Value Conflicts Revisited: Muslims, Gender Equality, and Gestures of Respect

Abstract This is a study on the inclusion of Muslims in liberal democracies in the presence of value conflict. We focus on handshaking controversies that appear to pit gender equality against religious freedom. The possible outcomes seem mutually exclusive: either conservative Muslim minorities must conform to the norms of the majority culture, or non-Muslim majorities must acquiesce to the legitimacy of conservative Muslim ideas. Using a trio of experiments to replicate our results, we demonstrate the efficacy of introducing alternative gestures of respect. Presented with a substitute gesture of respect – placing the ‘hand on heart’ – non-Muslim demands for Muslim conformity drop dramatically. The results of the handshaking experiments call out a general lesson. Thanks to the ingenuity and versatility of cultural customs to signal respect, value conflicts can be open to resolution in everyday encounters without minorities or majorities having to forsake their convictions.


Introduction
How are Muslim minorities to be included in liberal democracies?Sharing the values of the larger society is part of the answer, just as the largest number of Muslims do (Choi, Poertner, and Sambanis 2022a;Choi, Poertner, and Sambanis 2022b;Velthuis et al. 2022a;Velthuis, van der Noll, and Verkuyten 2022b).Yet conflicts between liberal values and conservative injunctions of Islam, most conspicuously but not exclusively on issues of gender equality, continue to be political flashpoints.Non-Muslims tend to view Muslims with conservative beliefs about gender equality as prototypical while viewing norm-complying Muslims, even though a majority, 'as exceptions who are not typical of their group' (Choi, Poertner, and Sambanis 2022a, 194).When such bias on the part of non-Muslims is the issue, time and mutual adaptation can favour conciliation (Choi, Poertner, and Sambanis 2022a).But an immediate challenge deserves attention: how should liberal democratic societies deal with conflicts of liberal and traditional Islamic values in hard cases, here and now, when observant Muslims believe in conservative, religiously grounded ideas 1 and seek to exercise their rights to religious freedom? 1 To avoid stigmatizing concepts when describing people who believe in Islamic ideas that proscribe touching members of the opposite sex, we use the terms traditional, observant, or conservative.In our usage, these concepts differentiate Muslims who believe it is wrong to shake hands with members of the opposite sex from other Muslims who do not believe in these conservative Islamic ideas or who are not observant of them.We want to underline both that Muslims are commonly fine with handshaking and that, for some Muslims, not shaking hands is a matter of religious conviction.
Hard cases make bad law but good social science.Public refusals of religiously observant Muslims to shake hands with members of the opposite sex in public encounters are a recurrent flashpoint in several Western European countries (cf.Carol and Koopmans 2013;Carol, Helbling, and Michalowski 2015;Joppke 2013;Koopmans et al. 2005).The president of France refused citizenship to a Muslim woman who refused to shake hands at a citizenship ceremony (Orgad 2021).Muslim students in Switzerland were forced to shake hands with teachers of the opposite sex because refusing to do so violated common courtesy, asserted gender inequality, and rejected Swiss values (Orgad 2021).A Muslim man in Sweden failed a job interview for not shaking hands with the female interviewer (Bob 2012).So, our test case is an observant Muslim's dilemma when the cultural norm is to shake hands in public encounters.For a traditionalist, to shake hands with a member of the opposite sex violates a religious injunction; for a non-Muslim who believes in liberal values, his refusal to shake hands with her violates her right to equal treatment.
Ours is a simple idea.Both Muslim and non-Muslim cultures have extensive repertoires of symbols and practices, among them gestures of respect.Gestures of respect are translatable and, therefore, substitutable across cultures.Substitute gestures of respect, we show, are key to dramatically reducing conformity pressures on observant Muslims in everyday situations where the underlying value conflict remains deep.
Conveying respect is the key, but respect is not a simple idea.Accordingly, our first step is to outline a theory of respect.The core of the theory is a distinction between two types of respectappraisal and recognition (Darwall 1977).The relevance of this distinction is owed to participant observation, personally witnessing conservative Muslims demonstrating recognition respect by placing their hand on their heart instead of shaking hands. 2 To test the hypothesis of a reduction in pressure to conform in response to a substitute gesture of respect, we conducted three independent handshaking experiments in Germany, where the handshaking custom is the norm.In each experiment, responses are observed either in the presence or absence of a substitute gesture of respect.To establish scope conditions, we analyze responses according to three related but distinct sources of support for minorities: internal motivation to control prejudice, immigration attitudes, and political orientation.

Meta-Conflicts and a Theory of Respect
Conflicts of values are the department of normative theorists (see, for example, March 2009).Conflicts of practices and customs, that is, how people should and do behave toward each other, are the stuff of everyday life.Intuitively, gestures of respectlistening attentively, waiting your turn in lineare solvents of social friction in everyday encounters.But everyday observation of gestures of respect as social solvents notwithstanding, the concept of respect itself is not transparent.
Disrespect offers a foothold for a theory of respect.A traditionally observant Muslim not shaking hands with a member of the opposite sex is an act of omission, narrowly construed.The perceived insult lies in a presumption of choice: he could have shaken hands but chose not to.Disrespect is thus the inference to draw.We respond not to the act in isolation but 'to the complex of the act and its motives' (Eidelson 2015, 77, italics in original).
More precisely, what does the disrespect consist in?One meaning of respect is 'an attitude of positive appraisal' of [a] person's merit or achievements (Darwall 1977, 38).But a person who believes in gender equality cannot be required to commend a Muslim man for refusing to shake a woman's hand.Recognition respect is different from appraisal respect.It is the respect we owe another person in virtue of her being a personanother human being like ourselves 2 The meeting to promote better relations between Muslims and non-Muslims took place at Bergen City Hall on Saturday, 28 May 2016.
who has moral weight and responsibility, whose intrisic worth therefore requires acknowledgement.Recognition respect is thus 'an affirmative obligation that one take to heart the various normative upshots of her being a person and regulate one's action accordingly' (Eidelson 2015, 74-5).The distinction between appraisal respect and recognition respect proves to be key.
Abstractly formulated, the logic of conflict over gender and handshaking is strict.The traditionally observant Muslim is asked to do what his convictions condemn as wrong, shaking the hands of a member of the opposite sex.The non-Muslim majority is asked to accept what their convictions condemn as wrong, treating women differently and worse.The alternatives are mutually exclusive and mutually exhaustive.Either the traditionally observant Muslim must conform to the values of the larger society, or the larger liberal society must accept the legitimacy of values that it deems illiberal.
Concretely formulated, the logic of conflict over gender and handshaking is more permissive in so far as majorities and minorities mutually recognize that each is 'entitled to have other persons take seriously and weight appropriately the fact that they are persons in deliberating about what to do' (Darwall 1977, 38).The clash of deeper-lying values remains; the traditionally observant Muslim is honour bound not to shake hands with a member of the opposite sex, and the non-Muslim is correspondingly bound to honour gender equality.But now, rather than the minority having to yield to the majority or the majority having to yield to the minority, there is an alternativeaffirmation of mutual regard through a substitute gesture of respect.The gesture of the observant Muslim placing his hand on his heart to greet a woman constitutes acknowledgement of 'an affirmative obligation that one take[s] to heart the various normative upshots of her being a person and regulate one's action accordingly' (Eidelson 2015, 74-5).So far as this substitute gesture of respect is recognized as a gesture of recognition respect, the conflict over handshaking can be resolved without first having to wait for a resolution of the conflict of values underlying it.Our first hypothesis is: H1: A substitute gesture of respect, specifically placing a hand on the heart, can be an effective substitute for shaking hands.
By effective, we mean reducing majority pressure on conservative observant Muslims to conform to the customs of the larger society that are at odds with their religious convictions.The conditional, 'can be an effective substitute', flags the need to specify scope conditions.Who is likely to be responsive to a Muslim employing a substitute gesture of respect, and who is least likely, and why?In thinking about this question, it is important to take into account the heterogeneity of preferences in the non-Muslim population.
In liberal democracies, we have witnessed for decades a society-wide increase in commitment to culturally liberal values, including, but not limited to, tolerance, openness to minorities, and an appreciation of cultural pluralism (Inglehart 2018;Inglehart and Welzel 2007).This increase in commitment to liberal cultural values has sparked a conservative cultural value backlash (see, for example, Norris and Inglehart 2019).Issues of inclusion and diversity regularly divide cultural liberals and cultural conservatives.In contrast, our concern is with issues of inclusion and diversity that divide cultural liberals themselves, when they must choose between two liberal values, minority inclusion and gender equality (Chong, Citrin, and Levy 2022;McClosky and Brill 1983).Based on this reasoning, we formulate a hypothesis about scope conditions: H2: The substitute gesture of respect will be most effective among those who have more culturally liberal values and left-leaning political orientations.

The Handshaking Experiments
The handshaking experiments are designed to directly test H1.They have two main experimental factors.In the first randomization, Muslims refused to shake hands without a substitute gesture of respect or refused to shake hands but, instead, placed a hand on their heart.In the second randomization, the refusal to shake hands is directed at non-Muslim women, specifically, or gender is left unspecified.
Replication is a vital concern.Pre-registration provides evidence that the results observed are the results anticipated.But evidence that results are anticipated is not evidence that they are replicable.To demonstrate that an experiment's results are replicable necessitates doing it again.This study is not pre-registered.Instead, it introduces an alternative, a fusillade design allowing multiple independent tests of the hypotheses.Requiring empirical proof of replicability by repeating an experiment not once but twice sets a high standard for replication.
The hypothesis to test is that a substitute gesture of respect compensates for a refusal to shake hands with a member of the opposite sex.To provide direct evidence of the replicability of our results, we conducted the same experiment three times, varying potentially relevant contextual details to test for generalizability.An imam refusing to shake hands with a non-Muslim woman arguably will evoke a stronger negative reaction than a Muslim child doing so.
Specifically, in The Politician Experiment, respondents read: Some local politicians were invited by the local Muslim community centre to the annual spring festival.To welcome them, the local politicians shook hands with the hosts.Some imams did not shake hands with the [politicians/female politicians].[Instead, they put their hand on their heart in greeting/No mention].After the meeting, some complained that Muslims do not follow the customs of our country.They asked that next time, everyone present be asked to shake hands.
In The Teacher Experiment, respondents read: The local school organized a welcome day for newly arrived families with school children.Among them were also some Muslim families.To welcome them, the teachers shook hands with the families.Some of the Muslim children did not shake hands with the [teachers/female teachers].[Instead, they put their hand on their heart in greeting/No mention].
After the meeting, some complained that Muslims do not follow the customs of our country.They asked that next time, everyone present be asked to shake hands.
In The HR Manager Experiment, respondents read: A local company organized an assessment day to fill vacancies.Several of the invited candidates were Muslim.To welcome them, the recruiters shook hands with the candidates.Some of the Muslim candidates did not shake hands with the [HR managers/female HR managers].[Instead, they put their hand on their heart in greeting/No mention].After the assessment day, some complained that Muslims do not follow the customs of our country.They asked that next time, everyone present be asked to shake hands.
In all three experiments, respondents were then asked: To what extent are you for or against such a request?(1, 'totally against'; 2, 'rather against'; 3, 'rather in favour'; and 4, 'fully in favour'). 33 There are arguments for 5-point Likert scales in preference to 4 and 4-point in preference to 5. The main benefit of a 5-point scale is that it contains a middle category that allows respondents who truly have no view on an issue to record it.A disadvantage is that respondents may choose the middle category to avoid discomfortsuch as having to think The challenge is to devise a procedure for conducting three identical experiments except for variations in social settings without responses to one affecting responses to either of the others.To achieve this, we introduce a fusillade design.The three experiments are each assigned to a randomly selected third of the sample.This fusillade design thus enables testing the same hypothesis three times simultaneously, eliminating the possibility that participation in one experiment could affect the outcome in any other.Together, this yields the 2 × 2 × 3 between-subjects factorial design summarized in Fig. 1.

The Sample
The three handshaking experiments were administered to a sample of N = 2,654 respondents recruited from an online panel provided by the commercial survey provider Respondi.Such convenience samples have been shown to generate results comparable to traditional probability-based survey samples (Coppock and McClellan 2019;Mullinix, Leeper, Druckman, and Freese 2015). 4 The data collection was fielded between 29 September and 17 October 2021.Quotas were implemented to match the general German population in terms of gender, age, and education.Half of our respondents are female (50 per cent) with an average age of 47 years (SD: 15 years).Some 23 per cent obtained the highest school-level education (Abitur). 5 The study was introduced to the respondents in very general terms as researching 'the attitudes of the population in Germany on various political issues'.The experiment was placed in the middle of the survey questionnaire, preceded by other questions on Muslim integration (acceptability of new mosques, halal food, Muslim holidays, and political rights) as well as political variables such as populism or threat perception.Possible question order effects were sought to be minimized through the use of digression blocs and questions about matters unrelated to Muslims, inserted before the experiments regarding Muslims.
About 87 per cent of respondents passed two (post-treatment) attention checks. 6Tables A1 and A2 in the appendix provide descriptive statistics for those who passed and those who failed the checks.Overall, there are a few notable differences.Respondents who passed the attention checks tend to be slightly older, slightly better educated, a bit higher on internal motivation to control prejudice (IMCP), and less likely to have a strong Christian identity (see Table A3 in the Supplementary Materials).However, to avoid post-treatment bias, we do not exclude respondents who failed the attention checks in our main analysis (Montgomery, Nyhan, and Torres 2018).Re-running our results on the restricted sample (including only those who passed both attention checks) yields robust results (see Table A8 and S12-14 in the Supplementary Materials).
hard about an issue or responding in a way that goes against what they perceive to be widely shared societal norms.We have chosen a combination of a 4-point scale and the possibility of selecting 'next' for respondents who do not want to record an opinion.The share of people who choose next is low (N = 6 or 0.2 per cent).The 4-point scale can be one reason why so many respondents in our study insist on conformity in the baseline conditions (Fig 4).If so, the attempt at avoiding inflated measures of pro-minority views was successful.That noted, we have used 4, 5, and 7-point Likert scales in previous work, and we have not found an occasion where one led to different main effects than the others (cf.Ivarsflaten and Sniderman 2022).

4
To be sure that the results were not an artefact of the convenience sample, we replicated a part of the experiment in a probability sample in a different Western European country where handshaking is also the norm.The results replicate; see Figure A2 in the appendix.5 Table A0 in the appendix compares vote intentions in the sample used in this study with vote intentions in other German samples studied during the same month.The table shows that the sample used in this study contains a larger share of farright (AfD) voters than in other studies.Since voters with a far-right orientation are the least likely to respond to the alternative gesture of respect, the estimated effects reported in this study are likely to be conservative compared to what would have been found in the other samples.
Tables A4, A5, and A6 provide balance checks for the different experimental conditions of the three handshaking experiments.The treatment conditions are well balanced.The only exception is a slight gender imbalance in the 'local community centre' setting.But, correcting this imbalance by controlling for gender (see Tables A7-A14) does not change our results.All in all, any differences in outcomes cannot be attributed to the compositional differences of the experimental groups.

Additional Measures
To provide assurance that our results are not the product of an idiosyncratic choice of measures, three different indicators of inclusionary values and left-leaning political orientations are used to examine H2.The indicators are conceptually and operationally distinct, yet they have all been shown to be associated with openness to minorities (Blinder, Ford, and Ivarsflaten 2013;Harteveldt and Ivarsflaten 2016;Helbling 2012;Hetherington and Weiler 2009).Note that all measures were asked after the experiment (and are, therefore, strictly speaking, post-treatment), but thanks to a buffer of general political questions, priming by the experiment is unlikely (Montgomery, Nyhan, and Torres 2018; but see Albertson and Jessee 2023).
The first indicator is the IMCP.It is measured by four survey items (Cronbach's alpha 0.85): 'I try to act without prejudice towards Muslims because it is personally important to me '; 'I get angry with myself when I have a prejudiced thought'; 'I try to act without prejudice towards Muslims because due to my own convictions'; and 'I don't want to appear racist, not even to myself'.Answers were registered on a 5-point scale (1 -'Strongly disagree'; 2 -'Rather disagree'; 3 -'Partly agree'; 4 -'Rather agree'; 5 -'Strongly agree').This indicator has been used to predict a reduction in the tendency to discriminate against Muslims (Blinder, Ford, and Ivarsflaten 2019) and the likelihood of voting for the extreme right (Blinder, Ford, and Ivarsflaten 2013;Harteveldt and Ivarsflaten 2016).
The second indicator is general immigration attitudes, combining the following three items into a single scale (Cronbach's alpha 0.92): 'In general, would you say it is good or bad for the German economy that immigrants come here?' (0 -'bad for the German economy' to 10 -'good for the German economy'); 'And would you say that cultural life in Germany in general is undermined or enriched by immigrants?' (0 -'undermines cultural life in Germany' to 10 -'enriches cultural life in Germany'); and 'Will immigrants make Germany a worse or better place to live?' (0 -'worse place to live' to 10 -'better place to live').While the question of immigration is conceptually distinct from minority inclusion, several studies have shown that attitudes towards immigration and Muslim inclusion in contemporary Europe are predicted by similar dispositions ( Helbling 2012).
The third indicator is differences in political orientation measured conventionally as selfplacement on the political left-right ideological spectrum.It is captured by the 11-point standard item: 'In politics, people sometimes talk about "left" and "right".Where would you place yourself on this scale, where 0 is left, and 10 is right?'.It is well established that those who identify on the left in contemporary politics are more likely to favour minority inclusion than those on the right (Hetherington and Weiler 2009).Descriptive statistics of these scales are provided in Tables A1  and A2 in the Supplementary Materials.

Results
Figure 2 presents the main results of the three handshaking experiments.It gives the percentages of those who are 'rather' or 'fully in favour' of compulsory handshaking across the two experimental conditions (along with 95 per cent confidence intervals).As expected, we find in the baseline conditions broadly based pressure to conformto insist on compulsory handshakingespecially if this involves a refusal to shake hands with women specifically.Three-quarters of the respondents think Muslims should be asked to conform to the handshaking custom if they refuse to shake hands with female politicians (76 per cent), female teachers (70 per cent), and female HR managers (74 per cent).These numbers are significantly higher than in the condition that speaks of refusing to shake hands with politicians (44 per cent), teachers (52 per cent), and HR managers (59 per cent) when gender is not mentioned.
The core hypothesis (H1)that the introduction of the substitute gesture effectively reduces conformity pressures on Muslimsis supported by the descending lines across all panels in Fig. 2. The effects are not merely statistically significant, they are substantively large.Even when gender is emphasized, placing a hand on the heart reduces the insistence on shaking the hands of politicians by 24 percentage points, teachers by 20 percentage points, and female HR managers by 21 percentage points (confidence intervals as indicated in Fig. 2).If the gender of politicians, teachers, and HR managers is left unspecified and the substitute gesture is introduced, only around 40 per cent of the non-Muslim public still insists on conforming to the practice of handshaking.In sum, the substitute gesture applied to women effectively undermines the massive insistence on cultural conformity to the handshaking standard.Applied in a gender-neutral manner, the substitute gesture is even more effective. 7It turns the tables completely.This evidence is in line with H1 that the substitute gesture is effective in reducing pressures on minorities to conform. Figure A1 in the appendix repeats the analysis of the handshaking experiments separately for female and male respondents.We find no important differences in the insistence on cultural conformity and compulsory handshaking across the genders, notwithstanding the plausibility of a conjecture that women may be more affronted by a refusal to shake hands than men.Table A 15 in the Appendix shows the results of a follow-up experiment to test whether the finding holds if men were singled out by Muslim women for refusals to shake hands.The table shows that the substitute gesture also significantly lessens conformity pressures in this situation.
Figures 3-5 assess scope conditions (H2), employing three distinctive indicators of culturally liberal attitudes -IMCP, immigration attitudes, and political ideology.We report predicted probabilitiesderived from logistic regression models8 of favouring compulsory handshakingwith full three-way interactions between the indicators, handshaking and female politicians/teachers/ HR managers (see Tables A9-A11 in the Appendix).
Two broad patterns emerge.First, as expected, the demand for cultural conformity in general strongly and significantly declines with higher levels of IMCP (Fig. 3), higher scores on immigration attitudes (Fig. 4), and lower scores on the left-right self-placement scale (Fig. 5).Second, in line with the scope condition hypothesis (H2), the effect of placing a hand on the heart is generally stronger for respondents who score high on IMCP and immigration attitudes and low on left-right self-placement.Across Figs 3-5, we do not observe one instance of respondents with an exclusionary or ideological far-right orientation (those scoring below 3 on IMCP, below 4 on immigration attitudes, or above 7 on left-right ideology) responding affirmatively to the substitute gesture of respect.Those who score high on indicators associated with an exclusionary mindset insist on conformity in the presence of an alternative gesture of respect.
All the main variations across the three figures are on the side of respondents open to inclusion.That said, there are some notable differences across the figures and panels.IMCP is the indicator that produces the most consistent moderation pattern.In Fig. 3, five of the six panels show that those scoring high on IMCP (4) respond affirmatively to the 'hand on heart' treatment.It is only in the local school scenario where gender is not mentioned that the substitute gesture has no effect on those who score high on IMCP.But, instructively, the reason is not a failure of the substitute gesture to dampen pressures to conform but a lack of conformity pressure altogether in the control condition in this variant.
Support for H2 is only somewhat less consistent in Fig. 4, where the effect of the substitute gesture is moderated by immigration attitude.In four of the six panels, the results are in line with the scope-condition hypothesis: the effect of the substitute gesture is present at scores in the middle or at the higher end of the scale.As with the IMCP indicator, the strongest effect of the substitute gesture of respect is seen when women are explicitly targeted for handshake refusals.We also explicitly tested for an interaction between the manipulations of 'hand on heart' and 'female politicians/teachers/ HR managers'.This was done using a simple OLS regression with a multiplicative interaction term for two experimental conditions.While the effect of 'hand on heart' significantly interacts with female politicians in the local community centre scenario, this is not replicated in the local school and local company scenarios where the effect is additive.See Table A7 in the Supplementary Materials.Relying on only one experiment would have misled us, which is evidence of the value of fusillade designs.4, it is nonetheless safe to say that the main findings also hold for political ideology.Those on the left are less insistent on cultural conformity and more likely to accept suitable substitutes for handshaking.Furthermore, the substitute gesture is especially important to them when their commitment to equal treatment of women is explicitly put under pressure.

Discussion
When Muslims hold liberal values on a par with what is common in society at large, the challenge that needs to be addressed is that of misperceptions or native bias.Previous work has pointed out that citizens in Europe think Muslims are more likely to hold conservative values than they do and respond favourably when it is made clear to them that Muslims share their values and norms (Choi, Poertner, and Sambanis 2022a;Choi, Poertner, and Sambanis 2022b; see also Hainmueller and Hopkins 2014).
For a liberal society aiming to be true to liberal ideals, reality, not misperception, is no less demanding a challenge.Conflicts between liberal values and the values of Islam are genuine for traditionally observant Muslims.And the fact that they are a minority within a minority is just the point.Liberal societies owe a duty of protection to minorities, which very much includes religious minorities.Diversity and cultural pluralism are flagship values of contemporary liberal societies, especially so among citizens on the centre-left.But so, too, is a larger understanding of equality, which very much includes gender equality, especially among citizens on the centre-left.Confronted with an imam refusing to shake hands, the responses we observed were overwhelmingly in favour of insisting on conformity among citizens across the political spectrum.
The main contribution of this study is to point to a way out of these value stand-offs, acknowledging the agency of both the majority and the minority.The challenge is to pinpoint what non-Muslims owe Muslims and what Muslims owe non-Muslims when each believes the other is wrong.We have proposed that the key is a theoretical distinction between two types of respect.Liberal non-Muslims do not have an obligation to commend the beliefs of conservative Muslims about the rights and responsibilities of womenappraisal respect.However, they do have an obligation to take to heart the weight of their religious edictsrecognition respect (Darwall 1977, 38;Eidelson 2015, 74-5).
Our results demonstrated that respect, understood as recognition respect, can be a solvent of value conflicts.Thanks to the 'fusillade' design, three experiments were conducted simultaneously, each on a random third of the sample, assuring the independence of all tests.These three tests provide reassuringly alike results, notwithstanding potentially significant differences in the social setting.The fusillade design thus provides confidence in the results, notwithstanding this being a one-off survey.
Still, it is fair to ask, would we observe in real life what we have observed in these three survey experiments?Two different questions need to be distinguished.The first is the sincerity of 'benign' responses to minorities in a public opinion interview.Are majorities only saying what they believe they should say about minorities in order to present themselves in a socially desirable light?Insincere tolerance is the concern: saying that Muslims need not shake hands, even though, in truth, they believe that they should be held to the same standard as everyone else in order to avoid an appearance of bias.The results of the handshaking experiments are unequivocal.All three demonstrate that, absent a substitute gesture of respect, the explicit consensus is that Muslims need to conform to the handshaking custom.
The insistence we observe on the handshaking custom in the absence of an alternative gesture of respect also alleviates additional concerns about the experiment being fielded during the COVID-19 pandemic when insistence on handshaking could have reasonably been expected to be lower.The very high levels of insistence on handshaking across experiments in the absence of the alternative gesture of respect indicate both that respondents felt free to express their sincere

Figure 1 .
Figure 1.Treatment conditions in the handshaking experiment.

Figure 2 .
Figure 2. Results of Handshaking Experiments.Estimates in the percentage who are 'rather in favour' or 'fully in favour' of compulsory handshaking with 95 per cent confidence intervals.Tests for the difference between 'no mention' and 'hand on heart' are based on two-sided t-tests.+p < 0.1, *p < 0.05, **p < 0.01.See Table A7 in the Appendix for full regression results. 7

Figure 3 .
Figure 3. Results of Handshaking Experiments by Respondents' Internal Motivation to Control Prejudice (IMCP).Estimates are percentages of 'rather in favour' or 'fully in favour' of compulsory handshaking with 95 per cent confidence intervalsbased on logistic regression models with full three-way interactions.SeeTable A9 in the appendix for full regression results.

Figure 4 .
Figure 4. Results of Handshaking Experiments by Respondents' Immigration Attitude.Estimates are percentages of 'rather in favour' or 'fully in favour' of compulsory handshaking with 95 per cent confidence intervalsbased on logistic regression models with full three-way interactions.See Table A10 in the Appendix for full regression results.

Figure 5 .
Figure 5. Results of Handshaking Experiments by Respondents' Political Left-Right-Ideology. Estimates are percentages of people who are 'rather in favour' or 'fully in favour' of compulsory handshaking with 95 per cent confidence intervals.Based on logistic regression models with full three-way interactions.See TableA11in the appendix for full regression results.

Figure 5
Figure5further inspects scope conditions using the general measure of left-right orientation.While there are certainly nuances, and patterns are somewhat less pronounced in this figure compared to Figs3 and 4, it is nonetheless safe to say that the main findings also hold for political ideology.Those on the left are less insistent on cultural conformity and more likely to accept suitable substitutes for handshaking.Furthermore, the substitute gesture is especially important to them when their commitment to equal treatment of women is explicitly put under pressure.