To accept or not to accept: Level of moral concern impacts on tolerance of Muslim minority practices.

Living with diversity requires that we sometimes accept outgroup practices that we personally disapprove of (i.e., tolerance). Using an experimental design, we examined Dutch majority group members' tolerance of controversial practices with varying degrees of moral concern, performed by a culturally dissimilar (Muslims) or similar (orthodox Protestant) minority group. Furthermore, we examined whether arguments in favour or against (or a combination of both) the specific practice impacted tolerance. Results indicated that participants expressed less tolerance for provocative practices when it was associated with Muslims than orthodox Protestants, but not when such practices elicit high degrees of moral concern. This indicates that opposition towards specific practices is not just a question of dislike of Muslims, but can involve disapproval of specific practices. Argument framing did not have a consistent effect on the level of tolerance for the practices.

Background

In many Western societies, the hotly debated issues around cultural diversity boil down to concrete practices and behaviours of ethnic and religious minority groups like Muslims (Cesari, 2013). Should we tolerate Muslim teachers’ refusal to shake hands with parents of the opposite sex? Should we allow civil servants to wear a headscarf to work? Should a city or state permit the founding of Islamic primary schools? It is around these concrete questions that multiculturalism and living with diversity is truly put to the test, and ways of life collide. It is one thing to endorse freedom of speech, but another thing to apply this freedom to a religious leader calling gays and lesbians inferior people. And it is one thing to endorse freedom of education, but another to accept religious schools teaching children undemocratic principles. Yet, tolerance for dissenting beliefs and practices is a key condition for citizenship and democracy (Sullivan & Transue, 1999). Tolerance is not the absence of prejudice, but rather refers to the acceptance of norms or practices that one considers wrong, but not beyond bearing (Verkuyten & Yogeeswaran, 2017).

While people can reject certain practices because of their prejudicial feelings towards a group, they might also be intolerant of that kind of practice more broadly. For example, one can resist the idea of Muslims establishing an Islamic primary school because one feels negatively towards Muslims, or because one thinks that religion has no place in education entirely. Thus, people can object to a particular practice of a group because they dislike the group or because they disapprove of the practice itself. Group distinctions might be irrelevant for tolerance of highly offensive acts because such acts are rejected under all circumstances and independent of who is doing it (generic tolerance; Hurwitz & Mondak, 2002). However, people may be more tolerant of practices of groups that are more aligned with the national majority than a minority group engaging in the same action when the act is of less moral concern (discriminatory tolerance).

In the present work, we examine whether majority members are willing to tolerate particular controversial practices of Muslim minorities (a culturally dissimilar group) relative to orthodox Protestants (a culturally similar group). Our aim was to go beyond the existing research by testing whether there is greater tolerance for provocative practices when they are carried out by dissimilar relative to similar outgroup members regardless of the level of moral concern evoked by such practices, or whether such differences would disappear when the practice involves strong moral concern. While previous research has focused on majority group members (in)tolerance of specific practices performed by Muslims (e.g., Saroglou, Lamkaddem, Van Pachterbeke, & Buxant, 2009; Van der Nol, 2014; Verkuyten & Slooter, 2007), it has not considered the possibility that similar reactions are elicited when such practices are performed by a more culturally similar group. Furthermore, existing research has not examined whether toleration depends on reasons for or against acceptance (but see Gieling, Thijs, & Verkuyten, 2012). Here, we explored whether pro, contra, or dual arguments for a specific practice further moderated the impact of group membership and type of practice on tolerance examining the impact of framing effects on tolerance. We examined these issues using vignettes in an experiment design involving a large national sample from the Netherlands, which is one of the most secular countries in the world.

Moral spectrum and type of controversial practice

Skitka, Bauman, and Sargis (2005) proposed that objections towards particular practices can differ in the extent to which moral concerns are involved (see also Rozin, 1999). Moralized entities and activities tend to lead to avoidance and rejection rather than toleration (Ellemers, 2017). In contrast to subjective preferences and social conventions (Turiel, 2002), people tend to believe that matters of morality are objective, universally true, and thereby applicable regardless of group boundaries (Skitka, Bauman, & Mullen, 2008). Things that are considered right are simply right and what is wrong is wrong, independent of who is doing it (Turiel, 2002). If, for example, one has a strong moral conviction that gender inequality is wrong, one is likely to believe that gender inequality is wrong in all cultures and religions. With moral issues, people focus more on their principles and ideals rather than on authorities, group differences, and social identities (Skitka & Morgan, 2014).

Researchers from different fields have shown that people are less accepting of divergent beliefs and practices that are viewed as moral issues, and that acceptance for moral issues is less context‐sensitive than for nonmoral issues (e.g., Cole Wright, Cullum, & Schwab, 2008; Ellemers, 2017; Skitka et al., 2005). Seeing an issue as (im)moral tends to result in greater rejection, independent of the moral emotions that might be involved and relatively independent of the context (Cole Wright et al., 2008; see also Wainryb, Shaw, Laupa, & Smith, 2001; Wainryb, Shaw, & Maianu, 1998). For example, early adolescents who endorse the value of cultural diversity tend to tolerate practices that raise subjective and conventional considerations, but are less accepting of practices that are perceived as moral transgression (Gieling, Thijs, & Verkuyten, 2010). These findings suggest that people may show varying degrees of tolerance towards a culturally similar versus dissimilar groups’ practices depending on the degree of moral concern evoked by the practice. Specifically, we expect that if a practice strongly raises moral concerns, people will not accept the practice regardless of who enacts it. However, for practices moralized less, majority members will be less tolerant of a culturally dissimilar group (Muslims in this work) engaging in the practice compared to more culturally similar group (orthodox Protestants in this work).

Framing of controversial practices

Tolerance involves a trade‐off between reasons to accept dissenting norms and practices and reasons to reject them (Verkuyten & Yogeeswaran, 2017). Reasons to accept norms and practices are often linked to civil liberties. For example, freedom of speech is an important argument for accepting hate speech against a minority group. Thus, one way in which people's tolerance of dissenting practices might be increased is by framing these acts in terms of civil liberties. However, questions of tolerance are multi‐faceted and typically evoke oppositional arguments as well. Reasons to reject specific practices often revolve around concerns about harm and social order (Gieling et al., 2012). For example, people's attitudes towards hate speech against minority groups such as gays and lesbians may not only be based on considerations of freedom of speech, but also considerations about the offence and harm towards targeted individuals. Additionally, people are often not just exposed to one set of considerations about an issue or problem, but rather to competing arguments (Sniderman & Theriault, 2004). A number of studies have shown that the impact of any given argument on individuals’ opinions is neutralized by the introduction of a competing argument that rebuts the first (e.g., Brewer & Gross, 2005; Sniderman & Theriault, 2004). For example, one study indicated that the level of rejection of ethnic‐specific school policies increased when these policies were framed in terms of the cost for the majority population, in comparison with when no such frame was used (Van Londen, Coenders, & Scheepers, 2010). However, when the emphasis on costs was combined with an argument about the positive implications of educating ethnic minorities, the level of rejection of the policies was not different than in the condition in which no arguments were given (but see Gieling et al., 2012). Thus, the effect of one consideration appeared to be neutralized by introducing a counter‐consideration.

In the present study, we did not make specific predictions about the impact of arguments on tolerance for various religious practices, but rather explored whether tolerance of controversial practices depended on providing a pro, contra, or dual argument for tolerating the practice. The central assumption in framing research is that people are sensitive and responsive to different, often contradictory, considerations about an issue, making their opinion dependent on the way that the issue is presented (Chong, 1993; Nelson & Kinder, 1996). By framing an issue in a particular way, attention is directed towards positive or more negative considerations that affect people's views about the issue (e.g., affirmative action, welfare policies, and civil liberties; Brewer & Gross, 2005; Druckman & Nelson, 2003; Nelson, Clawson, & Oxley, 1997). However, framing effects are particularly likely when people have somewhat ambiguous or relatively weak feelings about a particular issue. In contrast, strong feelings and attaching personal importance to an attitude make it resistant to change (Howe & Krosnick, 2017). Attitudes tend to be strong and therefore difficult to change when they reflect moral values and convictions. Research has shown that important attitudes are resistant to framing effects (Lecheler, de Vreese, & Slothuus, 2009). For example, students who attached more importance to their anti‐egalitarian attitudes at the start of a course on gender equality were more likely to successfully resist changing their attitudes towards equal rights as a result of taking the course (Sevelius & Stake, 2003).

Context of the present research

We examined the level of tolerance people express towards provocative practices when they are committed by members of two groups that vary on a continuum of cultural similarity: Muslims and orthodox Protestants (both representing around 5% of the Dutch population). In general, people are expected to show higher levels of tolerance towards orthodox Protestants than Muslims engaging in a controversial practice they disapprove of because the former are less likely to be perceived as a threat to national norms, values, and beliefs (Wirtz, van der Pligt, & Doosje, 2016). First, orthodox Protestants have Dutch ancestry, which is not the case for Muslims. Second, Christianity, and Protestantism in particular, is considered a historical component of Dutch culture and society, whereas Islam is often perceived as incompatible with Dutch national identity (Sniderman & Hagendoorn, 2007). Research has shown that even non‐believing Dutch adolescents express less negative attitudes towards Christians compared to Muslims (Verkuyten & Thijs, 2010), suggesting that such differential reactions do not simply refer to religious in‐group bias. Furthermore, Dutch natives are more strongly opposed to political representation of Muslim immigrants from Turkey as compared to Christian immigrants from Turkey (Verkuyten, Hindriks, & Coenders, 2016), indicating that religious group membership is important in people's judgements. Tolerance is thus expected to be lower when specific practices are performed by Muslims as compared to orthodox Protestants.

However, as stated earlier, we expect tolerance of specific practices to further depend on the extent to which a particular practice evokes moral concern. Specifically, we focused on tolerance of three practices that have been demonstrated to evoke varying degrees of moral concern among Dutch majority members (Gieling et al., 2010; Van Doorn, 2015; Verkuyten & Slooter, 2007, 2008): the founding of separate religious schools, the exclusion of women from religious boards, and the public expression of views that gays and lesbians are inferior people. Calling gays and lesbians inferior people is found to be most strongly considered a moral issue because it causes psychological harm to the people concerned. The exclusion of females was considered less harmful, but goes against the principle of gender equality, while the founding of religious schools mostly raised concerns about social integration. The general expectation is that tolerance will be lowest for the statement denigrating gays and lesbians, followed by the exclusion of women and then the founding of religious schools, as the latter are the least moralized practices. Furthermore, since moral criteria are considered absolute, the group membership of the one performing the disapproved moral act should not matter. Thus, we expected individuals to be (in)tolerant of the statement denigrating gays and lesbians irrespective whether it is made by a very dissimilar group like Muslims or a culturally similar group like orthodox Protestants (generic tolerance). In contrast, we expected individuals to be less tolerant of Muslims (relative to orthodox Protestants), in particular when it comes to the founding of religious schools as it is of the least moral concern (discriminatory tolerance).

Method

Sample and procedure

The data were collected in May 2016 by the research institute I&O that maintains a panel that is representative for the Dutch population in terms of age, gender, education, and province of residence. Panellists were randomly invited via email to participate in an online questionnaire of a large Dutch University in which we were able to embed a between‐subjects experiment. Sample size was determined before data collection using the statistical software G*Power (Faul, Erdfelder, Buchner, & Lang, 2009).1 In total, 1,225 Dutch majority group members participated. The majority of the participants (53.3%) did not belong to any religious community and participants who affiliated with a religious group mostly belonged to a non‐orthodox Protestant denomination (24.2%), followed by Catholics (19.9%). Men were slightly overrepresented in the sample (57.1% male) and ages ranged from 18 to 90 years (M = 56.53, SD = 12.82). The political self‐placement scale (Jost, 2006) was presented at the end of the questionnaire: left (18%), centre‐left (18%), centre (33%), centre‐right (16%), or right (15%).

Design and materials

Following previous research, three vignettes were used about specific practices that in recent years have caused considerable public debate in Dutch society: founding of religious primary schools, exclusion of women from religious boards, and homophobic statements by religious authorities. Importantly, these practices have been debated in relation to both Muslims and orthodox Protestants and have been found to differ in terms of the moral concerns that they raise (Gieling et al., 2010, 2012; Van Doorn, 2015; Verkuyten & Slooter, 2007). For example in one study, it was found that around four in five adolescents considered the homophobic statement as wrong and harmful, whereas around one in three considered the founding of Islamic schools to be wrong (Gieling et al., 2010). Furthermore, in a post‐hoc test on a national sample (N = 218) and following Skitka et al. (2005), respondents were presented with a single‐item measure that directly asked to what extent their moral principles formed the basis of their views on each of the three practices. Pairwise comparisons indicated that the homophobic statement was rated as significantly (p s < .001) more moral (M = 6.62, SD = .78) compared to the other two and that the exclusion of women (M = 6.21, SD = 1.18) was considered significantly more moral than the founding of Islamic schools (M = 5.53, SD = 1.43), F(3, 215) = 6.32, p < .001 (Pillai's).

Next to the scenario, the vignettes included arguments referring to core values such as social order and civil liberties. In the ‘contra argument’ condition, scenarios were framed in terms of social order or equality, whereas in the ‘pro argument’ condition, the scenario was framed in terms of civil liberties. Both values were presented together in the ‘mixed argument’ condition, and the control condition did not include any references to values. Furthermore, the actors mentioned in the scenarios were either orthodox Protestant or Muslim, resulting in a 2 (Target Group: orthodox Protestant vs. Muslim; between‐subjects) × 4 (Framing: Pro argument vs. contra argument vs. mixed arguments vs. no argument; between‐subjects) × 3 (Practice: founding of religious schools vs. exclusion of women vs. homophobic statement; within‐subjects) design. The scenarios were as follows:

(Freedom of education is very important in our society [pro]. It is very important that children can integrate in society and this is best achieved with mixed schools [contra].) Some organizations are founding [Islamic/strict Protestant] schools where only [Muslim/orthodox Protestant] children are accepted. What should the government do about these schools?

(Freedom of speech is very important in our society [pro]. It is very important not to hurt or insult other people groundlessly [contra].) A(n) [Imam/priest of an orthodox Protestant church] recently gave a speech in which he stated that gays and lesbians are inferior human beings. What should the [Mosque/Church] administration do about it?

(Freedom of assembly is very important in our society [pro]. Equality of man and woman is a crucial value [contra].) Recently, there was a discussion about someone who was not accepted in the administration of a [Mosque/strict Protestant church] because she is a woman. What should the congregation do about it?

Measures

Tolerance

After reading each vignette and based on previous research (see Gieling et al., 2010; Verkuyten & Slooter, 2007), participants indicated their tolerance for the scenario choosing between the following possible actions of the responsible authorities: (1) ‘do nothing and simply allow it’, (2) ‘try to convince them not to do it, but allow it if they don't agree’, (3) ‘in between’, (4) ‘try to convince them not to do it, but forbid it if they don't agree’, and (5) ‘simply forbid it’. We focused on participants’ view about what the responsible authorities should do because the exercise of tolerance presupposes the power to interfere and participants themselves are not in a position to do so but they can appeal to authorities to act (Horton, 1996; Verkuyten & Yogeeswaran, 2017). The scales were subsequently reversed so higher values reflected more tolerance. Although strictly speaking the scale consisted of five answer categories that are logically arranged in a meaningful order, we treated it as a continuous measure. In two experimental studies using similar vignettes and the same response scale, it was found that the findings are similar when using a continuous or ordinal measurement scale approach (Sleijper, 2018). Participants also indicated their feelings towards each scenario (−3 ‘very negative’ to 3 ‘very positive’), which allows us to investigate whether the reactions to the scenarios are indeed a matter of tolerance (i.e., acceptance of disapproved practices) and whether the three scenarios do indeed differ in the negative feelings that they elicit.

Method of analysis

Preliminary analyses and randomization checks were performed to determine the necessity of control variables. Randomization checks suggested to control for Protestant affiliation because Protestants were not equally distributed across the eight experimental conditions, Pearson χ2(7) = 14.10, p = .05. There were no differences (p s > .50) between experimental conditions in the distribution of gender, age, education, and political self‐placement. Thus, only the variable Protestant affiliation (1 vs. not Protestant = 0) and its interaction with target group condition were included as control variable.

We first performed a between‐subjects multivariate regression analysis on tolerance ratings for each one of the three scenarios. Independent variables were target group and framing conditions. Since there were no missing values in the data, multivariate regression analysis is an appropriate method of analysis because it does not assume sphericity and yields high power as well as low chances of type I error for complete data (Gueorguieva & Krystal, 2004). Furthermore, multivariate regression analysis may be performed despite the skewed distribution of the dependent variables as it is relatively robust against violations of multivariate normality (Finch, 2005). We examined the assumption of homogeneity of variance–covariance matrices with a Box M test that yielded significant results, F(42, 2374510.7) = 3.53, p < .001. The violation of the assumption was accepted, as cells with larger sample sizes mostly produced larger variances and covariances than cells with smaller sample sizes yielding a conservative estimation of the alpha level (Tabachnick & Fidell, 2014).2 We compared coefficients across tolerance outcomes using the Wald tests (Kodde & Palm, 1986).

Results

Descriptive findings

We first examined whether participants generally felt negative about the practices depicted in the three scenarios. About three quarters of the participants expressed negative feelings about founding of religious schools (M = −1.47, SD = 1.43) and exclusion of women (M = −1.92; SD = 1.40). Concerning the homophobic statement, negative feelings were even more widespread with 94.5% of the participants disapproving such statements (M = −2.44, SD = .91). For all three scenarios, mean scores of feelings were well below the neutral zero mid‐point of the scale, p s < .001, and even below the score of −1, p s < .001. Feelings towards the three practices were moderately correlated (r = 0.41 to 0.48).3

Explaining tolerance

Average tolerance varied considerably across practices (see Table 1). A series of Wald tests revealed that the intercepts (see Table 2, columns 2–4) differed significantly from each other for all pairwise comparisons, F(1, 1,218) > 21.02, p < .001. As expected, participants were least tolerant of a homophobic statement and most tolerant of founding religious schools, with the exclusion of women in between. Further, results indicate a significant overall effect of target group, F(3, 1,216) = 54.68, p < .001, ηp 2 = .12, with practices being tolerated less when committed by Muslims compared to orthodox Protestants (see Table 1). As expected, tolerance ratings were significantly lower when religious schools were founded by Muslims compared to orthodox Protestants, whereas tolerance ratings of a homophobic statement (moral principle of harm) and the exclusion of women (principle of equality) did not differ significantly between Muslims and orthodox Protestants (see Table 2, column 2–4 for results of the multivariate regression estimating effects on each tolerance outcome, separately).

Table 1

Tolerance ratings by experimental conditions and practices

	Religious schools		Exclusion of women		Homophobic statement		N
	M	SD	M	SD	M	SD
Overall	2.407	1.390	2.308	1.381	1.553	.934	1,225
Target group condition
Protestant	2.85	1.44	2.29	1.44	1.65	1.06	621
Muslim	1.95	1.77	2.32	1.31	1.45	.78	604
Framing condition
Control	2.41	1.41	2.41	1.44	1.45	.90	315
Pro argument	2.41	1.41	2.28	1.38	1.61	.95	306
Contra argument	2.38	1.37	2.14	1.30	1.51	.87	288
Both arguments	2.43	1.38	2.38	1.38	1.63	1.00	316
Religion
Protestant affiliation	2.886	1.443	2.882	1.446	1.795	1.104	297
No or other affiliation	2.253	1.338	2.124	1.308	1.476	.860	928

Table 2

Multivariate analysis of variance (MANOVA, column 1) and its underlying regression model with tolerance ratings for scenarios as dependent variables (column 2–4), N = 1,225

	Multivariate factor of tolerance ratings (1)		Tolerance for specific practices
			Founding of religious schools (2)		Exclusion of Women (3)		Homophobic Statement (4)
	Pillai's V	F	B	SE	B	SE	B	SE
Target group	.119	54.68***
Protestant (ref. cat.)
Muslim			−.710***	.085	.160	.088	−.084	.060
Framing condition	.014	1.95*
Control (ref. cat.)
Pro argument			−.014	.104	−.122	.108	.163*	.074
Contra argument			−.100	.105	−.278*	.109	.044	.075
Both arguments			−.066	.103	−.067	.107	.152*	.073
Religious affiliation	.067	29.24***
Non‐protestant (ref. cat.)
Protestant affiliation			.893***	.117	.969***	.122	490***	.083
Religion * target group	.018	7.22***
Protestant A. *Muslim			−.681***	.173	−.441**	.180	−.412*	.123
Intercept			2.657***	.088	2.154***	.092	1.430***	.063
F			35.101		14.450		9.597
R 2			.147		0.067		0.045

*p < .05; **p < .01; ***p < .001.

Arguments

Overall, there was a small and only marginally significant effect for framing condition, F(9, 3,654) = 1.92, p = .05, ηp 2 = .005. Furthermore, for the three separate practices, the evidence for the effects of arguments in favour or against tolerance is weak and mixed (see Table 1). A positive effect of a pro argument on tolerance was only found for the scenario of a homophobic statement (Table 2, column 4). Compared to the control condition (no argument), participants displayed higher levels of tolerance when they read a pro tolerance argument. Additionally, participants tolerated the exclusion of women somewhat less when they read a contra tolerance argument (Table 2, column 3). Further, the results indicate that mixed arguments (favouring and opposing combined) did not have an effect on the level of tolerance of the three practices.4

Protestant participants

In general, participants belonging to a Protestant denomination were more positive (M = −1.26, SD = 1.21) towards the different practices as compared to non‐religious participants and participants belonging to another denomination (M = −2.03, SD = .84), F(3, 1,221) = 51.70, p < .001, η2 = .11. Compared to non‐Protestant participants, Protestant participants also exhibited higher levels of acceptance towards the practices, F(3, 1,216) = 29.24, p < .001, ηp 2 = .07. This was especially the case when the practices were performed by orthodox Protestants as indicated by a significant negative interaction effect between Protestant affiliation and target group, F(3, 1,216) = 7.22, p = .001, ηp 2 = .02. This interaction effect was found for all three practices, but appeared especially strong for the religious schools and less so for the exclusion of women and the homophobic statement (Table 2, columns 2–4).

The finding that participants with Protestant affiliation exhibited higher acceptance when practices are performed by orthodox Protestants compared to Muslims may also hold for Christians in general who were more positive towards the practices (M = −1.54, SD = 1.13) than non‐religious participants (M = −2.08, SD = .81), F(3, 1,221) = 31.49, p < .001, η2 = .07. Therefore, Protestant affiliation and its interaction with target group were replaced by Christian affiliation and the respective interaction with target group in an additional multivariate regression analysis. As compared to non‐Christians, Christians were more accepting when practices were performed by orthodox Protestants than Muslims. Thus, biased acceptance judgements based on the target group were not particular to Protestants but held for Christians in general. This result might be due to the fact that Protestants, and Christians, more generally, are more positive towards the different practices when performed by orthodox Protestants, which would mean that their acceptance is less a matter of tolerance (objection without interference). Therefore, we conducted additional analyses that are reported below.

Additional analysis

The sample used for the main analysis included a small share of participants that had positive feelings towards the practices, which indicates that for them the issue is not a question of toleration. An additional multivariate regression analysis was performed including only participants who displayed negative or neutral feelings towards the three practices (N = 1,079).5 Results are similar to the main analyses, with the exception that the interactions of Protestant affiliation and target group were non‐significant for tolerance of the exclusion of women and the homophobic statement. These different results can be attributed to the fact that half of the participants with positive feelings towards the practices performed by orthodox Protestants were themselves Protestants. These participants with positive feelings were included in the main analysis but not in the additional analysis. Thus, the additional analyses indicate that Protestants who express positive feelings towards the practices performed by Protestants drive the interaction effects observed in the main analysis.

Discussion

In many Western societies, there have been heated debates about the accommodation of specific cultural and religious norms and practices within society. Some argue that opposition to minority practices (e.g., by Muslims in the west) indicates general dislike and prejudice (Spruyt & Elcharus, 2012), whereas others claim that it can be a question of disapproval based on general principles (Sniderman & Hagendoorn, 2007). We compared tolerance of three controversial practices performed by either Muslims (a culturally dissimilar group) or orthodox Protestants (a culturally similar group). This allowed us to assess whether Muslim practices are tolerated less and whether this depends on the nature of the particular practice. Our findings demonstrate that differential toleration of Muslim practices may depend on the extent to which the practice evokes moral concerns.

First, participants were most negative and least tolerant of a homophobic statement that challenges moral norms of preventing harm and equal respect. Participants were less negative and somewhat more tolerant towards the exclusion of women from a religious board. The tolerance for the founding of religious schools was highest. These findings support previous research that demonstrated that these three practices differ in how strongly they raise moral concerns (Gieling et al., 2010; Van Doorn, 2015; Verkuyten & Slooter, 2007) which we also found in our post‐hoc study on a national sample. Research in social domain theory (Turiel, 2002) and on moral convictions (Skitka & Morgan, 2014) indicates that people perceive their moral beliefs to be objectively and universally true making something wrong independent of who is doing it. In line with this, we found that participants were less tolerant of Muslims (vs. orthodox Protestants) founding separate religious schools (an issue low in moral concern), but that there was no difference in tolerance for the homophobic statement and exclusion of women (the issues of greater moral concern) regardless of whether these were done by a Muslim or orthodox Protestant. According to Hurwitz and Mondak (2002), threat perceptions and group affiliations tend to be irrelevant to tolerance of highly offensive acts as they are rejected under all circumstances. In Dutch society, calling gays and lesbians inferior people and excluding women go against important moral principles (Dagevos, Andriessen, & Vervoort, 2016) and are not tolerated, neither for Muslims nor orthodox Protestants.

Interestingly, tolerance did not systematically depend on the presentation of pro, contra, or mixed arguments for accepting the practices. As compared to when no arguments were made, pointing out a value favouring tolerance only increased tolerance of the homophobic statement. Similarly, only tolerance for the exclusion of women was found to be lower when an argument against the practice was presented. Presenting a mixture of pro and contra arguments did not have any clear effects, which has also been found in other research (Brewer & Gross, 2005; Sniderman & Theriault, 2004). There are at least two possible explanations for these findings.

First, due to floor effects, framing may not decrease tolerance in the case of highly offensive practices that evoke very strong objection such as making a homophobic statement or excluding women (Slothuus, 2008). Strong attitudes are relatively resistant to arguments and framing effects and difficult to change (Howe & Krosnick, 2017; Lecheler et al., 2009). This is especially the case when attitudes are based on moral beliefs that tend to be considered objective and universal (Skitka & Morgan, 2014).

Second, framing effects may only be observed in the case of strong arguments (Chong & Druckman, 2007) and the persuasiveness of the arguments used in the experiment might not have been strong enough. Only referring to core values related to the controversial practices may not suffice to alter considerations, especially since participants’ attention span may be relatively low when completing online questionnaires. Additionally, convincing the intolerant to become tolerant is quite difficult (Gibson, 2006) and therefore might require the use of vivid framing methods such as videos and newspaper articles with emotional appeals (Chong & Druckman, 2007).

Limitations and future directions

There are some limitations to the current study, which may be addressed in future research. For example, while we focused on concrete cases rather than abstract principles and used realistic and debated issues instead of unfamiliar and hypothetical scenarios, we were only able to examine tolerance of three practices. Future research could focus on other practices that clearly differ in terms of the moral concerns that they elicit, and measure moral convictions in addition to, for example, attitude strength (Skitka & Morgan, 2014). Furthermore, future work could assess individuals’ actual experiences with the particular practices and one's knowledge of them. It would also be useful to examine the individual endorsement of, for example, free speech and the importance attached to public order and social cohesion (Peffley, Knigge, & Hurwitz, 2001). For example, people can clearly prioritize one value over another, but experiences of value conflict are also possible because competing values are simultaneously considered equally important.

Additionally, while we focused on the importance of the religious group and specific practices and arguments for tolerance, future studies could examine other dimensions that are likely to be important. For example, the sense in which people are expected to be tolerant (social, political), the social implications of tolerance, and the underlying beliefs (e.g., informational, cultural, moral) of the people engaged in the disapproved practices might all be relevant to study. Practices that are more visible and that have more far‐reaching societal consequences might be less tolerated, and dissenting practices based on other informational beliefs compared to other moral beliefs might be tolerated more (Ellemers, 2017; Verkuyten & Slooter, 2007).

Furthermore, participants were asked what responsible authorities should do about specific practices because tolerance typically implies the power to prohibit or seriously interfere with conduct that one finds objectionable (Horton, 1996). However, apart from the controversial nature of the practices, participants might have other reasons (e.g., anti‐government, anti‐authority) for why they think that responsible authorities should (not) take steps against specific practices. Yet, this would not explain that the three practices were not tolerated similarly and that participants expressed less tolerance for some practices when associated with Muslims than orthodox Protestants.

Finally, it could be argued that the five response options to measure tolerance constitute separate categories rather than a continuous scale. However, the categories have been treated as a continuous scale in previous studies (e.g., Gieling et al., 2010; Verkuyten & Slooter, 2007), the scale is strongly related to the continuous ‘feeling’ measure (note 2), and research has demonstrated that the findings are not sensitive to treating the measure as a continuous or an ordinal scale (Sleijper, 2018). Future research may, however, benefit from using a different Likert‐type format or develop a more detailed understanding of the different possible meanings of the response categories used to measure tolerance.

Conclusion

In contrast to existing research on tolerance of minority practices (e.g., Van der Nol, 2014; Verkuyten & Slooter, 2007), here we have compared tolerance of different practices enacted by culturally similar versus dissimilar groups. This systematic comparison allowed us to assess whether opposition towards minority (Muslim) practices is not just a question of dislike and prejudice towards the group (discriminatory tolerance), but can also involve disapproval of specific practices independent of the actor (generic tolerance). The findings support the notion that (in)tolerance and prejudicial attitudes are distinct phenomena (e.g., Crawford, 2014; Klein & Zick, 2013; Van der Noll, 2010). Social psychological research tends to equate tolerance with openness, being well disposed towards cultural others, or having a generalized positive attitude towards them (see Verkuyten & Yogeeswaran, 2017). In this understanding, tolerance is the opposite of dislike, disapproval, or prejudicial attitude. By contrast, the classical understanding of tolerance involves putting up with something that one disapproves of or is negative about. Tolerance keeps negative attitudes and beliefs from becoming negative actions and thereby makes living with differences possible. A diverse and peaceful society does not require that we all like each other, but it does necessarily mean that people have learned to tolerate one another. We have to agree how to disagree, and social psychological research should make an important contribution to our understanding of how people think about different types of provocative practices and develop tolerant and intolerant judgements.