may note the attempt to find a correlation between antisemitic manifestations and the fluctuations of the economic conjuncture (Epstein 1993); the military situation in the Middle East (Eddy 2014); or Muslim immigration to Europe (Feldman and Gidley 2018). For sure, further research needs to be undertaken in order to ascertain whether observed and perceived expressions of antisemitism reflect contingent situations related to occurrences in specific times and places or rather reflect a manifest or latent feature inherent in society and scarcely sensitive to changing circumstances.
The FRA 2018 Study
This paper develops a new analysis of the data collected in the second survey on Jewish people’s experiences and perceptions of hate crime, discrimination, and antisemitism, undertaken in 2018 at the initiative of the European Union’s Agency for Fundamental Rights, FRA. The main results were published in FRA (2018a). I re-elaborated the original data taking advantage of Similarity Structure Analysis (SSA), a technique of data processing based on Facet Theory (see below).
The FRA 2018 survey built strongly on the experience and methodology developed for the 2012 FRA survey on discrimination and hate crime against Jews, which covered nine EU Member States (France, the UK, Germany, Hungary, Belgium, Italy, Sweden, Latvia, and Romania. Data for the latter country were omitted in the final analysis because they were of poor quality). The conceptual infrastructure was provided for the most part by an academic advisory committee that included several leading specialists on issues of contemporary European Jewry convened in 2012 by the Jewish Policy Research Institute, JPR, in London. FRA was mainly interested in assessing the practical instances of physical aggression and harassment and the extent of Jewish responses to them, as well as the awareness of existing laws aimed at protecting the Jewish community. These aspects are not dealt with in detail in the present study, which mainly focuses on perceptions and experiences of antisemitism contents, transmission channels, and perpetrators. Such topics were discussed and defined by the advisory committee reflecting the themes covered in the previous section of this paper.
The 2012 FRA survey generated a crop of research papers, which examined the issues both in the integrated perspective of the eight countries covered, and in the local framework of specific country reports. The latter related to the United Kingdom (Staetsky and Boyd 2014), Italy (DellaPergola and Staetsky 2015), Belgium (Ben-Rafael 2014), France (Cohen 2012 and 2013), Sweden (Dencik and Marosi 2017), Germany (Glöckner 2013), Hungary (Kovács 2013a), and Latvia (Kovács 2013b).
In 2017, a stakeholders’ consultation was carried out in Vienna at the initiative of the FRA to elaborate the feasibility plan of a new study to be conducted in 2018 in 13 countries – Austria, Belgium, Denmark, France, Germany, Hungary, Italy, Latvia, the Netherlands, Poland, Spain, Sweden, and the United Kingdom. An expanded academic committee reflecting the larger number of participating countries advised on the design and implementation of this survey. The FRA’s 2018 survey collected data from 16,395 self-identified Jewish respondents (age 16 or over) in 12 EU Member States (the data for Latvia were omitted from the final analysis because of poor quality). Countries were selected for the study primarily because of their relatively larger Jewish population size but also in view of their locations in the different regions of the continent: north, south, east, and west. The 12 EU countries covered correspond to 97 % of the EU core Jewish population, estimated at above one million in 2018 (DellaPergola 2019).
In preparation for the 2018 survey, the 2012 survey questionnaire was subjected to a review that resulted in changes in some of the questions. Efforts to reduce the survey length were taken, with a view to minimizing the respondent burden. This included reviewing possible questions for deletion and reducing the number of items and answer categories in individual questions. Some questions were deleted; others were streamlined, rephrased or repositioned in the questionnaire to improve the flow when answering the questions. The questionnaire was also revised to establish a design compatible across most common, latest operating systems (such as Microsoft Windows, Apple’s iOS, Linux) and that also could work on different types of devices, including desktop and laptop computers, tablets, and smartphones that could be used for completing the survey. Questionnaire revisions aimed at retaining comparability with the 2012 survey to the extent possible. The 2018 survey questionnaire is available in a separate report (FRA 2018c).
The survey collected data through an open online survey, which was open for respondents to complete during seven weeks in May-June 2018. Eligible participants were all those self-defining as Jews, aged 16 or over, and resident in one of the survey countries. The questionnaire was administered online and could be accessed via an open web link that was publicized on the FRA website, through extensive advertisement in Jewish organizations, Jewish media outlets, and social networks. People who had connected with the survey were asked to snowball to acquaintances who might not have been aware of it. Some doubts remain whether the most marginal and unconnected layers of the Jewish public actually had or wanted to have access to the survey instrument. Information collected about the patterns of affiliation of respondents provided a reasonable sense of confidence that the survey did reach the peripheral fringes of the community (Staetsky 2019a). In some cases, participation of Jews belonging to the Haredi sectors of the Jewish community may have been lower than average.
A consortium of Ipsos MORI and the Institute for Jewish Policy Research (JPR), both based in the United Kingdom, managed the survey data collection under the administrative supervision of FRA staff. National research teams of academic experts and local researcher and community liaison points in each survey country supported the survey implementation.
JPR collected information on the size and composition of the Jewish population in each country and on Jewish community structures in the European countries involved, identified ways to raise awareness about the survey among Jews in the selected countries, and implemented a capillary communication strategy. Ipsos MORI ensured the technical set-up of the survey, including the translation of all survey materials, development of the survey website, and compliance with the standards of data security, privacy, and confidentiality.
Throughout data collection, responses were monitored using the online monitoring tool provided by Ipsos. This enabled monitoring of response levels on a daily basis and enabled observing the impact of particular communications campaigns by different organizations across the survey countries and checking the distributions of responses by age, sex, geography, and Jewish affiliation to assess how the communications campaigns were reaching difference segments of the target population. Because of the voluntary and self-selected nature of respondents, the sample cannot be considered strictly representative as would be a random probabilistic sample of the target population. Comparisons between the survey results and Jewish population distributions by age, sex, and major geographical region available from other national and Jewish community sources showed relatively modest amounts of bias. For the purpose of population characteristics description, the sample can be weighted according to the distributions in those other sources. As the present paper is concerned with relationships between variables rather than with population profiles, unweighted data were used here.
The data collection outcomes confirmed the experience of similar online surveys that the launching day is critical. In this case, over 4,000 responses were obtained on the first day alone, constituting nearly a quarter of the total sample. Following the processes undertaken to assess the quality of the data and cleaning of the dataset, the final dataset included 16,395 completed questionnaires across the 12 valid survey countries, excluding Latvia. The average time for survey completion was 33 minutes, and the median duration was 27 minutes.
The response rate in 2018 was significantly higher than in 2012 (see Appendix table). In the seven countries where data can be compared for both dates, the number of valid respondents increased from 5,663 in 2012 to 13,083 in 2018, an increase of 131% or more than double. The number of respondents increased especially in France (+233%), the UK (+222%), Germany (+103%), Belgium (+79%), and Sweden (+47%). This improved coverage of the Jewish population testifies to considerably more efficient advertising of the survey, enhanced access to the web, and also probably greater awareness of and interest in the topic investigated.
Perceptions of Antisemitism inside and outside the Jewish Community
One important research question emerging from the various sources of data reviewed so far is whether the existence of antisemitism is perceived to an equal extent by those who are its target and by the many more around them who constitute the majority of population. The correlation that may exist between internal and external perceptions of antisemitism at the country level can be judged by comparing different studies that have addressed, respectively, Jewish or general population samples in the same places and at more or less the same time. Several such studies undertaken in recent years allow for such comparisons.
In Figure 1, I compared the ADL data for 2013 and the FRA data for 2018 for the 12 countries included in the latter study. It is important to note that the ADL study of total populations concerned primarily the cognitive perceptions of antisemitism, whereas the FRA study of Jewish populations also covered its experiential aspects. The FRA data reported in Figure 1, however, relate to a cognitive question about the concern with antisemitism as a societal issue in the country of residence; thus making it comparable with the ADL data. In the figure, the horizontal axis measures the diffusion of antisemitic prejudice in society at large – assumed here as the explanatory variable – and the vertical axis measures the Jewish perception of the phenomenon – assumed here as the dependent variable.
In the ADL study, the proportion of the total population rated as antisemitic ranged between a low of 4% in Sweden and a high of 45% in Poland, while among Jews in the FRA study, perceptions of antisemitism as an important issue in society ranged between a low of 56% in Denmark and a high of 95% in France. It thus clearly demonstrates how much more sensitive Jews are than others when assessing antisemitism and the environment within which it develops. The general ranking of countries along the two measures, however, was overall very consistent, showing a 52% positive correlation. The determination quotient (R2 = 0.2728, or 27%) indicates how much of the variance in perceptions by Jews (the variable on the vertical axis) is statistically explained by attitudes among the general public (the variable on the horizontal axis). If the apparent outlier case of Denmark is omitted, the variance explained is somewhat reduced to 20%. The result is statistically significant although not very powerful. The data in Figure 1 suggest quite clear regional differences between northern, western, and eastern European countries. Regional differences within the EU were clearer from the ADL than from the FRA survey data.
Figure 1. Comparing measures of antisemitism in 12 EU Countries – FRA 2018 percentage assessing antisemitism as a very serious or fairly serious issue in society vs. ADL 2013 percentage of antisemitic respondents.
Key to countries: AT: Austria; BE: Belgium; DE: Germany; DK: Denmark; ES: Spain; FR: France; HU: Hungary; IT: Italy; NE: The Netherlands; PL: Poland; SE: Sweden; UK: United Kingdom.
A similar comparison between the same ADL data and the 2012 FRA data (based on eight countries) resulted in a much higher correlation and explained variance: 54%. The six-year gap between the two FRA surveys may account for part of the inconsistency. Clearly, the sensitivity of Jews to perceived antisemitism is higher than on average among the surrounding population. In a reported context where perceptions of antisemitism were increasing rapidly, however, patterns of change in each country seem to have gone in somewhat different directions, thus reducing the strength of the immediate relationship between perceptions among the general population and within the Jewish community.
The 2018 FRA study of Jewish perceptions of antisemitism and the European Council’s 2018 Eurobarometer study of antisemitism perceptions among society at large provide similar comparisons. The Eurobarometer survey was undertaken in all 28 member countries of the European Union in 2018. Interestingly, the 12 countries investigated in the nearly simultaneous FRA survey happened to be those that registered the highest perceptions of antisemitism among the Barometer 28, with the sole exception of Spain. In this case, the comparison can be improved by distinguishing in both surveys those who perceive antisemitism as a very important or very serious problem and those who view it only as a fairly important or fairly serious problem. Looking first at those viewing antisemitism as a very serious issue (Figure 2), their incidence according to the Eurobarometer ranged between a low of less than 10% in Poland, Spain, Hungary, and Denmark, and a high of over 35% in Sweden. According to the FRA survey, the range was between 15% in Denmark and 65% in France. Jewish perceptions again appeared significantly higher than total perceptions. Statistically, country variance across the Eurobarometer respondents explained 17.5% of the country variance among FRA respondents. This coefficient of determination is considerably lower than that observed on the basis of the ADL study.
Figure 2. Comparing measures of antisemitism in 12 EU Countries: “Antisemitism is a very serious problem in your country.” Percentages – FRA 2018 vs. Eurobarometer 2018
Key to countries: AT: Austria; BE: Belgium; DE: Germany; DK: Denmark; ES: Spain; FR: France; HU: Hungary; IT: Italy; NE: The Netherlands; PL: Poland; SE: Sweden; UK: United Kingdom.
The relationship between general and Jewish perceptions was even weaker with reference to the perception of antisemitism as the aggregate of a very serious or fairly serious issue (Figure 3). The respective percent frequencies increased from a low of 20% in Spain and a high of 80% in Sweden according to the Eurobarometer, and from a low of 55% in Denmark to a high of 95% in France according to the FRA. With the possible exceptions of Denmark and Spain, the range of intercountry variation in acknowledging the seriousness of the antisemitism threat in society was somewhat reduced. Consequently, the amount of country variance in Jewish perceptions explained by country variance in general perceptions was a modest 10.4%.
Looking at these correlations, it is remarkable how Poland and Sweden occupied diametrically opposed positions in the ADL study seen above, versus the Eurobarometer study. The frequencies for Spain and Hungary were among the higher among the countries examined according to the ADL criteria and among the lower according to the Eurobarometer, although the percentages themselves were not that different. Possibly, the disparity partially is a reflection of the five-year lag between these two surveys of general public opinion, and their different methodologies. Part probably depended on the different respondents’ evaluation of antisemitism as a phenomenon permeating the society of the respective countries (Eurobarometer), as against the personal attitudes by respondents with regard to Jews (ADL).
Figure 3. Comparing measures of antisemitism in 12 EU Countries: “Antisemitism is very serious or fairly serious problem in your country. ” Percentages – FRA 2018 vs. Eurobarometer 2018.
Key to countries: AT: Austria; BE: Belgium; DE: Germany; DK: Denmark; ES: Spain; FR: France; HU: Hungary; IT: Italy; NE: The Netherlands; PL: Poland; SE: Sweden; UK: United Kingdom.
This observation prompted me to compare the findings of the ADL and Eurobarometer studies, both aimed at the total population, but with significantly different formulations of the dependent variable being investigated. As already noted, in the case of Eurobarometer, it was the assessed significance of antisemitism in society; in the case of the ADL, it was the assessed percentage of people with antisemitic views. The results are presented in Figure 4.
An extraordinary inverse relationship emerged between the findings of the two surveys, with an amount of explained variance of 18.5%, which if not exceptionally high, was nevertheless significant. These data are enlightening because they reveal the possible contraposition between personally harboring antisemitic prejudices, and perceiving antisemitism as a problem in society. Moreover, once again a definite clustering of European countries into regional areas with distinct perceptional patterns emerges: the low range of the ADL measure, tending to be higher on the Eurobarometer measure, included northern countries such as Sweden, Denmark, the Netherlands and the United Kingdom; the higher of the ADL range combined with the lower on the Eurobarometer range included countries of the former Soviet bloc area such as Poland and Hungary; the intermediate range included western countries such as France, Belgium, Germany, Austria, and Italy; Spain was in an outlier position, the lowest by Eurobarometer and among the higher by the ADL.
Figure 4. Comparing measures of antisemitism in 12 EU Countries – Eurobarometer 2018 percentage assessing antisemitism as a very serious or fairly serious issue in society vs. ADL 2013 percentage of antisemitic respondents.
Key to countries: AT: Austria; BE: Belgium; DE: Germany; DK: Denmark; ES: Spain; FR: France; HU: Hungary; IT: Italy; NE: The Netherlands; PL: Poland; SE: Sweden; UK: United Kingdom.
In any case, the amount of coherence that appeared to exist between the two general studies (ADL and Eurobarometer) and the FRA specifically Jewish study, while not absolute, consistently showed more acute and sensitive perceptions of antisemitism among Jews than among the general public. This may reflect a situation in which the perception of antisemitism has become over time an integral part of the patterns of Jewish identification – as will be argued later in this paper.
Cognitive, Experiential and Affective Domains
Another important question concerns the definition of the relevant boundaries of interest when researching perceptions of antisemitism. As is well known from general research in the field of social psychology, the whole gamut of human attitudes and perceptions can be classified into three main domains: cognitive/intellectual, behavioral/experiential/ instrumental, and affective/emotional (Tooby and Cosmides 1992). Likewise, in the study of antisemitism, we may imagine the total attitudinal and perceptional space of an individual or of a collective of individuals as a bi-dimensional shape. Figure 5 schematically exemplifies the expected partition of that space among those three main perceptional domains. In a multiple variable analysis aimed at better understanding perceptions and experiences of antisemitism, each of the variables included should ideally fall within any of the three delineated domains. This expectation would be true if we assume that the questions or indicators actually included in a given study succeeded in covering the entire human perceptional space and its domains.
Most historical and social scientific studies about antisemitism have stressed either a cognitive or a behavioral dimension. Considerable existing information inappropriately confounds these two aspects, whose empirical incidence is actually widely different: for the most part, cognition of antisemitism is more diffused than actual experience of it. Surprisingly, the third domain, the affective-emotional, has often been neglected in serious social scientific discourse. The affective/emotional aspect has been kept at the margins or basically ignored in studies of antisemitism, with the exception of specialized clinical-psychological studies directly focused on those aspects (e.g., Weinfeld et al. 1981, Helmreich 1996). And yet, I argue, it is crucially important to answer the question: How does antisemitism affect a person’s emotional and affective perceptions? For example, does antisemitism generate among Jews anxiety, anger, fear, aggressiveness, passivity, activism, loneliness, solidarity, creativity, or other feelings? Does it strengthen or weaken previously held Jewish religious or ethnic identities? Does it strengthen or weaken Jewish community relations and networks? Does it strengthen or weaken empathy, solidarity, and national identification with the country of residence of affected Jews? This still appears to be an uncharted research area.
Figure 5. Illustrative distribution of perceptions of antisemitism by main domains.
Each of the above-mentioned contexts requires different analytic tools and probably also different explanatory frameworks. To exemplify some of the emerging differences, I show in Table 1 a short synthesis and comparison between variables related to antisemitism as investigated at the cognitive and at the experiential level in the 2018 FRA survey of Jews in 12 EU countries. The data compare the highest frequencies reported among the different perceptional options that were offered to respondents within each selected area of interest.
In each instance chosen, the reported frequencies were higher at the cognitive level than at the experiential level. This accents the difference between awareness or belief that a given expression or behavior is antisemitic and actually hearing that expression or experiencing that behavior. Table 1 in particular synthetizes selected contents that Jews may perceive as offensive concerning alleged characteristics of the Jews themselves, the Holocaust, or the State of Israel. The frequency ranking of three major areas of contents appears to be reversed in the experiential data versus the cognitive data. The table also compares cognitive and experiential frequencies regarding the most used channels of transmission of such negative contents and frequencies of discrimination possibly (cognitive) versus actually (experiential) suffered.
Table 1. Highest frequencies reported regarding selected antisemitism perceptions and experiences in 12 EU countries, 2018
| Area of concern | Cognitive | Experiential |
| Contents of antisemitism | 95% | 51% |
| Negative on Holocaust | 95% | 35% |
| Negative on Jewish characteristics | 94% | 43% |
| Negative on Israel | 85% | 51% |
| Channels of transmission | 89% | 80% |
| Anti-Jewish discrimination | 49% | 21% |
Source: FRA 2018a, Figures 5, 6, 7, Tables 5, 7, 8.
These examples indicate that it is imperative carefully to specify the meaning of the data presented in any given analysis of the empirical evidence; unfortunately, this has not been a firm rule in past and ongoing research about antisemitism. In the present study, reflecting the database available and its limits, attention was devoted to the profile of and differences between cognitive and experiential domains. Unfortunately, relevant aspects in the affective domain were not investigated in the FRA study, and therefore their incidence cannot be compared with that of the other domains. The emerging evidence strongly suggests that future research on antisemitism should take into consideration all possible domains of the human perceptional sphere.
Similarity Structure Analysis – SSA
In the case of data analysis that focuses on quantitative analysis, the dominant mode in the literature has been the presentation of simple tabulations of data frequencies. Plain description or simple juxtaposition of results, beyond a first impression, does not usually generate more complex insights about theories or processes that would require the simultaneous processing of vastly larger quantities of information. A great improvement has been the use of causal inference, where a given variable is posited as the one to be explained (the dependent variable) and one or more other variables are posited to be the explanatory factors. This enables verifying the overall validity of a given hypothesized model and the respective explanatory suitability of each variable included in the model (e.g., Zick et al. 2011; Rebhun 2014; Cohen 2018).
The present paper demonstrates the use of Similarity Structure Analysis (SSA) (Guttman 1968; Amar and Levy 2014) in developing an original and thus far little exploited approach to the study of Jewish perceptions of antisemitism. SSA is part of the broader concept of Facet Theory, which is an approach to theory construction integrating research planning with data analysis. SSA aims at exploring the interrelationships that exist among large numbers of variables rather than focusing on explaining only one at a time. SSA computes the correlations between each of the different answers, in this case based on the original FRA survey response distributions within each of the 12 EU countries. In SSA, correlations are based on rank orders of variables’ values and not on the actual metrics of their quantitative differences. To make the concept of correlation absolutely clear to the general reader, if two or several respondents gave identical answers to two different questions, this means that the inherent contents of those two questions were very similar, and the two questions were strongly correlated. If the answers provided by different respondents were different, the correlation between those questions was weaker or even negative. The SSA software transforms these correlation coefficients – stronger or weaker – into appropriate distances – respectively, shorter or larger – between points on a bi-dimensional map, each point representing one variable. The emerging visual configurations are helpful in assessing the overall contents of subject matter and its logical partitions. Different spatial domains detected on a map – each with its own shared homogeneous contents visually similar to the example shown in Figure 5 above – represent a higher order of generalization concerning the one or more variables included in each domain. Such conceptual domains can be hypothesized a priori or left to be determined by the software. Each domain thus detected may contain one or more variables according to the number of pertinent questions that were asked in the survey instrument.
SSA maps shown later in this paper represent the entire configuration of all possible distances between each variable and each of the others – as resulting from the different response options provided by over 16,000 respondents in the 12 countries investigated. This is the largest number of Jewish respondents ever obtained in a research project undertaken in Europe. The total response obtained and the inter-country differences in consensual versus polarized perceptions of antisemitism provide the empirical evidence needed to evaluate the internal articulation and plausible structure of the topic observed.
SSA has the advantage of not being very sensitive to imperfect representation of the target population because of biased sampling. The method focuses on associations between variables rather than on their frequencies. It has no explicit pretension of searching for causal-directional relationships between variables, although it may help to establish causality hypotheses. It rather attempts to present a picture of logical interrelationships and affinities between different concepts and variables, thus refining the understanding of connections or incompatibilities between relevant domains of contents. Causal directionality of those associations will sometimes be self-evident although under other circumstances, it may remain not adjudicated. SSA’s main goal is to stimulate theory and hypothesis formulation toward a better holistic understanding of the topic studied. It may be useful to develop new typologies or to demonstrate the inconsistency of existing ones.
Each of the SSA maps that will be presented below is the equivalent of a bi-variate or multi-variate statistical table in which one dimension is the 12 countries observed, and the other dimension is the category distribution of the antisemitism-related variable (or variables) in question. Moderate rotations in the different map displays do not have analytic meaning. Some rotations were introduced in the graphics that follow to make the results more easily readable and immediately comparable. All the original tables were published in the original FRA report (FRA 2018a).
To exemplify the approach adopted in this paper, I reproduce in Table 2 one of the tables from the FRA 2018 survey report. These data constitute the basis for the elaboration of Figure 6 presented in the next section of this paper. In the figure, as will be seen, the HUDAP software (Amar and Toledano 2001) transforms the original table into a matrix where the differences between the values reported in each row – representing the different response options – are modulated according to the differences across the different columns – representing the different countries included in this study.
Modulating the categories in each variable by the 12 countries covered in the 2018 survey amounts to attributing the same weight to each country. This is obviously incorrect in terms of the size of the respective Jewish populations: the core Jewish population in the 12 counties ranged in 2018 between 450,000 in France and 4,500 in Poland (DellaPergola 2019). On the other hand, the actual final target and recipient of the antisemitic message – if any – is the individual Jew. Had we modulated the tables by the over 16,000 respondents, and not by the 12 countries, the large numerical impact of the major Jewish populations would have minimized the impact of the smaller ones, thus biasing the total picture in the direction of France, the UK and, to some extent, Germany. By choosing to present the following data unweighted by the size of a country’s Jewish population, the underlying hypothesis is that there exists a strong dependency of Jewish life on the unfolding of cognition and experiences at the national level of each country. I thus make this hypothesis explicit, while acknowledging that alternative analyses conducted at the individual respondent level might produce somewhat different results. Experience from previous research suggests that those differences would not affect the main thrust of the analysis, as in most cases inter-country perceptional differences would tend to prevail over intra-country differences.
Table 2. Assessing social and political issues as a problem, by EU Member State (%) a, b, c, d, e.
a Out of all respondents (n=16,395); country results are unweighted; 12 country average is weighted.
b Question: B02. To what extent do you think the following are a problem in [COUNTRY] (Items as listed in the table)?
c The results presented in the table are the sum of answer categories ‘a very big problem’ and ‘a fairly big problem.’
d The social issues are listed in descending order according to the average of the 12 countries.
e For each country, the three most serious problems – as assessed by the respondents – are highlighted in the table.
Source: FRA (2018a, Table 1).
Key to countries: AT: Austria; BE: Belgium; DE: Germany; DK: Denmark; ES: Spain; FR: France; HU: Hungary; IT: Italy; NE: The Netherlands; PL: Poland; SE: Sweden; UK: United Kingdom.
By the same token, each of the subsequent SSA maps in this paper was derived from one or more tables in the original FRA report (FRA 2018a). The present study indeed largely complements the original 2018 FRA report by using the same database, although differently. Rather than describing facts and frequencies, as prevalent in most research including the original report, I aim at uncovering the deeper layers of explicit and latent meaning of the antisemitic phenomenology in the European Jewish context. One original contribution of this paper is the combination of various tables into one conglomerate map, thus transforming the original bi-variate analysis into a multiple-variable analysis. After computing the correlations emerging from intercountry variation not on one but simultaneously on several variables, the product is as if the pertinent maps of the respective variables had been superimposed one over the other. This helps to detect and integrate relationships between categories and concepts originally presented as separate items. I trust the reader will find the following analysis intuitively accessible and innovative, no less enlightening than the original materials, and a useful complement to the initial publication of the FRA 2018 survey results.
Sergio Della Pergola, demografo
(continua)
(da Jewish Perceptions of Antisemitism in the European Union, 2018: A New Structural Look. Analysis of Current Trends in Antisemitism. Berlin: De Gruyter, and Jerusalem: SICSA, ACTA, 40, 2, 2020)