All time references are in CEST
Surveying Social Norms: Challenges for Survey Research |
|
| Session Organisers | Dr Ivar Krumpal (University of Leipzig) Mrs Anna Lena Fehlhaber (Leibniz University Hannover) Dr Anatol-Fiete Näher (Hasso-Plattner-Institut Potsdam) |
| Time | Tuesday 18 July, 09:00 - 10:30 |
| Room | |
Monitoring collective action as well as individual behavior and attitudes via surveys and big data has posed new challenges to the survey discipline. This session aims at presenting and discussing current survey research on social norms. We aim to discuss best practices, new challenges and innovative designs that address both methodological and substantive problems related to the emergence, enforcement, change and decay of social norms. The explanation and measurement of norm compliance / deviance are also of central interest in our session.
We invite submissions that address these issues and/or present potential solutions. We also invite applications of norm research from any discipline. In particular, we are interested in studies that (1) deal with substantive problems and applications of survey research, such as norm compliance, deviant behavior, ethical preferences in regards to allocation or trolley problems; (2) present current survey research focusing on public opinion in regards to the emergence of new social norms, values and the production of collective goods; (3) deal with methodological problems such as nonresponse, social desirability bias or sampling issues presenting innovative designs and solutions addressing these problems; (4) discuss the use of artificial intelligence in the collection and analysis of data on social norms; (5) present experimental survey research (e.g. factorial surveys, conjoint experiments, field experiments) and statistical procedures to analyze such data (e.g. conjoint analysis); (6) integrate innovative experimental designs in well-established, large-scale population surveys of the general population; (7) discuss best practices in surveying social norms.
Keywords: Social norms, social desirability bias, nonresponse bias, factorial surveys, conjoint experiments
Ms Annelie Brüning (Utrecht University) - Presenting Author
Professor Wojtek Przepiorka (Utrecht University)
Professor Tali Spiegel (Utrecht University)
Professor Tanja van der Lippe (Utrecht University)
Most survey-based studies on the normative underpinnings of gendered practices rely on measures of personal attitudes (i.e. gender roles) only. This involves the attitudes people have towards statements such as “Men should participate in housework to the same extent as women”. However, recent advancements in the measurement of norms suggest that (1) measuring social norms also requires the elicitation of people’s beliefs about what other people think and do, and that (2) these beliefs exert a greater influence on respondents’ behavior than their attitudes. Yet, there is little evidence on how the elicitations of these attitudes and beliefs in surveys affect each other. We conduct a survey experiment on a diverse sample of 1484 Dutch respondents that are employed and 24 to 65 years old. First, we explore to what extent discrepancies exist between respondents’ attitudes towards traditional gender roles and these respondents’ expectations regarding the prevalence of these attitudes and practices in their social environment. Large discrepancies would indicate that current gender role measures insufficiently capture the normative underpinnings of gendered practices. Second, we examine sequence effects – whether eliciting attitudes and beliefs after each other prompts respondents to align or distinguish their attitudes and beliefs. Third, we assess if sequence effects vary based on the closeness of the reference group used in social norm inquiries. We find respondents hold, on average, significantly less traditional gender role attitudes than they believe to be held or practiced in their social environment. Moreover, our results indicate a small sequence effect in one of the three gender spheres we studied. We find no evidence for a moderating role of the reference group on sequence effects. Our results demonstrate the need for measuring attitudes and beliefs in gender norm inquiries and point to potential caveats when measuring them in surveys.
Mr Matthias Klingler (University of Zurich) - Presenting Author
Mrs Benita Combet (University of Zurich)
Mrs Larissa Fritsch (University of Zurich)
Mrs Sandra Gilgen (University of Zurich )
In nearly all Western countries, parents’ time investment has increased over the last fifty years, with higher SES parents investing more (Altintas, 2016). Based on extensive qualitative research in the US, Annette Lareau (2011) showed that not only time investment varies by SES but that parenting practices and norms also show status differentials. More precisely, parents with lower SES emphasize providing for their children’s material needs and granting them autonomy (termed ‘accomplishment of natural growth’), while parents with higher SES focus on structuring their children’s lives, fostering their talents, and empowering them to advocate for their interests (termed ‘concerted cultivation’).
This finding has not been widely challenged, although it is based largely on qualitative research conducted in one country - the USA. The lack of further research is even more surprising given that norms are known to vary widely across countries (Begall et al., 2023).
Goal of our research is to examine the universality of the content of parenting norms. To address these research gaps, we investigate this question in a country with great cultural and institutional heterogeneity – Switzerland. In a survey experiment, we have participants rate a fictional parenting situation representing a ‘natural growth’ versus a ‘concerted cultivation’ style, following an approach similar to Ishizuka (2019, based on the US population). The survey experiment is part of a population-representative sample of individuals included in the 2nd wave of the Swiss Assisted Reproduction Longitudinal Study (CHARLS, Büchler, 2023). The aims are to examine whether parenting norms differ by individual characteristics (e.g., SES, gender), whether different norms are attributed to groups (e.g., mothers vs. fathers), and whether they vary by macro-level factors (e.g., cultural regions within Switzerland). In addition, we aim to refine the measurement of parenting norms by disentangling different components of the
Dr Martina Kroher (Leibniz University Hannover) - Presenting Author
Social norms provide a framework for the social interaction of groups and societies. Using students as an example, we analyse the norm of honesty in exams and whether or how this norm has changed over the last decade. Deviations from the honesty norm are measured at the same German university at four different points in time (2013, 2016, 2020, and 2024) in order to identify changes in the norm as well as in the violations of the norm.
The focus of the analyses is on the self-reported (mis-)behaviour in exams and seminar papers of all students enrolled at this university in the corresponding semester. The respondents had to admit if they had ever cheated in a test (using crib sheets or copying answers from neighbours) or plagiarised in any way in a seminar paper. We are aware that this direct questioning on such sensitive issues will lead to some socially desirable or dishonest answers, nevertheless this so identified lower threshold may help to shed some light on the otherwise unknown extent of this deviant behaviour.
We always conducted a full survey, i.e. every enrolled student was informed by email and invited to participate in the survey. In total, between 1200 and 2600 students took part in one of the four online surveys.
Preliminary findings indicate a general downward trend in self-admitted misconduct, with students adapting their behaviour to the advancing technological capabilities. This leads to the highest share of cheating in the use of artificial intelligence in seminar papers. The contribution will examine the changing importance of different types of academic misconduct over time, taking into account reasons for and against cheating, field of study as well as socio-demographics.
Dr Celine Wuyts (KU Leuven) - Presenting Author
Professor Geert Loosveldt (KU Leuven)
Socially desirable responding (SDR) poses a challenge in survey research, especially when social norms influence how respondents answer questions about sensitive or socially valued traits. While direct measures of SDR have been developed to detect impression management ("faking-good"), they often require additional scales, increasing survey length and response burden. Indirect measures, particularly those derived from existing personality scales, offer a promising solution, yet their effectiveness remains underexplored. One avenue for further exploration is the use of existing scales such as the 21-item Schwartz Portrait Values Questionnaire (PVQ-21), which is included in the European Social Survey core questionnaire and measures ten basic value orientations. Values like fairness and kindness are culturally esteemed and therefore prone to over-reporting, making them suitable for use as indirect measures of SDR. This study explores the potential of using the PVQ-21 to extract such indirect SDR measures and compares their effectiveness in explaining response patterns with a direct SDR measure based on the Marlowe-Crowne social conformity scale.
Specifically, this study evaluates indirect measures of SDR based on item response models and simple proportions of desirable scale endpoint responses. Using data from the PVQ-21 in Belgium Round 8, we validate these indirect measures against a direct SDR measure included in the country-specific questionnaire. We also explore patterns of these indirect measures across other rounds and countries.
Our analysis addresses three key research questions: (1) How predictive are indirect measures of impression management as measured by a direct impression management scale? (2) How do indirect SDR measures vary across rounds and countries? (2) What is the relationship between SDR measures and attitudes towards income redistribution, immigration, and homosexuality, indicating prevailing social norms? This study contributes to the literature by enhancing our understanding of social norms across different political contexts.
Miss Ladin Toplu (Bogazici University) - Presenting Author
In Turkey, attitudes toward abortion have traditionally been measured using a 0-10 acceptability scale in values surveys and the importance of context in shaping abortion attitudes has long been neglected. Notably, context-dependent abortion attitudes in Turkey were last measured in the European Values Survey in 2008, despite significant political and societal shifts since then. The current hostile rhetoric of political leaders toward abortion in Turkey and the global rise of far-right ideologies, and the accompanying restrictive policies highlight the need for updated data.
To address this gap, Turkish Social Values Study includes three novel context-based questions: (1) approval of abortion for an unmarried woman, (2) approval of abortion for a married couple, and (3) approval of abortion when a woman’s health is at risk. Additionally, the survey introduces an unprecedented question that measures whether the respondents agree or disagree with the following statement: “Abortion should be easily accessible to everyone”. The aim of this new variable is to measure public opinion on the accessibility of abortion, which has never been systematically assessed in Turkey before. Hence, this research focuses on the following question: “How do the demographic determinants of attitudes toward abortion change across different scenarios? What are the determinants of being for or against accessible abortion in Turkey?”.
This novel approach responds to the need for data in a sociopolitical climate where abortion debates are increasingly polarized. Furthermore, it aims to go beyond binary measures and capture nuanced attitudes that reflect real-life complexities more accurately and contribute to a more thorough understanding of reproductive autonomy. The findings have the potential to inform policies that advocate for universal abortion access and contribute to the broader discourse on reproductive justice in Turkey and beyond.
Ms Emma Zaal (University of Groningen) - Presenting Author
Dr Yfke Ongena (University of Groningen)
Professor John Hoeks (University of Groningen)
Social Desirability Bias (SDB), the tendency to provide a socially favorable image in self-reports, is a challenge when investigating sensitive topics in surveys. SDB results in underreporting norm-violating behavior (e.g., substance use) and overreporting norm-conforming behavior (e.g., exercising). To address this issue, we conducted an experimental study in which we employed so-called face-saving strategies to reduce SDB. A face-saving approach aims to ease traditional norms by reassuring respondents that deviating from typical or expected behaviors is sometimes unavoidable, thereby reducing the discomfort associated with acknowledging such actions. We conducted a survey in two neighborhoods in Groningen, the Netherlands, which featured questions prone to SDB, such as on willingness to volunteer and assist neighbors, as well as on perceived social cohesion, disturbances, safety, and financial difficulties.
We included a face-saving preamble aimed at normalizing the potential unpopular opinion of having difficulties living alongside individuals from different ethnic backgrounds. This preamble employed downtowning, phrasing that aims to weaken the strength of a statement or social norm (e.g., “It can sometimes be challenging to [...] It’s also very normal to find this difficult.”). Additionally, we offered face-saving answer options, for instance by incorporating degrees of truth (answer options reflecting partial agreement, like “occasionally”) and credentialing (offering justifications for norm-noncompliance, like “no, but I don’t have time”). Our findings revealed that participants were more likely to provide socially undesirable responses when presented with face-saving options compared to a traditional yes/no format, while a preamble did not affect SDB. Our results replicate results of other recent studies demonstrating the effectiveness of face-saving in reducing SDB. We explored face-saving in other question formats and operationalizations as well, offering deeper insights into the conditions under which face-saving is effective.
Ms Monica Gerber (Universidad Diego Portales) - Presenting Author
Mr Cristóbal Moya (DIW Berlin / Universität Bielefeld)
Measuring participation in violent collective action through surveys presents significant challenges. People may be reluctant to report engaging in such behavior due to concerns about legal consequences or because it is socially undesirable to admit. However, in Chile, following the 2019 social uprising, there was widespread public support for the use of violence as a means of protesting against perceived social injustice and inequality.
We conducted a survey after the uprising (n = 2594), asking respondents whether they had engaged in violent protest actions and whether they justified such actions. Surprisingly, 11% reported verbally abusing the police, 7% reported participating in street blockages, and 3% reported throwing stones or bottles during a demonstration. These findings offer valuable insight into why individuals may disclose violent behavior despite the potential risks of self-reporting.
To further explore these patterns, we conducted a latent class analysis combining both self-reported violent actions and support for such actions. Three distinct classes emerged: (1) those who participated in and supported violence, (2) those who neither participated in nor supported violence, and (3) those with minimal participation but high support for violence. The third group is particularly interesting, as it suggests the possibility that some individuals in this category may have engaged in violence but refrained from admitting it.
We then explored the variables predicting membership in these classes. Individuals who participated in or supported violent collective action exhibited lower levels of police legitimacy and higher exposure to police repression. Notably, both police legitimacy and repression played a key role in differentiating between those who merely supported violence and those who reported direct involvement in it. These findings raise new questions about whether self-reporting violent actions in surveys becomes more likely in highly politicized contexts where distrust toward the targeted group is widely shared.
Dr Marija Radoman (Research Associate, Faculty of Philosophy, University of Belgrade) - Presenting Author
This research seeks to explore how societal norms, context, and traditional gender roles permeate psychotherapeutic education, potentially affecting both training and therapeutic processes. The central research questions of this study include examining how societal attitudes are reflected in therapist education and to what extent they influence therapists’ approaches to gender issues. The study uses a mixed-methods approach to explore the integration of feminist knowledge in psychotherapy. It examines whether feminist concepts such as gender roles and patriarchal attitudes can be effectively operationalized within the field, and how quantitative and qualitative methods provide complementary or contradictory results regarding these issues.
The quantitative component involves an online survey of therapists from post-Yugoslav countries to assess the prevalence of gender stereotypes and patriarchal views in practice. Although this approach provides a broad understanding of therapists' attitudes, it struggles to capture the deeper, subjective factors influencing therapy, particularly regarding gender identity and power. The qualitative part of the research, based on interviews with female psychotherapists, offers richer insights into personal narratives and subjective interpretations. While it provides a deeper understanding of how therapists construct and respond to gender norms, it lacks generalizability due to the smaller sample size.
By combining both approaches, this study highlights the strengths and limitations of each methodology. The quantitative data offer a wider statistical overview, while the qualitative interviews provide a more nuanced understanding of how feminist concepts are operationalized in therapeutic practice. Ultimately, the study reflects on the effectiveness of these methodologies in capturing the complexities of feminist concepts in psychotherapy.
Miss Joana Nunes (ISCTE-IUL) - Presenting Author
Ms Sven Waldzus (ISCTE-IUL)
Collective action (CA) is a driver of social change starting at grassroots level. Previous studies have established the role of identification, morality, injustice and efficacy (SIMCA model) in predicting CA. However, the role of social norm adherence and social influence have been overlooked. In two experiments, descriptive norms were manipulated in the context of the housing movement in Portugal. In Experiment 1 (N=235), participants responded to the survey both in an online and controlled environment and were randomly assigned to two distinct conditions. They were shown a fictitious text about the high vs low participation rates of a reference group. Experiment 2 (N=330) had a similar design, except that a control group was added and participants responded to the survey only online. After the presentation of descriptive norms information, we incorporated a self-persuasion technique, where individuals were asked to type a text about the reasons behind high/low participation according to the condition assigned, providing in the same methodology both quantitative and qualitative components. In the two studies, the manipulation was effective, given that participants in the high participation norm condition estimated that their close social network also participated more in the housing movement (measured norms). However, contrary to our predictions, there was no total effect of the manipulated descriptive norms on participants’ intentions to engage in CA. Nevertheless, the whole SIMCA model mediated the relationship between descriptive norms and CA intentions measures. Specific distinct significant indirect effects were found in the two studies. As the reference group affected by grievances differed between the two studies (young, Experiment 1; Portuguese, Experiment 2), we suggest that the relevance attributed to specific SIMCA factors is context and social identity dependent. This study provides both relevant theoretical and methodological knowledge about the role of descriptive norms.
Dr Katrin Drasch (FAU Erlangen-Nürnberg) - Presenting Author
This presentation discusses lessons insights from a pilot study using a factorial survey to examine third person judgments about when an older person should be placed in a retirement home. It assesses societal attitudes towards institutional elderly care and the trade-off between freedom and security while addressing the social norm of "retirement home as last resort." The theoretical framework is based on a Rational Choice approach and the theory of successful aging (Steverink et al., 1998), incorporating an age gradient (Ormel et al., 1999). A web-based factorial survey was developed following Auspurg and Hinz's (2015) recommendations and recent methodological considerations (e.g., Treischl & Wolbring, 2022). The survey included D-optimal balanced dimensions at the vignette level (resolution IV-design; D-efficiency 93.9) and varied individual characteristics and personal circumstances, such as social embeddedness and care availability, across a high- and low-cost scenarios. The dependent variable measured participants' recommendation for a described person moves into a retirement home on a 7-point Likert scale. We found that several of the theoretical mechanisms indeed apply but the explanatory power and more important, the difference between recommendations for the fittest person as compared to the most fragile person are small. This makes us conclude that in German society, the reputation of care homes seems really poor and the Covid situation made this even worse. While aiming to assess behavioral intentions to infer broader societal attitudes, the overall unsatisfactory results lead us to question whether such inferences to the societal macro level are plausible due to differing aggregation processes influenced by social interdependencies or whether such inferences should better be avoided.
Dr Vittoria Offeddu (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan)
Dr Maria Cucciniello (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan; Department of Social and Political Sciences, Bocconi University, Milan)
Dr Elisabetta Colosi (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan)
Dr Lorenzo Lucchini (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan; BIDSA Bocconi Institute for Data Science and Analytics, Bocconi University, Milan)
Miss Laura Leone (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy; Nuffield College, University of Oxford, Oxford; Department of Sociology, University of Oxford, Oxford)
Dr Duilio Balsamo (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy; BIDSA Bocconi Institute for Data Science and Analytics, Bocconi University, Milan)
Dr Francesco Bonacina (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan)
Dr Chiara Chiavenna (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan; BIDSA Bocconi Institute for Data Science and Analytics, Bocconi University, Milan)
Miss Elena D'Agnese (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan; Department of Economics and Management, University of Pisa, Pisa; Department of Statistics, Computer Science, Applications, University of Florence, Florence) - Presenting Author
Professor Alessia Melegaro (DONDENA Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan; BIDSA Bocconi Institute for Data Science and Analytics, Bocconi University, Milan)
Recent epidemics have shown that human behaviour can significantly affect the emergence and the transmission of infectious diseases. However, current epidemic modelling approaches are suboptimally equipped to incorporate behavioural components, because they are seldom grounded on behavioural science theory, and behavioural survey data is often unsuitable to directly inform model design.
In this work, we propose an interdisciplinary framework for the collection of behavioural survey data aimed at parametrizing epidemic models.
Between March and July 2024, we administered a cross-sectional survey to representative samples of adults of six European countries to investigate factors potentially influencing preventive behaviours under different epidemic conditions. We embedded validated behavioural theory into epidemic model design from the outset, tailoring the behavioural data collection to specific modelling needs. Before the full launch, the questionnaire was evaluated by experts from different scientific disciplines and the general public to ensure an appropriate balance between the validity and functionality of the collected behavioural data and its applicability to epidemic modelling.
The collected data will allow us to amplify traditional modelling methodologies. For instance, by expanding the existing definition of “social contacts” beyond in-person interactions to include discussion contacts, and assessing not only respondents’ opinions on multiple topics, but also how they compared to those of their social circles, we constructed discussion matrices to inform how interactions influence individual decision-making processes preceding behavioural choices. Our data also allows us to identify the psychological antecedents of behaviour, e.g. willingness to vaccinate, and to account for changes in willingness over time, capturing its dynamic relationship with changing epidemic conditions.
The proposed approach provides critical insights for incorporating behavioural factors into epidemic models through data-driven approaches, increasing our ability to design novel computational epidemic models that provide a more accurate and thus more policy-relevant representation of reality.
Dr Alisa Remizova (GESIS – Leibniz Institute for the Social Sciences) - Presenting Author
Dr Maksim Rudnev (University of Waterloo)
Measuring religiosity in the European Social Survey: Cross-temporal and cross-country measurement invariance
Secularization theory suggests that religiosity declines over time, yet the study of religious change across countries faces a critical challenge: the potential noninvariance of religiosity measurement. Previous research has indicated that measures of religiosity are culturally specific, raising doubts about the universality of religiosity and its measurement across diverse countries. Nevertheless, it was demonstrated that religiosity measures can achieve approximate invariance across some European countries. However, whether these measures are invariant not only across countries but also over time has yet to be investigated. Without temporal invariance, observed changes in religiosity may reflect shifts in the understanding of religiosity and its measurement, rather than true societal trends, thereby undermining the validity of comparisons over time.
This study examines the measurement invariance of religious involvement (RI)—captured by the frequency of an individual's religious attendance, frequency of praying, and self-assessed religiosity indicators—across 11 rounds of the European Social Survey (2002–2024). Using a one-factor model of RI and the alignment method, we find that while exact invariance across rounds is largely absent, approximate invariance holds in all countries, providing reliable, aligned factor means for assessing trends in RI in European countries over time. Further analysis reveals patterns of religious decline across Europe, while some countries exhibit stable or even increasing levels of RI, diverging from the overall trend. Additionally, we discuss the polarization of countries based on the varying dynamics of their RI.
Mr Peer Keßler (University of Greifswald, Institute of Psychology, Department Health and Prevention) - Presenting Author
Mr Anatol-Fiete Näher (University of Potsdam, Hasso Plattner Institute for Digital Engineering)
Mr Ivar Krumpal (Leipzig University, Institute of Sociology)
The COVID-19 pandemic has intensified the debate about fairness in medical allocation decisions. Emergencies require careful balancing of highly conflicting normative principles. Triage decisions involve criteria such as maximizing treatment benefits, treating people equally, prioritizing those worst off, and promoting societal value.
Emanuel et al. (2020) argue for maximizing treatment benefits as the leading criterion. More specifically, they argue to choose the persons with the highest chance of survival and the most life-years ahead, using a random lottery only when patients are equal in these dimensions. This approach aligns with a rank-ordering solution to triage (Veatch 1995), ranking treatment benefits over equal treatment.
In democratic societies, criteria on ethical decisions must at least coincide with the convictions of the general population to ensure public trust in institutions and social cohesion. Measuring preferences in ethical dilemma situations can give insights into the perceived fairness of ethical decisions and guidelines.
Despite extensive research on ethical preferences in triage, no study, to our knowledge, has directly examined preferences aligned with the rank-ordering approach recommended by Emanuel et al. (2020). To this end, we conducted a conjoint experiment with a sample of a broad international public (N=1,998). Participants were asked to fictionally allocate a respiratory ventilator to one of two patients in need that would otherwise not survive. The patients were represented by randomly generated profiles. Additionally, participants could decide to allocate the respiratory ventilator by random chance.
We find that participants choose an allocation by random chance more likely when the patients in question have equal chances of survival and are equal in age. Even though we find similar but less pronounced effects for other attributes, our results speak in favor of preferences for a rank-ordering approach to triage.
Mr Guido Priem (KU Leuven) - Presenting Author
Dr David Kretschmer (University of Oxford)
Professor Eva Jaspers (Utrecht University)
When is a religious expression acceptable, and when is it intolerable? With religious diversity being increasingly a topic of debate in Western European societies, it becomes important to study where the public draws this boundary. While existing experimental research has highlighted a range of factors that shape tolerance towards different religious practices, the investigation of these determinants is scattered over a range of studies. An important question, therefore, is which contextual factors influence people's decisions to tolerate or reject religious expressions when other relevant factors are taken into account. Moreover, these studies focus on average effects, potentially obscuring subgroup variations. This raises a second question: How do individual characteristics shape the impact of these attributes on tolerance for religious expressions?
This study aims to address these issues by using a single-profile conjoint experiment embedded in the LISS-panel infrastructure. We present respondents to a series of instances of religious speakers speaking to an audience and ask respondents to judge if they should be allowed to speak or not. We randomly vary who is speaking, what is said, and where it takes place, allowing us to separately estimate the impact of each attribute on decisions to tolerate religious expressions. To detect individual differences, we use a novel machine learning method based on Bayesian Additive Regression Trees (BART) to systematically explore how the dominant attribute in the decision varies according to a range of socio-political characteristics included as pre-existing data in the LISS panel. Leveraging the breadth of pre-existing data in the LISS panel, we rely on an extensive list of background information on the respondents to asses this heterogeneity. In doing this, this study is the first to model how the interplay between individual characteristics and contextual attributes shapes tolerance decisions about religious expressions.