Positive Association Between Ritual Performance and Perceived Objectivity of Moral Norms

ABSTRACT

We examined the relationship between religious rituals and how people perceive moral norms. Prominent anthropological theories propose that rituals charge associated moral norms with objectivity such that moral norms are perceived as absolute and independent of time and space. We used two cross-sectional datasets to test this hypothesis and conducted five correlational studies with three culturally distinct populations. The results, supported by meta-analysis of our effect sizes, show a positive association between attending collective religious rituals and perceiving moral norms as objective. Moreover, increased saliency of the characteristic aspects of ritual form, namely the perceived invariance, and digitalizing and materializing potentials, was associated with increased reporting of moral norms as objective. Overall, this manuscript provides initial support for theories suggesting that ritual behavior helps ground moral norms by affecting perceptual mechanisms related to norm processing.

Introduction

Moral norms exhibit exceptional cross-cultural variation. For instance, while the 16^th-century Aztec community would consider a king`s failure to sacrifice one of his children as immoral, Christian communities across Medieval Europe used accusations of ritual killings to stir moral outrage against Jewish minorities. The high level of norm variance also applies within a single cultural context wherever we consider norm development in time (e.g., the disappearance of physical punishment of children at school in Western societies). Although norms change in time, individuals often perceive their group’s moral norms as true and eternal, while other groups’ norms as false and temporary. This distinction carries essential questions regarding the process through which norms are fabricated to appear objective and everlasting.

Recently, psychologists started to investigate folk intuitions and beliefs regarding the nature of morality (for a review, see, Pölzler, 2018), finding that people tend to perceive the realm of morality almost as objective as the realm of factual propositions (Goodwin & Darley, 2008). Importantly, researchers investigating potential factors that influence perceived moral objectivity suggested that aspects of religious conviction such as declared religiosity (Yilmaz & Bahçekapili, 2015), religious grounding of morality (Goodwin & Darley, 2008), and belief in supernatural punishment (Sarkissian & Phelan, 2019) increase moral objectivity. Some measurements of moral objectivity even include a sub-scale called “Divine Truth” that is explicitly linked to religiosity (together with “Independent Truth” and “Universal Truth,” see, Zijlstra, 2019).

However, while bringing an informative focus on religion, we propose that the past research neglected a crucial predictor – ritual behavior – and argue that ritual may be the necessary facilitator of the effects found for the various religion-related variables. Although religious beliefs and behaviors present an intricate complex where one influences the other, beliefs alone cannot sustain and perpetuate religious traditions. It is through ritual performance that beliefs are communicated, negotiated, and sustained in a population (Lang & Kundt, 2020; Shariff et al., 2014; Sosis, 2004, 2019). Thus, in this paper, we (1) develop a theory of how rituals may enhance perceived objectivity of moral norms and (2) test this theory across six studies.

Moral reality inside minds

For the purpose of this research, we use the concept of moral objectivity to represent an umbrella term for a variety of features that the human mind attributes to moral norms. In opposition to moral subjectivity and relativity, moral objectivity is a belief that moral norms exist independently of time and space (universality), circumstances (absoluteness), and humankind (realism). For example, Goodwin and Darley (2008) showed that people regard factual statements (e.g., “Boston is further north than Los Angeles”) almost as objective as ethical statements (e.g., “robbing a bank in order to pay for an expensive holiday is a morally bad action“) and more objective than statements about social conventions (e.g., “wearing pajamas and bathrobe to a seminar meeting is wrong behavior”) and preferential statements (e.g., “Frank Sinatra was a better singer than is Michael Bolton”). Similarly, Nichols and Folds-Bennett (2003) found that children see ethical beliefs (beliefs about the harm caused to living individuals) as more objective than taste and aesthetic preferences.

Crucially, Goodwin and Darley (2008) examined various types of grounding of ethical beliefs, and their results suggest that religious grounding of ethics is robustly correlated with moral objectivism. Using the items from the Ethics Position Questionnaire (EPQ) – the moral perception measure developed by Forsyth (1980) – Yilmaz and Bahçekapili (2015) found that moral objectivity was predicted by self-reported religiosity and that priming participants with a concept of god was associated with higher reported objectivity scores. Subsequent research (Sarkissian & Phelan, 2019) established a positive correlation between belief in hell and moral objectivity. This study also revealed that for participants affiliated with the Abrahamic religions, priming with divinity increased moral objectivity and, additionally, that priming with moral objectivity increased the perception of god as more punishing. More generally, morality-concerned, omnipotent, and punishing gods (Beheim et al., 2021), as well as broad supernatural punishments (Watts et al., 2015), are argued to play a crucial role in the evolution of large and complex societies by motivating individuals to cooperate with other believers (Lang et al., 2019; Norenzayan et al., 2016; Purzycki et al., 2016).

While we agree that religious beliefs play an essential role in the evolution of morality and, specifically, in moral objectivity, we consider the picture where moral objectivity is supported by the belief system alone incomplete. Religious traditions vary in how interventionist and punishing their gods are, yet even the traditions with less moralizing deities usually have a set of morals forged with a more fundamental element: formal public displays of commitment to religious norms. That is, rituals.

Rituals

Ritual studies traditionally stress the symbolic dimension of rituals (see, Bell, 2009 for a review). However, while rich in cross-cultural variation in employed symbols, rituals also exhibit cross-culturally recurring patterns in which ritual acts are performed in rigid and repetitive sequences that are highly exaggerated (Boyer & Liénard, 2006; Rappaport, 1979; Tonna et al., 2019). As argued by Rappaport (1999), rigid and repetitive performance affords reliable transmission of certain information and strengthens the transmission’s quality and reception.

Importantly, rituals provide a platform where participants can reliably communicate information about their acceptance of group norms symbolically linked to the rituals. The reliability of such communication is achieved by materially anchoring the communicated message through visible and tangible ritual performance (e.g., the number of piercings in the performer’s body; see, Xygalatas et al., 2021). This is crucial because rituals are more often than not concerned with states that lack physical properties (e.g., commitment or prestige; Chvaja & Řezníček, 2019), especially pronounced in the domain of moral norms (e.g., loyalty, fairness, righteousness), and raises questions about how norms can be perceived as objective when lacking material properties. We argue that the distinctive aspects of ritual behavior facilitate the materiality-objectivity link for moral norms.

Ritual aspects promoting norm objectivity

Inspired by Rappaport (1999), we identify three main aspects of the ritual form that facilitate norm objectivity. Namely, rituals are physically performed, use clearly delineated elements, postures, and utterances charged with meaning, and are invariant in the general structure of performance. These aspects have important consequences for the establishment of moral orders because these aspects allow to 1) anchor moral orders in the corporeal performance, 2) fashion clearly distinct states regarding the existence and applicability of moral norms, and 3) create an impression of eternally existing orders by invariably repeating the ritual.

First, ritual has materializing potential because ritual performance lends the solidity of objects to something previously immaterial and makes it directly accessible to the senses. By enacting a ritual that symbolically binds its performers to a specific set of norms, these norms are materialized in the performance itself. For example, when European medieval lords accepted a new vassal, they performed a ritual where the “lord grasped the clasped hands of the vassal” (Muir, 2005, p. 36). Then the vassal and the lord exchanged a kiss. Such formal performances materialize the specific norm that relates to the newly achieved relationship between the lord and the vassal. Of course, such materialization is further supported by the association of norms with the supernatural order and other interlocked aspects of a religious worldview. Still, the sheer power of the physicality of the ritual form, being repetitively associated with the norm, would enhance the perception of norms as objective.

Second, ritual digitalizes continuous, analog internal states into publicly observable states with clearly defined gradation. In other words, ritual expresses the complex and ambiguous states and concepts that are hard to quantify (e.g., prestige) in discrete quantitative units and, thus, facilitates comparison between them (e.g., a sitting order at a ceremony). The digital representation reduces vagueness and increases the clarity of the communicated message. The greater this reduction, the greater the clarity and unambiguity of distinction. In many cultural rituals, the ritual form brings this reduction to the binary extreme where there is only a “yes/no” message left. For example, the norms related to dating may be blurry on what is and what is not allowed for the dating couple regarding interactions with other potential partners. Norms instigated through the ritual performance of the marriage ceremony are not vague or unclear anymore but rather sharply delineated (e.g., no extra-pair copulation).

Third, the fact that rituals are often rigid and their structure is invariant (i.e., they must be performed in an exact prescribed manner) charges the target of ritual performance (i.e., norms) with seeming endurance and changelessness. For example, while the content of readings during the general Roman Catholic Mass changes weekly to such a degree that they might never repeat in a year, the structure of when to stand, sit, chant, kneel or shake hands during the mass does not vary. In fact, this structure is so invariable that an Order of the Mass (Ordo Missæ) has existed in nearly the same form for centuries.

In summary, we hypothesize that the more frequently one attends religious rituals, i.e., rituals that are symbolically linked to morality, the more one would perceive morality as objective. Another hypothesis is that an increase in perceived objectivity of norms would be associated with the perception of ritual as physically anchoring those norms (materialization), clearly delineating them (digitalization), and making them eternal (invariability). To enhance our tests’ robustness and cross-cultural credibility, we conducted six correlational studies on various populations using primary and secondary datasets. While we assume, for simplicity, that religious rituals are implicitly linked to morality through religious symbols, utterances, and preaching included in rituals, we also test whether ritual is associated with objective morality more strongly for participants who explicitly link morality to religion or believe that their god is moralizing and punishing.

First, we analyzed the association between rituals and norm objectivity in two preliminary studies using the measurements from the previous literature on the cross-sectional and cross-national sample datasets (Studies 1A and 1B). Then, we developed a novel scale for measuring two crucial facets of perceived norm objectivity: perceived norm universality (independent of time and space) and reality (independent of humankind) and tested this relationship in a Czech sample (Study 2). Third, we replicated the analysis of Study 2 on the sample from Study 1B (Study 3). Fourth, we replicated those analyses on a US Christian sample and explored whether the intensity of perceiving individual ritual aspects (materialization, digitalization, invariance) would also predict perceived norm objectivity (Study 4 and pre-registered Study 5). Finally, we conducted a meta-analysis of studies 1B-5. Note that while our theory speaks to the causal direction, that is, rituals and their aspects causing people to perceive moral norms as more objective, our tests are only correlational. We will turn back to this issue in the final discussion.

Study 1A

For our first study, we utilized existing data collected by Pew research center within the project Religious Landscapes in 2014 (Pew Research Center, 2015), representative of the US population and World/European Values Survey (we will refer to the collapsed data as WVS) representative of various countries between 1981 and 2004 (EVS, 2021; Haerpfer et al., 2021). We used these datasets to test the general relationship between ritual attendance and norm perception across different world regions.

Method

Materials from all studies in this manuscript are accessible from the Supplementary Materials (SM) document. Collected data and analytical codes can be found at https://osf.io/y3r4g/?view_only=b8f39767c14c4969a57a06fea32957d9.

Measures

In the Pew data set, participants were asked which of the two statements they agree with: “There are clear and absolute standards for what is right and wrong or whether something is right or wrong often depends on the situation.” In the WVS data set, participants chose from two statements: “There are absolutely clear guidelines about what is good and evil. These always apply to everyone, whatever the circumstances; or there can never be clear and absolute guidelines about what is good and evil. What is good and evil depends entirely upon the circumstances at the time.” These items were used as our measure of moral absoluteness – one instance of the moral objectivity concept. We coded choosing the first option as 1 and choosing the second option (together with the answer neither/both equally) as 0. Those who refused to answer or did not know were not included. Collective ritual attendance was measured on a 6-point scale in Pew and 7-point scale in WVS, where 1 = never and 6/7 = few times a week. As theoretically relevant confounds, we adjusted the models for various religious beliefs such as subjective religiosity (WVS), belief in hell (WVS, Pew), heaven (WVS, Pew), and for religious guiding of morality (Pew). Finally, we included into the model demographic controls such as age, sex, income, education, religious denomination (WVS, Pew), and state (Pew) in which participants live (country in WVS). In WVS, we were able to control for the size of the city where respondents live and for cultural interdependency through language families. For more details on the items and data sets, see section S1 in SM.

Analysis

Data were analyzed in R, version 4.0.3 (R Core Team, 2020). To analyze the general relationship between rituals and moral objectivity, we used four logistic regression models for Pew data and six for WVS. The first model included only collective ritual attendance as a predictor. The second model added age and sex as covariates. The third model added financial satisfaction (WVS) as a covariate, income (Pew), education (Pew, WVS), city size (WVS) as fixed effects, and state (Pew) and country (WVS) as random effects. The fourth model also controlled for religious variables: religious tradition or denomination, belief in hell and heaven as proximal concepts for supernatural punishment and reward (WVS, Pew), religious guidance of morality (Pew), and subjective religiosity (WVS). In WVS analyses, we included a fifth model with the year of the data collection as a continuous covariate and a sixth with language families as fixed effects. Random effects were modeled using package lme4 (Bates et al., 2015). Finally, after conducting all studies, we included an additional analysis to this study pertaining to the interaction of religious guiding of morality and ritual attendance. We built logistic models in four steps to analyze Pew data and in six steps for the analysis of WVS data. Unlike the analysis of the general relationship, we collapsed age and sex with other demographic confounds. Due to missing data, the models’ N varied from 34,502 to 26,865 for Pew analyses and from 217,925 to 33,849 for WVS analyses. We report odds ratios (OR) with 95% CI.

Results

Our prediction was that ritual attendance will positively predict the probability of being a moral absolutist. The first model revealed that the probability of endorsing absolute standards for right and wrong increased as the ritual attendance increased (Pew: OR = 1.37, 95% CI = [1.35, 1.39], p < .001; WVS: OR = 1.19, 95% CI = [1.19, 1.20], p < .001). The effects remained positive and confident across all modeling steps, yet the effect sizes decreased after adding all relevant covariates (Pew model 4: OR = 1.13, 95% CI = [1.11, 1.16], p < .001; WVS model 6: OR = 1.11, 95% CI = [1.09, 1.12], p < .001). Model 4 fitted to the Pew data suggested that participants reporting the lowest ritual participation had 27% chance of being absolutists. For participants who reported the highest ritual participation, the chance of being absolutists was 41%. In the WVS dataset, model 6 estimated that participants who scored the lowest on the ritual scale had, on average, 37% probability of being absolutists while this probability was 52% for those scoring the highest on the ritual scale. To calculate these estimates, we used the package effects (Fox & Weisberg, 2018). For details on models, see Tables S3 and S4 in SM.

To see whether some extreme countries do not drive the worldwide results, we conducted logistic regression with ritual attendance and belief in moralizing gods for each country separately, finding that out of 74 countries, the ritual variable was in a positive and statistically significant association with moral absolutism in 59 countries (see Table S5 in SM for details, and for visuals see the figure available at the osf.io page of this study).

Next, since the variable of religious guidance of morality included in Pew data taps clearly into the symbolical connection between ritual attendance and moral norms (and is associated with the probability of being absolutist: OR = 2.32, 95% CI = [2.18, 2.47], p < .001), we conducted an ad-hoc interaction analysis. Our prediction was that for people who guide their moral beliefs by religious teachings, the association between collective rituals and moral absoluteness will be stronger. The models revealed strong and stable interaction between anchoring morality in religious belief and ritual attendance (OR_interaction = 1.11, 95% CI = [1.07, 1.15], p < .001; see, Figure 1). This effect was even stronger when we controlled for various confounds (OR_interaction = 1.15, 95% CI = [1.10, 1.20], p < .001). See Table S6 in SM for further details.

Figure 1. Study1A: The relationship between collective religious rituals and absolute morality (with 95% confidence intervals). We observed a stronger association between the frequency of ritual performance and moral absolutism for participants who symbolically connect ritual attendance and morality through religious guidance in moral topics (compared to those who do not). The data were collected by Pew research center.

Discussion

In large and representative samples of US respondents and the world population, participation in religious collective rituals was positively associated with the belief that standards for what is right (good) or wrong (evil) are clear and absolute rather than depending on specific circumstances. These effects held even when we controlled for several important confounds related to religious beliefs. Ritual attendance, therefore, explains a unique variation in moral absolutism/universalism in people who share the same religious traditions and levels of belief in several religious concepts. Moreover, those who explicitly anchor their morality in religious teachings experience a stronger relationship between collective rituals and moral absoluteness than those who anchor morality in other ways. Therefore, our assumption about the necessity of a symbolical connection between rituals and moral norms was supported. However, these results need to be interpreted with caution. First, the measure provided by Pew and WVS taps only one potential factor of moral objectivity, namely absolutism/universalism. Second, multiple causal processes led to missing data in both datasets, and some of them could affect our results. For instance, in the WVS data, we were not able to determine why is the specific observation missing (e.g., it was not asked or participant refused to answer or did not know). Therefore, we conducted another preliminary Study 1B that used the norm objectivity scale already utilized in previous research, recruited a cross-national sample, and controlled the process of how missing data were generated.

Study 1B

To keep our inferences cross-nationally applicable, we utilized our ongoing data collection at the pilgrimage site to Santiago de Compostela in Spain to collect additional data on norm objectivity. The pilgrimage is attended by pilgrims from all around the world (37 nationalities in our sample) and, crucially, for both religious and secular reasons (Amaro et al., 2018).

Method

Participants

Pilgrims were recruited in Santiago through random sampling from the streets. The survey was finished by 322 participants, but due to missing data (see SM section S2 for details), the samples for individual analytical steps varied from 302 (172 females, M_age = 37.55, SD_age = 14.77) to 275. Participants were not compensated.

Measures

We used a 3-item construct (“There exists a single moral code that is applicable to everyone, regardless of any individual person’s beliefs or cultural background.”; “It is possible to compare different cultures by a single, universal standard of moral rightness.”; “If two people really disagree about a particular moral problem, then at most one of them can be correct, since moral problems cannot have multiple correct answers.”; Cronbach’s α = 0.65) taken from Sarkissian and Phelan (2019) to measure objective morality (M = 3.76, SD = 1.49) on a scale from 1 (totally disagree) to 7 (totally agree). Collective ritual participation was assessed by self-reported attendance (M = 2.29, SD = 1.68) on scale from 1 (never) to 7 (every day). Belief in moralizing gods (split-half reliability = 0.67, M = 2.78, SD = 1.66) was assessed by two items (e.g., “I believe that God will punish all sinners.,” “I believe that God knows everything we do or think.”) on a scale from 1 (totally disagree) to 7 (totally agree). This latent construct as well as other latent constructs in this manuscript, were created by averaging the respective variables. We also measured income, education, and nationality as control variables (see section S2 in SM).

Analysis

As a general modeling strategy in this study and the rest of the manuscript (except when explicitly describing a different strategy), we used a three-step modeling approach to assess the association between norm objectivity and ritual participation. In the first step, we assessed the relationship between norm objectivity and the frequency of ritual participation. In the second step, we adjusted these effects for demographic covariates (note that nationality was modeled as a random effect). In the third step, we added belief in a moralizing god to the model. We report standardized beta estimates (i.e., all continuous and ordinal predictors were standardized by z-scoring them before the analyses) from the first and third steps in the main text. This standardization method allows us to compare the effects of two essential predictors for our analyses, namely the frequency of ritual participation and the intensity of belief in moralizing gods. The results of all models can be found in SM. For testing constructs that we created, we used Principal Component Analysis (PCA; package psych [Revelle, 2020]) for the study in which the scale was developed and Confirmatory Factor Analysis (CFA; package lavaan [Rosseel, 2012]; cutoff criteria described by Hu & Bentler, 1999) in the following studies.

Results

We predicted that the frequency of participation in collective rituals should be positively associated with moral objectivity and detected this positive association using a linear regression (β = 0.15, 95% CI = [0.03, 0.26], p = .010). However, adjusting this model for belief in moralizing gods (β = 0.24, 95% CI = [0.10, 0.38], p = .001) dramatically decreased the magnitude and confidence of the ritual effect (β = 0.04, 95% CI = [–0.10, 0.18], p = .564). See the first row in Figure 2 and Table S8 in SM. To exclude the possibility that multicollinearity drove the decrease in the effect sizes from model 1 to model 3, we checked the Variance Inflation Factors (VIF) for moralizing gods (1.49) and collective rituals (1.53), which was below levels indicating multicollinearity. However, the mutual correlation between moralizing gods and rituals was rather high (Pearson’s r = 0.53).

Figure 2. The association between the frequency of collective ritual participation and norm objectivity (full scale) from studies 1B (first row), 2 (second row), 3 (third row), 4 (fourth row), and 5 (fifth row). The first column shows regression lines estimated based on the model with collective rituals as the only predictor (step 1 model). The second and third columns show the association of collective rituals and moralizing gods with norm objectivity while controlling for each other (and for other co-predictors; step 3 model), and the last column represents the simple relationship between collective rituals and moralizing gods. This figure suggests that the stronger the association between rituals and moralizing gods, the lower the association between rituals and moral objectivity after controlling for moralizing gods.

Discussion

We found that the frequency of participation in collective religious rituals was positively correlated with the perception of moral norms as objective. However, adding belief in moralizing gods to the model removed the predictive power of collective rituals. These results encouraged us to examine the relationship between rituals and norm objectivity in more detail in a series of four following studies.

First, while we chose the 3-item scale assessing norm objectivity in Study 1B due to its simplicity of use in the field setting, this scale inadequately captured the variety of norm objectivity aspects. Specifically, our aim was to create a measurement that would reflect the variety of perceptions associated with the concept of moral objectivity, namely universality, absoluteness, or reality of moral norms in a Platonic meaning (studies 2–5).

Second, to provide further robustness to our investigation, we inspected the relationship between rituals and moral objectivity in three distinct populations with varying degrees of ritual attendance and belief in moralizing gods (Czech university students: Study 2, Santiago pilgrims: Study 3, Christian US population: Study 4 and 5). See, Table 1 for a comparison of ritual frequency and belief in moralizing gods across studies (and populations).

Table 1. Means and standard deviations of religiosity-related variables in studies 1B-5. The scale applies to both measures (collective rituals and moralizing gods).

	Collective rituals	Moralizing gods	Scale
Study 1B (International pilgrims)	2.29 (1.68)	2.78 (1.66)	1–7
Study 2 (Czech students)	1.91 (1.42)	2.47 (1.43)	1–6
Study 3 (International pilgrims)	2.99 (1.66)	3.19 (1.72)	1–6
Study 4 (US Christians)	3.14 (1.73)	4.68 (1.24)	1–6
Study 5 (US Christians)	3.30 (1.73)	4.72 (1.26)	1–6

Third, while we found an association between the frequency of ritual performance and norm objectivity that supports our theory, it does so only in broad strokes and does not investigate the ritual aspects that, according to our theory, should be associated with norm objectivity. Therefore, we inspected whether the specific aspects of ritual behavior (materialization, digitalization, invariability) are associated with moral objectivity (Study 4 and 5).

Since we developed the new instruments of moral objectivity in the Czech language, we decided to first test the validity of the novel instruments in the Czech student population, then in the international pilgrim population, and finally, in the US population.

Study 2

To tap into various facets of moral objectivity, we developed a survey instrument measuring norm universality across cultures and ages, norm independency of human race, truth-aptness of the moral judgments, and absoluteness of moral standards.

Method

Participants

Using the Facebook groups of various Czech universities, we recruited 429 students to fill out an anonymous online survey (approx. 10 min., no reward). After excluding the participants who stopped filling out the survey before the norm objectivity measures (N = 144, see SM section S3 for details), the total sample consisted of 285 participants (197 females), but due to missing data, analyses with covariates were performed on 262 participants (M_age = 23.98, SD_age = 5.00).

Measures

Norm objectivity was assessed by 10 items, each on a scale from 1 (totally disagree) to 6 (totally agree). PCA suggested a three-item solution, but the time universality showed poor internal consistency (α = 0.59). Thus, we forced items to load on two factors which yielded a clear separation of items between universality across space and time and reality (for wording and factor loadings, see, Table 2, for details on the construction of the scale, see section S3 in SM). Thus, we created subscales of universality (α = 0.80, seven items, M = 2.52, SD = 0.81) and reality (α = 0.70, three items, M = 2.60, SD = 0.99) of moral norms. Moreover, we also collapsed all 10 variables and created one latent construct of norm objectivity (α = 0.80, M = 2.54, SD = 0.73). Both scales correlated strongly with the collapsed scale (r_universality = 0.92, r_reality = 0.67) and moderately with each other (r = 0.34), suggesting that scales are distinct but still tap into one concept. The frequency of collective rituals (M = 1.91, SD = 1.42) was measured on a scale from 1 (less than once a year or never) to 6 (at least once a day). To measure belief in moralizing gods, we used two items (“God or some other divine entity can punish immoral behavior.,” “God or some other divine entity can monitor our actions.”) measured on a scale from 1 (totally disagree) to 6 (totally agree). This measure showed a good reliability (split-half = 0.87; M = 2.47, SD = 1.43). As we sampled from a demographically homogenous student group, we included only controls of sex and age.

Table 2. Measure of moral objectivity with factor loadings. Bartlett’s test for sphericity: (χ2(45) = 904.40, p < .001), Kaiser-Mayer-Olkin test = 0.8.

Item	Universality	Reality
Moral values and norms actually exist rather than being constructed by humans.		0.71
The existence of moral values and norms is directly dependent on the existence of human species. (Reverse item)		0.81
Even if humankind would go extinct, certain actions would remain morally right and other actions morally wrong.		0.78
What is morally right today has been morally right also in the past.	0.58
Moral norms can change over time. (Reverse item)	0.38
If people in the past considered something as moral while we currently believe it is immoral, this is because people in the past were wrong.	0.52
There is just one morality, irrespective of other cultures.	0.85
It is possible to compare different cultures by a single, universal standard of moral rightness	0.84
If people with different cultural backgrounds disagree with each other regarding a moral obligation, it is possible that both are right. (Reverse item)	0.56
The same moral norms that apply to me also apply to all people around the world regardless of whether they come from different cultures or places.	0.80

Results

We predicted that the two individual factors of norm objectivity as well as the whole scale would be positively related to the frequency of participation in collective rituals. The universality of norms was positively predicted by collective ritual participation (β = 0.18, 95% CI = [0.06, 0.29], p = .003) but this relationship did not hold (β = 0.08, 95% CI = [–0.09, 0.25], p = .343) after adjusting the model for moralizing gods (β = 0.16, 95% CI = [–0.02, 0.33], p = .076). Similarly, perceived reality of moral norms was predicted by collective rituals (β = 0.35, 95% CI = [0.24, 0.46], p < .001) but again, disappeared (β = 0.06, 95% CI [–0.09, 0.21], p = .441) after adjusting the model for moralizing gods (β = 0.40, 95% CI = [0.24, 0.55], p < .001). Finally, analyzing the effects of collective rituals on the whole scale of norm objectivity revealed that collective rituals predict norm objectivity (β = 0.28, 95% CI = [0.17, 0.39], p < .001) and that adjusting the model for moralizing gods (β = 0.28, 95% CI = [0.12, 0.45], p < .001) substantially and negatively affected the predictive power of collective rituals (β = 0.09, 95% CI = [–0.07, 0.25], p = .284). See the second row in Figure 2 and Table S9 in SM. As the predictive power of collective rituals disappeared when the model was adjusted for moralizing gods, we computed VIF (moralizing gods = 2.07, rituals = 2.01) for the model and checked the association between collective rituals and moralizing gods (Pearson’s r = 0.71). Although the VIF scores were relatively low, the large correlation between rituals and moralizing gods suggests that multicollinearity may be one of the reasons for our finding that the effects of collective rituals and moralizing gods to some extent cancel each other out.

Discussion

We found that the frequency of participation in collective rituals predicted both norm objectivity factors as well as the whole scale. However, these effects were not reliably estimated when the models were adjusted for belief in moralizing gods. These results are consistent with the results from Study 1B. However, the detected correlation between ritual attendance and belief in moralizing gods did not allow us to disentangle the effects of these two factors.

Study 3

Study 3 assessed the relationship between the novel measure of moral objectivism and ritual participation in pilgrims to Santiago de Compostela. By using the same population as in Study 1B but a novel norm objectivity scale, we were able to test the speculation that poor measurement of norm objectivity affected the results of Study 1B while the low-religious population affected the results of Study 2.

Method

Participants

Participants were recruited via Facebook pilgrim groups in English. Participation was anonymous, voluntary, and not rewarded. In total, 178 respondents participated in the study, but only 137 (see section S4 in SM for details) completed the full questionnaire (90 females, M_age = 57.34, SD_age = 12.01). Participants reported 16 nationalities (see Table S10 in SM).

Measures

We used the same measure of norm objectivity as in Study 2 (English translation). CFA of the norm objectivity construct showed a good fit to the data (RMSEA = 0.080, SRMR = 0.061, TLI = 0.910, CFI = 0.932) and we created the two latent constructs of universality (α = 0.83, seven items, M = 3.11, SD = 1.06) and reality (α = 0.61, three items, M = 3.10, SD = 1.10) as well as the full scale collapsed (α = 0.84, M = 3.11, SD = 0.96). Both subscales correlated with the collapsed scale (r_reality = 0.77, r_universality = 0.96) as well as with each other (r = 0.56). Collective ritual attendance (M = 2.99, SD = 1.66) and belief in moralizing gods (split-half reliability = 0.81, M = 3.19, SD = 1.72) were measured in the same way as in Study 2. On the top of sex and age as controls, we measured education, income, and nationality. For details on measures see section S4 in SM.

Results

We predicted that the two aspects of norm objectivity would be positively associated with the frequency of collective rituals and that this relationship would be stable even after adjusting the model for belief in moralizing gods. Universality of moral norms was positively predicted by collective rituals (β = 0.24, 95% CI = [0.14, 0.34], p < .001). However, this effect disappeared (β = 0.13, 95% CI = [−0.07 0.32], p = .204) when we adjusted the model for belief in moralizing gods (β = 0.37, 95% CI = [0.17, 0.56], p < .001). Collective rituals also predicted the reality of norms (β = 0.44, 95% CI = [0.29, 0.60], p < .001), an effect that held (β = 0.31, 95% CI = [0.11, 0.50], p = .002) even after adjusting the model for moralizing gods (β = 0.26, 95% CI = [0.06, 0.45], p = .010). Finally, the full variable of norm objectivity was positively predicted by collective rituals (β = 0.45, 95% CI = [0.29, 0.60], p < .001), and this effect remained stable, albeit smaller, (β = 0.20, 95% CI = [0.01, 0.39], p = .036) after the addition of moralizing gods to the model (β = 0.37, 95% CI = [0.18, 0.56], p < .001). See the third row in Figure 2 and Table S11 in SM. Assessing the multicollinearity of predictors revealed lower VIF scores (rituals = 1.59, moralizing gods = 1.59) and a smaller correlation between moralizing gods and collective rituals (Pearson’s r = 0.59) compared to Study 2. These results are congruent with the assertion that more stable effects of rituals can be detected in populations where belief and practices correlate to a lesser degree.

Discussion

This study supported the reliability of our norm objectivity scale. Using this scale in a population with a moderate degree of religiosity, ritual attendance generally predicted two of the three factors of norm objectivity even after controlling for belief in moralizing gods. Based on these results, it appears that the failure to detect a unique association between norm objectivity and ritual participation after controlling for moralizing gods in Study 1B might be due to a poor measurement tool.

Interestingly, for the universality factor, moralizing gods overtook the predictive power of collective rituals, while collective rituals had generally larger predictive power than moralizing gods when predicting the reality of norms. These results suggest that the predictive power of ritual participation on the perceived universality of norm application is limited. The belief in moralizing gods (especially in religious traditions with a strong missionary component such as found in Christianity; Lang et al., 2019), on the other hand, maybe associated with the broader applicability of moral norms across cultures, helping the spread of such beliefs (Norenzayan et al., 2016).

Before testing this inference with Christian participants in the US, we improved our two key predictors: participation in collective rituals and belief in moralizing gods. Specifically, the extreme end of the ritual scale (participate “at least once a day”) was switched to “at least once a week” because Christians usually do not attend services more than once a day. To make the moralizing gods variable more precise, we added a question on whether gods care about interpersonal norms (and not only observe and punish transgressions). Finally, we also inspected the specific aspects of the ritual norm that should affect moral objectivity, namely ritual materialization, digitalization, and invariability.

Study 4

This study aimed to replicate Study 3 in a population of the US Christians while also testing the predictions about the specific aspects of ritual behavior that should promote moral objectivity. We addressed the potential issues with missing data by introducing only items with obligatory answers to a financially motivated sample.

Method

Participants

We recruited 302 Christian participants from the US population and paid them $1.7 for survey completion using the server prolific.co. Twelve participants who did not pass one of the two attention checks (participants were asked to choose a concrete option on a scale) were excluded. In total, data from 290 participants (129 females, M_age = 37.37, SD_age = 11.32) were analyzed. All participants completed the whole survey.

Measures

We employed the same scale as in Study 2 and 3 to measure norm objectivity. However, this scale showed a poor consistency with the US participants (RMSEA = 0.161, SRMR = 0.114, TLI = 0.669, CFI = 0.750). Exploring the intercorrelations between items (Figure S1 in SM) revealed that the reversed items formed a separate factor rather than loading on their respective factors of universality and reality. Hence, we removed these reverse items from the norm objectivity scale, which improved the fit to the data (RMSEA = 0.058, SRMR = 0.034, TLI = 0.964, CFI = 0.978). The resulting subscales comprised five items for universality (α = 0.76, M = 3.81, SD = 1.00) and two items for reality (split-half reliability = 0.81, M = 4.01, SD = 1.28). We also collapsed all seven items into one norm objectivity scale (α = 0.84, M = 3.87, SD = 1.00). Again, we correlated the subscales with the collapsed scale (r_reality = 0.84, r_universality = 0.96) as well as with each other (r = 0.67).

The novel construct of ritual aspects included 12 items measured on scale from 1 (totally disagree) to 7 (totally agree). PCA showed that items loaded clearly on the three expected factors, namely on a factor of materialization (α = 0.93, M = 4.84, SD = 1.46), digitalization (α = 0.88, M = 4.62, SD = 1.39), and invariance (α = 0.82, M = 4.75, SD = 1.27). We also created a collapsed scale of ritual aspects (α = 0.92, M = 4.74, SD = 1.16). See, Table 3 for exact wording of the used items and their factor loadings and Figure S2 in SM for intercorrelations.

Table 3. Novel items to measure the feelings about rituals. Bartlett’s test for sphericity: (χ2(66) = 2341.85, p < .001), Kaiser-Mayer-Olkin test = 0.91.

Item	Mat.	Inv.	Dig.
I feel that the rituals of my religion help to make tangible what is otherwise hidden to the senses.	0.90
I feel that the rituals of my religion help to flesh out concepts of my religion that are abstract.	0.87
I feel that the rituals of my religion help to focus our attention upon what is otherwise hard to grasp.	0.89
I feel that the rituals of my religion help to show the realness of the concepts of my religion.	0.77
I feel that the rituals of my religion have fixed steps that must be followed.		0.69
I feel that the essential parts of the rituals of my religion are the same all around the world.		0.76
I feel that the essential parts of the rituals of my religion are the same every time I participate.		0.86
I feel that the essential parts of the rituals of my religion were the same in the past and will be the same forever.		0.79
I feel that the rituals of my religion bring clarity to interpersonal relationships.			0.67
I feel that the rituals of my religion create clear boundaries.			0.90
I feel that the rituals of my religion decrease ambiguity.			0.61
I feel that rituals of my religion help to recognize differences between things.			0.72

Participation in collective religious rituals was measured on a 6-point scale as in the previous study, but we redefined point 6 as More than once a week (M = 3.14, SD = 1.73). To the scale of moralizing gods, we added the item “God or some other divine entity cares about whether we cheat or/and lie.” The whole construct had, thus, three items with very good internal reliability (α = 0.91, M = 4.68, SD = 1.24). In this study, we further included measures of the size of the city of residence and denomination. See section S5 in SM for details on measures.

Results

As in previous studies, we predicted that the frequency of collective rituals should predict norm objectivity even after adjusting the model for belief in moralizing gods (but note that the scale of norm objectivity somewhat differs from previous studies). Furthermore, we also predicted that perceiving rituals as materializing the insubstantial norms, making digital distinctions between rightness and wrongness, and invariably repeating the norms should correlate with perceiving norms as objective. We let the intercepts vary by denomination in our models.

Universality was positively predicted by collective ritual (β = 0.36, 95% CI [0.25, 0.46], p < .001) and the effect held (β = 0.22, 95% CI [0.10, 0.33], p < .001) even after adding moralizing gods (β = 0.33, 95% CI [0.22, 0.44], p < .001) into the model. Collective rituals also predicted the perceived reality of norms (β = 0.31, 95% CI [0.20, 0.42], p < .001), and this estimate remained stable (β = 0.18, 95% CI [0.07, 0.30], p = .002) even after adjusting the model for moralizing gods (β = 0.37, 95% CI [0.26, 0.48], p < .001). Averaging across these dimensions, norm objectivity was positively predicted by collective rituals (β = 0.37, 95% CI [0.26, 0.48], p < .001) and this effect remained stable, albeit smaller, (β = 0.22, 95% CI [0.11, 0.33], p < .001) after the addition of moralizing gods (β = 0.37, 95% CI [0.27, 0.48], p < .001) to the model. See, Figure 2 (fourth row) and Table S12 in SM. VIF scores (rituals = 1.32, moralizing gods = 1.20) and the correlation between moralizing gods and ritual attendance (Pearson’s r = 0.37) were lower than in previous studies, which is in line with our expectation that decoupling these two variables in a religious population would allow us to better separate their effects.

The effects of ritual aspects were analyzed in three steps. In the first step, we included only ritual aspects, in the second step, we also included demography and the frequency of ritual attendance, and in the third step, we included belief in moralizing gods. The composite variable of rituals aspects predicted norm objectivity (β = 0.39, 95% CI [0.28, 0.49], p < .001) and kept its predictive power (β = 0.13, 95% CI [0.01, 0.25], p = .041) after controlling for the frequency of participation in collective rituals and belief in moralizing gods. Examining the individual factors of ritual aspects revealed that invariability (β = 0.17, 95% CI [0.07, 0.28], p = .001) and digitalization (β = 0.14, 95% CI [0.02, 0.25], p = .018) both predicted norm objectivity even in the third-step model. Nevertheless, this was not the case for the materialization factor (β = −0.02, 95% CI [−0.14, 0.10], p = .685; although materialization predicted norm objectivity in the model without moralizing gods: β = 0.15, 95% CI [0.03, 0.27], p < .001). See Table S13 further details.

Discussion

Study 4 replicated the effects of previous studies on the US Christian population. The more frequently participants attended collective rituals, the more they perceived norms as objective. Moreover, the more participants perceived rituals as material and invariant, the more they perceived norms as objective, even when they attended rituals equally frequently and perceived their god as equally moralizing. In support of our conjecture from Study 2, the small correlation between collective rituals and moralizing gods found in this study suggests that collective rituals in Study 2 could not predict objectivity after controlling for moralizing gods because in populations with lower religiosity, the correlation between those two variables was too high to afford their separation.

Compared to previous studies, the scale of norm objectivity did not clearly separate into the two factors, but the reverse items loaded on one factor. Perhaps, scoring low on moral objectivity does not necessarily mean scoring high on moral subjectivity in the US Christian population. Rather than seeing objectivism and relativism as two opposites, the participants in Study 4 formed separate opinions about objectivism and relativism.

On the other hand, we were able to control for the Christian denominations in Study 4. This is important since different denominations could be concerned with morality with varying intensity and their rituals may include different behaviors potentially affecting the perceived moral objectivity. For example, protestant denominations are focused on social norms, strongly monitoring each other’s conduct, leading them to hold more homogeneous values and support impersonal political and legal institutions more than Catholics (Arruñada, 2010), and protestant ritual meetings are filled with longer sermons which include moral and social topics more often compared to Catholics (Pew Research Center, 2021a, 2021b). Another example are charismatic movements and Pentecostalism where rituals encompass singing and synchronous movements leading to ecstatic experiences of unity (Inbody, 2015). Experiencing commonality between individuals may increase perceived objectivity since moral objectivity depends on the shared agreement about the moral values (Goodwin & Darley, 2012), which might be salient to participants in charismatic and Pentecostal rituals. A similar logic applies to the size of the community and an average number of ritual participants. In rituals attended by more people, participants may feel that the values are shared among more individuals leading to higher moral objectivity. Although we were not able to control directly for the number of ritual participants,¹ we used a proxy of city size assuming that in larger cities, masses are attended by more participants. We conducted an additional study reported in SM Section S5.1 with secondary data and found preliminary support for this conjecture.

Study 5

To increase our confidence in the link between collective rituals and norm objectivity and between the perception of collective rituals and norm objectivity found across previous studies and especially in Study 4, we conducted a pre-registered (accessible from https://osf.io/y3r4g/?view_only=b8f39767c14c4969a57a06fea32957d9) replication of Study 4. We used precisely the same questions as in Study 4 for all variables. We also sampled from the same population, i.e., US Christians. The only difference between Study 4 and the current Study 5 was the sample size, which was aimed at 500.

Method

Participants

We recruited 507 participants via prolific.co and paid $0.86 for completing the survey, which took 5–10 minutes. After excluding those who did not complete the main variables of interest (7) and failed at one of two attention checks (14), the final sample size comprised 486 participants (221 females, M_age = 39.59, SD_age = 12.68).

Measures

CFA of the norm objectivity scale with the structure based on Study 4 (reverse items removed) showed a good fit to the data (RMSEA = 0.039, SRMR = 0.029, TLI = 0.989, CFI = 0.993). We created latent constructs of reality (split-half = 0.84, two items, M = 4.10, SD = 1.34) and universality (five items, α = 0.81, five items, M = 3.63, SD = 1.05). We also created collapsed variable of the full norm objectivity scale (α = 0.84, seven items, M = 3.87, SD = 1.01). Both subscales correlated with the collapsed scale (r_reality = 0.78, r_universality = 0.95), and with each other (r = 0.54).

CFA also supported a clear three-factor structure (RMSEA = 0.054, SRMR = 0.029, TLI = 0.976, CFI = 0.982) of the ritual aspects scale. As in the previous study, three latent variables were created: materialization (α = 0.93, M = 5.01, SD = 1.34), digitalization (α = 0.89, M = 4.80, SD = 1.29), and invariability (α = 0.79, M = 4.80, SD = 1.21). The collapsed scale of ritual aspects also showed a good reliability (α = 0.91, M = 4.87, SD = 1.07). Figure S4 in SM shows intercorrelations of the subscale.

We used the same items as in the previous study to measure belief in moralizing gods (three items, α = 0.92, M = 4.72, SD = 1.26) and collective ritual attendance (M = 3.31, SD = 1.73). We measured the same control variables as in Study 4.

Results

As in the previous studies, we hypothesized that collective rituals as well as their specific aspects would predict perceived norm objectivity even after controlling for belief in moralizing gods. Collective rituals predicted norm universality (β = 0.36, 95% CI = [0.27, 0.44], p < .001), and this effect only slightly decreased (β = 0.28, 95% CI = [0.19, 0.38], p < .001) after controlling for moralizing gods (β = 0.19, 95% CI = [0.10, 0.28], p < .001). The frequency of participation in collective rituals predicted perceived reality of moral norms (β = 0.38, 95% CI = [0.30, 0.46], p < .001). The relationship was weaker but estimated with confidence (β = 0.22, 95% CI = [0.13, 0.31], p < .001) when belief in moralizing gods (β = 0.37, 95% CI = [0.29, 0.46], p < .001) was added to the model. Finally, the same positive relationship was found between collective rituals and the full norm objectivity scale (β = 0.41, 95% CI = [0.33, 0.49], p < .001). Adding moralizing gods (β = 0.28, 95% CI = [0.19, 0.37], p < .001) into the model somewhat decreased the effect of rituals, which, however, remained stable (β = 0.29, 95% CI = [0.20, 0.38], p < .001). See the last row in Figure 2 and Table S15 in SM. VIF scores (rituals = 1.33, moralizing gods = 1.26) and the correlation of moralizing gods and ritual participation (r = 0.42) corresponded to the previous results and stable effect sizes of rituals in Study 4.

The collapsed scale of ritual aspects was in a strong positive relationship with the collapsed norm objectivity scale (β = 0.46, 95% CI = [0.38, 0.54], p < .001) and this effect remained stable (β = 0.27, 95% CI = [0.18, 0.37], p < .001) when collective rituals and moralizing gods were added to the model. Models with materialization (β = 0.17, 95% CI = [0.08, 0.27], p < .001), digitalization (β = 0.19, 95% CI = [0.10, 0.29], p < .001), and invariability (β = 0.22, 95% CI = [0.14, 0.30], p < .001) yielded qualitatively comparable results and this applies also to all individual norm objectivity factors. Details reported in Table S16 in SM.

Since each of the tested predictors (ritual performance, ritual aspects, moralizing gods) predicted a unique portion of the variance in norm objectivity, we conducted further exploratory (not pre-registered) regressions probing the mutual relationship of these variables. First, we tested whether belief in moralizing gods would predict higher norm objectivity even without ritual performance (and vice versa) by interacting belief with performance. The model revealed a positive interaction between belief in moralizing gods and collective rituals (β_interaction = 0.12, 95% CI = [0.03, 0.21], p = .008) in predicting norm objectivity, and this effect held (β_interaction = 0.12, 95% CI = [0.03, 0.21], p = .006) after controlling for various confounds. The comparison of intercepts in Figure 3 left (see Table S17 in SM for specific estimates) suggests that ritual performance without belief in moralizing gods does not by itself promote norm objectivity. Likewise, the regression slope of belief in moralizing gods for the lowest levels of ritual performance is comparably flatter than the high levels of ritual performance, suggesting that belief and performance are both necessary ingredients for the ritual effects on norm objectivity take place.

Figure 3. The relationship between norm objectivity and belief in moralizing gods (left) and ritual aspects (right) plotted for three frequencies of ritual attendance. For simplicity of illustration, we trichotomized the ritual performance frequency variable and collapsed estimates from the interaction models for the respective categories.

Furthermore, it may be that perceiving rituals as making the norms material, digital, and invariant as well as perceiving norms as objective may be driven by a third, untested variable. If this is so, the effect of ritual aspects should be independent of ritual participation frequency. We interacted ritual aspects with ritual performance frequency, finding that the more people attend rituals, the stronger is the effect of ritual aspects (β_interaction = 0.08, 95% CI = [0.00, 0.15], p = .037), and this interaction effect was even larger (β_interaction = 0.10, 95% CI = [0.03, 1.17], p = .007) after all the control variables and moralizing gods were added into the model (see Table S17 in SM). Analogically to the results of the interaction between belief in moralizing gods and ritual participation, Figure 3-right shows that a stronger perception of ritual aspects was more effective in the highest compared to the lowest levels of ritual frequency.

Discussion

This study replicated the results of Study 4 regarding the structure of both scales (norm objectivity and ritual aspects) and their mutual relationship. Furthermore, the reported results support the association between the frequency of collective ritual participation and norm objectivity. As in Study 4, these effects held even after controlling for belief in moralizing gods. Crucially, ritual aspects had the strongest effects compared to the measures of simple ritual participation and belief in moralizing gods. Finally, both ritual aspects and belief in moralizing gods interacted with collective ritual frequency in predicting norm objectivity.²

Meta-analysis of the conducted studies

Throughout our studies, ritual attendance had variable effects on moral objectivity when controlling for moralizing gods. Thus, we performed a meta-analysis of the conducted studies using standardized effect sizes and standard errors from regressions. The collapsed norm objectivity scale served as the outcome variable and ritual participation and moralizing gods together with other confounds served as controls (third-step models). We did not include Study 1A effect size due to a different nature of the explained variable (binary). The meta-analysis was performed using the R package meta (Balduzzi et al., 2019). Figure 4 plots the effect sizes and confidence intervals in a tree plot and summarizes the model information. The effect is identified within the confidence interval that excludes zero in both the common effect and random effects models.

Figure 4. Tree plot with further information on meta-analysis. TE stands for standardized effect sizes, seTE for standard errors.

General discussion

We provide initial evidence that collective rituals are associated with moral cognition, namely that the more people attend religious rituals, the more they perceive moral norms as objectively existing. These correlational results were stable in multiple populations with varying degrees of religiosity, although only in the high-religious population this effect remained stable after adding belief in moralizing gods into the model.

As ritual is theorized to generate a commitment to supernatural agents (Sosis, 2003), the association between ritual participation and norm objectivity was not detected in people who both attend rituals and believe in moralizing gods (or neither of them, as was the case in our Czech sample). Based on correlations between belief in moralizing gods and the frequency of collective ritual participation across our studies (Figure 2), it was only in the US Christian population where we observed the de-coupling of religious belief and behavior that allowed us to inspect these factors separately. We found that both variables predicted unique variation in norm objectivity, supporting the idea that belief and behavior constitute a mutually propelling system (Lang & Kundt, 2020; Sosis, 2019); rather than that one or the other would solely facilitate the perception of norm objectivity. The interaction between collective rituals and belief in moralizing gods (and religious guidance of morality) supports this assumption. Importantly, the more people felt that religious rituals are invariant and have materializing and digitalizing potential, the more they perceived moral norms as objective, an association that held even when we controlled for belief in moralizing gods and the frequency of ritual participation. As a matter of fact, ritual aspects had the largest beta coefficient of the three variables under scrutiny. Again, as with belief in moralizing gods, perceiving rituals as invariant and having materializing and digitalizing potential increased the strength of association between norm objectivity and frequency of collective ritual attendance.

Our results provide initial correlational evidence to the proposition that ritual serves as an essential building block of societies founded on moral norms (Rappaport 1999; Lang, 2019). Crucially, our results suggest that rituals may achieve this function not only through their symbolic content, which is cross-culturally highly variable but also through a cross-culturally recurrent form of ritual performance. While the belief-related content of any given ritual is, of course, essential for the realization of the group-specific norms (as suggested by the interaction analysis in Study 1A), it is the universally shared structure of ritual behavior that may help increase the perception of moral norms as objective which, in turn, promotes norm adherence (Rai & Holyoak, 2013; Young & Durwin, 2013). Further support for this conclusion is provided by developmental studies showing that children perceive norms as objective since early childhood (Nichols & Folds-Bennett, 2003), where the passing of norms from caregivers to children is often highly ritualized (Rossano, 2012).

However, while our theory speaks to a causal relationship between collective-ritual participation and norm objectivity, our data and results are correlational and should be interpreted as such. It is possible that people who score higher on norm objectivity are more likely to be attracted to attend religious rituals. For example, a previous study testing the causal relationship between belief in god and norm objectivity found that when people are primed with moral subjectivism, they are less confident about god’s existence (Yilmaz & Bahçekapili, 2015). We suggest that future research should use several types of experimental designs to directly test the causal relationship between ritual performance and moral objectivity. For instance, studies conducted in the controlled laboratory setting would allow us to discern the specific mechanisms that ritual behavior harnesses to promote norm objectivity by utilizing a variety of stimuli where individual ritual aspects could be manipulated. Studies in the field would allow us to test causality by utilizing the pre-post design during real-life collective rituals. Finally, to test whether rituals may affect norm objectivity in the real world, ritual attendance and perception of norm objectivity should be measured repeatedly within one sample in a longitudinal study.

It is also possible that a third variable caused an increase in both ritual attendance and norm objectivity. For example, a recent study found that endorsing moral objectivism is associated with the tendency to see structure in the world (Stanley et al., 2020), while another study showed that when participants are primed with mortality saliency, which is associated with loss of control over one’s life (Arndt & Solomon, 2003), they score lower on moral subjectivism (Yilmaz & Bahçekapili, 2018). It might be that the general trait of structure-seeking in people who experience uncertainty and loss of control in their lives may manifest in increased frequency of ritual participation as well as heightened moral objectivism. From the three aspects of ritual form we investigated, the feeling that rituals are invariant may serve as the expression of such structure-seeking. Indirect support for this speculation may be found in studies from multiple research lines. For example, people attend religious rituals more frequently after wars (Henrich et al., 2019) and during economic crises (Chen, 2010) and group threat, a situation of higher uncertainty and lower controllability, is associated with tightening of norms (Gelfand et al., 2011; Harrington & Gelfand, 2014). Another indirect support from a laboratory study, which operationalized ritualization congruently with our definition of ritual, complements this picture by showing that artificially evoked anxiety leads people to spontaneous ritualization displayed in rigidity and repetitiveness of movements (Lang et al., 2015). Future studies should directly test whether experimentally increasing uncertainty in people high on structure seeking would increase norm objectivity as well as ritualized behavior.

Nevertheless, even if we would assume a causal link between collective rituals and norm objectivity, other mechanisms than digitalization, materialization, and invariance may cause this effect. For example, attending a collective ritual where norms are publicly exposed may increase the perception that the norms are widely shared in the community and, therefore, increase the perception of norms as objective. Previous research has already found that commonly shared norms are perceived as more objective (Goodwin & Darley, 2012). To control for this explanation, we included a variable assessing the city size serving as the proxy for the size of the religious community in our models, but we do not know about the actual size of ritual communities for each participant.

Finally, our scale, but also the scales used in some previous studies (e.g., Zijlstra, 2019), refer to all moral norms, but a related study showed that some values and norms are perceived as more objective than others (Goodwin & Darley, 2012). Thus, another fruitful direction might be to investigate whether participation in a specific ritual may increase the perceived objectivity of the norm associated with that ritual. For example, attending a wedding ceremony may enhance the perception that norms related to loyalty are more objective but affect less (or not at all) the norm of not being nepotistic. Such a detailed examination of real-world ritual effects would provide another important test of the hypothesized relationship between ritual aspects and norm objectivity.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10508619.2022.2121454.

Additional information

Funding

Authors acknowledge support from the Big questions in the study of religion grant (MUNI/A/1435/2021) from Masaryk University, Czech Republic.

Notes

¹ The alternative explanation of the relationship between collective rituals and norm objectivity based on number of ritual participants was recommended to us during the feedback after all studies were conducted. Originally, city size was used as a confound due to the possibility that people in large cities may be more relative because of the faster and modern way of life.

² As we conducted interaction analyses of ritual aspects rather than mediation analyses (despite understanding ritual aspects as a mechanism that rituals harness to increase moral objectivity), it is necessary to clarify this decision. On the conceptual level, obvious aspects of rituals, namely their material potential, digital potential, and invariance, are mechanisms because without these features, according to our theory, the effects on moral objectivity would be lower or nonexistent. This, however, does not imply that the frequency of rituals increases the way people perceive rituals or that rituals became, e.g., more invariant as individuals attend them more often. Since mediation analysis forces researchers to directly assume the causality between three variables, interaction analysis seems less theoretically loaded. Thus, if the activity called ritual does not include the features (although only perceived) serving as the mechanism of the effect, the effect disappears.