Abstract
A growing body of research has documented connections between religious involvement and mental health. However, religion is complex and multidimensional. Religious witnessing, the interpersonal sharing of religious faith, is an important religious practice that has received little attention. Religious witnessing is a relatively unconventional behavior in contemporary American society, yet it can promote social interaction and belonging and has implications for personal identity and sense of self-worth. Using data from a 2010 national random sample (N = 1,342) of U.S. adults, we examine associations between religious witnessing and mental health and the moderating role of public and private religiosity. Mental health is measured using three classes of psychiatric symptoms (general anxiety, social anxiety, and paranoia). Results show that witnessing is related to positive mental health among more highly religious individuals and negative mental health among less religious individuals. Drawing from identity theory and authenticity research, we argue that the contingent impact of religious witnessing on mental health can be explained by (1) participation in social contexts and groups conducive to religious self-expression and (2) the interplay between witnessing, private religiosity, and feelings of authenticity.
Introduction
American religious institutions share their faith messages to nonbelievers in a variety of ways including through the production and marketing of material goods and through media such as direct mail, radio, television, and the Internet (Draper and Park 2010; Park and Baker 2007; Stieverman et al. 2015). Considerable evangelistic efforts also occur less formally through interpersonal interaction. In many instances, religious adherents engage in “witnessing,” the sharing of religious faith through interpersonal communication (Smith et al. 1998:65). Researchers have focused on religious witnessing as an important contributor to the growth of religious groups, the importance of social ties in witnessing, and how specific witnessing activities can influence social capital (Ebaugh and Vaughn 1984; Olson 1993; Priest 2007; Stark and Finke 2000). Religious witnessing may affect aspects of personal well-being such as mental health by promoting social interaction and integration and by providing opportunities to verify valued identities. However, in part due to data limitations, little to no empirical research has been devoted to the role of witnessing in shaping aspects of personal well-being such as mental health. The current study helps address this gap by examining the interplay between religious witnessing, personal religiosity, and three mental health outcomes—general anxiety, social anxiety, and paranoia—in a national sample of U.S. adults.
Background
Religion and Mental Health
Although we know little about the relationship between religious witnessing and mental health, research has repeatedly found connections between religious involvement and mental health (Ellison and Levin 1998; Hill and Mannheimer 2014; Koenig, King, and Carson 2012; Schieman, Bierman, and Ellison 2013; VanderWeele 2017). 1 Moreover, the relationship between religious involvement and mental health appears to be multifaceted, with religious affiliation, beliefs, experiences, and public and private behaviors exhibiting complex relationships with various mental health outcomes.
The literature on religion and mental health is not uniform in its results, finding both positive and negative effects. Researchers have found that frequency of service attendance at religious services is typically positively associated with mental health (e.g., psychological well-being) and inversely associated with mental health problems (e.g., psychological distress and major depression) (Baetz et al. 2006; Ellison 1991; Ellison, Burdette, and Hill 2009; Ellison et al. 2001; Maselko and Kubzansky 2006; Taylor, Chatters, and Nguyen 2013). Although much is still unknown about connections between religious attendance and mental health, explanations of the connections frequently center on the potential role of religion in social support and systems of meaning (Bradshaw and Ellison 2010; Ellison et al. 2009; Stroope, Draper, and Whitehead 2013). Though much of the research in this area has relied on cross-sectional data, longitudinal studies also have emerged in recent years (e.g., Chen and VanderWeele 2018; Childs 2010; Strawbridge et al. 2001). Though proving causation is not possible, some research has used fixed effects regression, which controls for unmeasured time-invariant characteristics and provides more rigorous tests of associations between religious change and mental health change over time (e.g., Croezen et al. 2015).
Religious involvement can also be negative for mental health. Research has found that negative social interactions within religious congregations are associated with mental health problems (Chou and Hofer 2014; Krause, Ellison, and Wulff 1998); individuals may experience shock or a sense of threat to valued relationships and identities when congregational interactions sour. Using the General Social Survey, Ellison and Lee (2010) found three dimensions of spiritual struggles—negative congregant interactions, divine struggles, and doubts—were associated with greater symptoms of anxiety and depression. Even religious service attendance can have negative ties to mental health. For example, Stroope et al. (2019) found that frequency of religious service attendance was associated with higher levels of trait anxiety in South Asians. Effects on mental health may depend on the interplay between religious attendance and other aspects of religiousness. For example, using a large national sample, Lim (2015:700) found mental health benefits in the case of high religious salience combined with high religious attendance as well as a mental health disadvantage in the case of low religiousness on one dimension combined with high religiousness on another. In addition, Lim and Putnam (2010) found that social embeddedness in a religious congregation had a contingent association with life satisfaction among U.S. adults such that embeddedness was more strongly predictive of life satisfaction for individuals with a firm sense of religious identity.
Private religiosity (e.g., prayer frequency) has received less research attention than has religious attendance, yet a complex pattern of findings has emerged. Some research on private religiosity and mental health has found positive associations, some has found negative associations, and some has found little evidence of a relationship (Ellison et al. 2001; Ellison et al. 2009; Hill and Cobb 2011; Shreve-Neiger and Edelstein 2004; Uecker et al. 2016). As with religious attendance, key in understanding these complex results is consideration of how private religiosity interacts with beliefs or other dimensions of religion. For example, cross-sectional (Bradshaw, Ellison, and Flannelly 2008) and longitudinal (Bradshaw and Kent 2018) studies have found that close conceptions of God are more strongly associated with mental health for individuals who engaged in frequent prayer (cf. Pollner 1989). These and other contingent findings from studies of public and private religiosity illustrate the importance of considering how different aspects of religion depend on each other in their associations with mental health (e.g., Abu-Raiya, Pargament, and Krause 2016; Ellison et al. 2013; Krause and Wulff 2004; Schieman 2008; Stroope, Walker, and Franzen 2017). In sum, recent research suggests that mental health may vary according to the religious circumstances in which religious practice occurs.
Identity Processes, Religious Witnessing, and Mental Health
How might religious witnessing affect mental health? Religious witnessing is both (1) a fundamentally social act that involves two or more individuals and (2) an act of self-expression, wherein one actor communicates his or her values, beliefs, and identity to others in a social encounter. In the language of identity theory (Burke and Stets 2009; Stryker 1980; Thoits 1983), religious witnessing constitutes a clear example of identity-related behavior. According to identity theory, identities are consequential for mental and emotional well-being because they are important sources of behavioral guidance, provide individuals with a sense of meaning and purpose in life, and create a sense of belonging within role-based groups and organizations (Thoits 1983). Social roles are linked to cultural meanings and expectations in broader culture regarding how one ought to act, and role participation involves socialization regarding appropriate behavior at the organizational, group, and interpersonal level. In this way, role participation provides individuals with information about “who they are in their own and others’ eyes and why and how they are expected to behave” (Thoits 2012:361), thus reducing anxiety and existential despair. In addition, since role identities often are embedded in social groups or organizations and role relationships are interdependent by their nature, role participation generally promotes a sense of belonging.
The institutional and organizational embeddedness of role identities provides individuals access to social networks and contexts where they can perform identities and receive valuable social feedback regarding role performances. Indeed, successful role performances should foster mental and emotional well-being, as research and theory suggest that positive emotion results from successful identity performances (Burke and Stets 2009). However, identities and identity-related behaviors are not unequivocal sources of positive emotion and enhanced well-being. For example, identity theory’s perceptual control model (Burke and Stets 2009) suggests that the emotions that result from role performances depend on the extent to which one is able to get others to validate and affirm their own self-held identity meanings in a given social encounter. In other words, role performances and identity-related behaviors produce positive emotions to the extent that others respond to our behavior in a way that makes us feel like they see us in the same way we see ourselves. However, if one is unable to verify an identity through role performance, then negative emotions, such as anger, guilt, or sadness, should result. Thus, from this perspective, mental health outcomes are viewed as a result of the emotions produced by attempts at identity verification in social encounters. To the extent that an identity (or identity-related behavior) serves as a consistent basis for self-verification, one’s emotional well-being and mental health should benefit. Conversely, if an identity (or identity-related behavior) serves as a consistent basis for nonverification, then one’s emotional well-being and mental health should deteriorate.
What can the identity theory model of mental health tell us about the interplay of religious witnessing, religiosity, and mental health? First, Thoits’s (1983) work highlights the importance of role-based groups and organizations for providing behavioral guidance, a sense of purpose and meaning, and feelings of belongingness through role-participation and identity expression. Thus, the mental health benefits of identity accumulation and expression may be contingent on the extent to which one’s identity is embedded in role-based institutions, organizations, and groups. Second, identity theory’s perceptual control model highlights the contingent nature of the relationship between identity-related behavior and mental health. Specifically, identity theory suggests that religious witnessing should promote (deteriorate) mental health to the extent that individuals are able (unable) to verify religious identities through the practice of religious witnessing.
In the following section, we turn our attention to identifying features of the self and social environment that can affect the success or failure of religious witnessing as an outlet for identity verification. Specifically, we point to two key features of the self-environment interface that likely matter for self-verification: (1) the availability of audiences that are receptive to religious witnessing and (2) authenticity as a fundamental self-motive.
Public Religiosity: The Availability of Receptive Audiences
At its core, the perceptual control model proposed by identity theory is fundamentally about the fit—or lack thereof—between self and environment. As with any identity, there are likely social contexts that are more conducive to the verification of religious identities than are others. Indeed, given levels of religious-secular social segregation and the secularization of many public spaces in the United States, this fact is particularly true of religious identities (DiPrete et al. 2011; Smith 2003). For instance, if a biology professor engaged in religious witnessing in a public university classroom, it would likely be met with disbelief, embarrassment, and in some cases anger from students. Although many students may agree with the professor’s religious views, American cultural norms strongly suggest that a college biology class is not the appropriate context for religious witnessing. If there are certain contexts that are more conducive to the verification of religious identities, then it is almost certainly true that there are individual differences in the availability of and access to contexts that are receptive to religious identity expression. Although large-scale features of social structure and cultural factors likely contribute to this, we focus here on what we consider to be the most proximal aspect of individuals’ social environment that can provide an outlet for the expression of deeply held religious belief: attendance at religious services.
In the language of identity theory, religious service attendance provides access to “mutual verification contexts” wherein a collection of individuals provide support for each other’s identities (Stets and Burke 2005). Religious service attendance provides the obvious benefit of providing a culturally appropriate context for the expression of religious views among like-minded others. Religious attendance also provides access to (1) networks of others with similar religious views, (2) other religiously affiliated groups and organizations, and (3) a group context within which to talk about proselytizing nonbelievers, proselytizing nonbelievers who may be present, and ultimately verifying religious identities. That is, religious attendance not only opens one context within which one can successfully perform religious identities but can open the door to many other opportunities for religious self-expression. In addition, religious attendance can provide an interpersonal context within which socialization regarding the appropriate narratives and contexts of religious witnessing and religious self-expression can occur.
Based on this, we expect that individuals who attend religious services regularly have more access to contexts that are conducive to religious self-expression, have more opportunities to verify their religious identities, and generally will be more successful in performing religious identities. Thus, for those with high levels of public religious participation, religious witnessing likely is undertaken in social contexts that are conducive to religious expression among like-minded others who have a vested interest in supporting one’s religious identity. On the other hand, for individuals who rarely or never attend religious services, religious witnessing is more likely to occur in contexts that are less conducive to religious expression among others who may not support one’s religious identity. In short, we expect that individuals with high religious attendance will be more likely to experience identity verification from religious witnessing whereas individuals with low religious attendance will likely experience identity nonverification via religious witnessing. This leads us to our first hypothesis:
Hypothesis 1: The relationship between religious witnessing and general anxiety, social anxiety, and paranoia will be more negative (protective) at higher levels of public religious participation (negative interaction of Witnessing × Attendance).
Authenticity and Private Religiosity
Although early work in identity theory focused on more situational identities (i.e., role identities and group identities), recent theory and research has placed more focus on the fact that the self-concept is a system of interrelated identities. Because of this, the concept of authenticity has experienced a recent resurgence in social psychological research and theory. Most empirical research investigating authenticity in social psychology tends to focus on how certain structural characteristics can lead individuals to report having engaged in inauthentic behavior or produce a general sense of inauthenticity (e.g., Didonato and Krueger 2010; Kiecolt 1994; Kiecolt and Mabry 2000; Sloan 2007). Further, research suggests that individuals’ subjective perception of authenticity is associated with various aspects of emotional well-being (Goldman and Kernis 2002; Lakey et al. 2008; Wood et al. 2008). Stets and Burke (2014) suggest that authenticity is an important self-motive and constitutes one of three key dimensions of self-esteem. They argue that verifying more general or global identities (e.g., person identities and social identities) is important for the authenticity dimension of self-esteem “because these meanings become core to the individual” (Stets and Burke 2014:416).
In general, identity theorists and other structural symbolic interactionists argue that individuals are motivated to adopt identities that are consistent with one’s global self-view because doing so bolsters one’s overall sense of authenticity (Lee 1998; Moore and Robinson 2006). Research also suggests that inauthentic self-expression can undermine individual well-being. For example, research on emotion labor (e.g., Hochschild 1979) suggests that inauthentic emotional expression can be cognitively and emotionally taxing. Further, research within psychology suggests that holding role-specific self-views that do not align with one’s global self-view is associated with reduced mental health (Bigler, Neimeyer, and Brown 2001; Diehl, Hastings, and Stanton 2001; Donahue et al. 1993). Applying this to our empirical case, we suggest that individuals with high private religious participation are more likely to view religious witnessing as an authentic expression of self than individuals with low private religiosity. Whereas religious attendance (i.e., public religiosity) can be motivated by extrinsic costs and benefits, we suggest that private religious activities are more directly linked to personal belief and core self-views. This leads us to our second hypothesis:
Hypothesis 2: The relationship between religious witnessing and general anxiety, social anxiety, and paranoia will be more negative (protective) at higher levels of private religiosity (negative interaction of Witnessing × Private Religiosity).
In sum, we expect that both private and public religious participation will moderate the relationship between religious witnessing and mental health. Though we make similar predictions for both private and public religiosity, we link each moderator to distinct aspects of the self-environment interface. Specifically, we expect public religiosity to moderate the effect of witnessing on mental health because religious participation is linked to more opportunities for the successful performance of religious identities. We expect private religiosity to moderate the effect of witnessing on mental health because private religious practice is indicative of the integration of religious belief into the core self-concept, suggesting that religious witnessing is more likely to be an act of authentic self-expression for those with high levels of private religious activity.
Data and Methods
Data
Data for the present study come from the 2010 Baylor Religion Survey, a random national sample of adults 18 years of age and older in the general U.S. population, conducted by the Gallup organization. While other waves of the survey exist, the 2010 wave used in the current analysis is the only wave including a topical module on health asking respondents a subsection of questions pertaining to mental health symptoms. A mixed-mode sampling design was used wherein telephone interviews were conducted using random-digit dialing of listed and unlisted numbers and self-administered mail-in questionnaires. The survey initially contacted 7,000 people via random-digit dialing. Approximately 2,500 of those contacted agreed to participate, and 1,714 of those who agreed to participate returned the survey. Thus, the overall response rate was 24.49 percent, and the contact-to-completion rate was 67.1 percent. The 2010 Baylor Religion Survey was found to be similar to the 2010 General Social Survey on comparable measures including gender, education, age, and religious service attendance (Thomson, Park, and Kendall 2018).
Mental Health Outcomes
Respondents were asked how frequently (never, rarely, sometimes, often, very often) they had experienced an array of mental health symptoms within the preceding month. This series of questions was used to create measures indicating level of symptomology associated with three classes of psychiatric symptoms: general anxiety, social anxiety, and paranoia. Factor analysis was used to confirm the fit of each grouping of three symptoms into single constructs. The questions used for each mental health measure are presented in Table 1, along with the respective factor loadings and internal consistency measures. The individual components of each outcome were combined, resulting in three summative measures ranging from 0 to 12, with higher scores representing higher levels of symptoms.
Factor Loadings and Internal Consistency Measures for Mental Health Outcome Variables.
Key Predictor Variables
We included three measures to assess religious witnessing and levels of public and private religiosity. Respondents were asked, “How often did you participate in the following religious or faith-based activities in the last month,” with “witnessing/sharing your faith” included among the options. Respondents’ answers to this question were used to create a four-point (not at all, 1–2 times, 3–4 times, 5 or more times) ordinal measure indicating the frequency of their witnessing within the preceding month. Public religiosity was measured through frequency of attendance at religious services. Respondents were asked, “How often do you attend religious services at a place of worship,” and presented with a nine-point scale ranging from never (0) to several times a week (8). This item was mean-centered in analyses (Aiken and West 1991).
In addition, a composite measure of private religiosity was created based on respondents’ answers to three questions. To measure subjective religiosity, respondents were asked, “How religious do you consider yourself to be?” and presented with four options ranging from not at all religious (1) to very religious (4). Other questions pertaining to private religiosity focused on practices such as prayer and reading sacred texts. Respondents were asked, “About how often do you spend time alone praying outside of religious services,” resulting in a six-point scale, ranging from never (0) to several times a day (5). Reading of sacred texts was assessed through a nine-point scale ranging from never (0) to several times a week (8) based on the question, “Outside of religious services, about how often do you spend time alone reading the Bible, Koran, Torah, or other sacred book?” These three items were standardized, with the resultant private religiosity index comprising the mean of the standardized items (Cronbach’s alpha = .84). 2
Control Variables
In addition to the key religion variables, we included a range of covariates to control for relevant factors such as sociodemographics, location, and religious affiliation. Dichotomous measures were used as indicators of sex (1 = female), race (1 = white), marital status (1 = married), urbanicity (1 = urban), and residence in the South (1 = South). Age was measured through the use of a continuous variable, while social class was measured through the use of two ordinal variables, including a four-point education measure ranging from less than high school to college graduate, and a seven-point income measure ranging from less than $10,000 to more than $150,000. Since political ideology may be a driver of both religious involvement and health (Hout and Fischer 2014; Pabayo, Kawachi, and Muennig 2015), we controlled political ideology using a seven-point scale, with higher values representing more liberal views. Our measure of religious affiliation relied on the RELTRAD classification scheme (Dougherty, Johnson, and Polson 2007; Steensland et al. 2000). The RELTRAD measure provides a categorization of religious denominations into seven categories based on denominational groupings: evangelical Protestant, mainline Protestant, black Protestant, Catholic, Jewish, other, and none. Evangelical Protestant is set as the reference because it is the largest category (see Table 2).
Descriptive Statistics.
Note. Data are weighted.
Analytic Strategy
Employing a complete case approach to missing data on study variables yielded an analytic sample of 1,342 cases. 3 We begin our analysis with bivariate comparison of means relative to religious witnessing across different levels of public and private religiosity. The analysis then regresses mental health outcomes on religious witnessing, public/private religiosity, and covariates using ordinary least squares (OLS) regression. We then incorporate interactions between religious witnessing and both public and private religiosity.
Results
First, we compared mean mental health scores and religious witnessing. These analyses yielded little evidence of a direct bivariate relationship between religious witnessing and mental health. However, the theoretical considerations underpinning our hypotheses suggest that this relationship may vary relative to the religiousness of individuals who witness. We then proceeded by incorporating analyses of religious witnessing and mental health relative to levels of religiosity.
As shown in Figures 1 and 2, the relationship between religious witnessing and mental health appears to vary according to levels of religiosity on the part of those who witness. Just less than half of the sample (49.7 percent) reported participation in public religiosity at levels below the mean. Of this part of the sample, about 14 percent reported prior month witnessing. Figure 1 shows mental health relative to high and low levels of public religiosity and witnessing. While religiosity and witnessing are treated as continuous variables in analyses, for simplicity of visual presentation, Figure 1 shows predicted levels of mental health and uses religiosity cut points, comparing those who witness with those who do not. Other comparisons could be made, but differences between those who witness and those who do not are likely important. In Figure 1, low–public religiosity individuals reporting witnessing experienced, on average, significantly more frequent symptoms (p < .05) for three outcome variables: general anxiety, social anxiety, and paranoia. In contrast, there appears to be little difference in mental health scores relative to witnessing among high–public religiosity individuals, as lower social anxiety scores were observed among the portion of the sample that reported witnessing.

Mental health relative to witnessing and public religiosity.
A similar pattern emerged with respect to the relationship between mental health and witnessing relative to private religiosity. Approximately 47.8 percent of the sample reported participation in private religiosity at levels below the mean value. Among this portion of the sample, just more than 10 percent reported engaging in religious witnessing in the preceding month. With respect to respondents with low levels of private religiosity, a higher number of symptoms are observed among those who witness, with significant differences (p < .05) found for all three outcomes, as Figure 2 illustrates. The effect of witnessing among those high in private religiosity was not as pronounced, as significant differences (p < .05) were found in only general anxiety. Taken together, these results suggest that there is a relationship between religious witnessing and mental health; however, this relationship is moderated by the religiousness of individuals who witness, with opposing effects observed between low- and high-religiosity individuals.

Mental health relative to witnessing and private religiosity.
On the basis of the aforementioned bivariate analysis, OLS models were then estimated to ensure that results were robust when controlling for relevant factors. We began this phase of our analysis by estimating a series of base models, one for each respective outcome variable. The results of this set of models are presented in Table 3. Consistent with our bivariate analysis, there does not appear to be a direct relationship between religious witnessing and mental health. One exception is that witnessing is significantly associated with lower scores on general anxiety.
Ordinary Least Squares Models Predicting Mental Health Symptoms (Unstandardized Coefficients).
Compared to less than high school.
Compared to less than $10,000.
Compared to evangelical Protestants.
p < .05. **p < .01. ***p < .001.
Guided by our hypotheses and the results of our bivariate analysis, we then extended our analysis with the inclusion of interaction terms in a series of models to assess the extent to which the relationship between religious witnessing and mental health varied according to individual religiosity. The results of models incorporating interaction terms are presented in Table 4. For each outcome, Model 1 features religious witnessing interacted with public religiosity. As shown, the interaction term is negative and statistically significant for all three mental health outcomes, indicating that associations between of religious witnessing and mental health problems are significantly reduced at higher levels of public religiosity. Specifically, at low levels of public religiosity, religious witnessing is generally positively associated with mental health problems. Conversely, for individuals with high levels of public religiosity, there is generally an inverse association between religious witnessing and mental health problems.
Ordinary Least Squares Models Predicting Mental Health Symptoms Relative to Religiosity Levels (Unstandardized Coefficients).
Compared to less than high school.
Compared to less than $10,000.
Compared to evangelical Protestants.
p < .05. **p < .01. ***p < .001.
To present this relationship, we graphed the predicted effects of each relationship (see Figure 3). As shown in Figure 3, there is a consistent relationship whereby more frequent witnessing is associated with greater mental health problems among those with low levels of public religiosity (though the marginal effect is nonsignificant for general anxiety; p = .347), whereas those with high levels of public religiosity experience fewer mental health problems as their level of religious witnessing increases. 4 For each of the three outcomes, mental health levels are similar when religious witnessing levels were low. As religious witnessing occurs more frequently, individuals who have high levels of public religiosity tend to experience improved mental health. In contrast, those who infrequently participate in public forms of religious practices experience increased mental health problems, resulting in a widening gap relative to level of public religiosity.

Mental health relative to witnessing and public religiosity.
A subsequent set of OLS models estimated similar ties between religious witnessing and mental health relative to levels of private religiosity. As shown in Table 4 (Model 2), for all three mental health outcomes, there was a statistically significant relationship between the interaction term and mental health scores, indicating that the association between witnessing and mental health varies according to level of private religiosity. Consistent with the prior set of models, coefficients for the interaction terms were negative for each of the mental health outcomes, indicating a generally positive association between religious witnessing and mental health problems at low levels of private religiosity. This is contrasted by a generally inverse association between religious witnessing and mental health problems among individuals with high levels of private religiosity.
Graphical representations of the interactive effect between religious witnessing and private religiosity for all three mental health outcomes are presented in Figure 4. As shown, there is a consistent pattern of greater mental health problems coinciding with increased witnessing among those with low levels of private religiosity. For those with high levels of private religiosity, a different pattern is observed whereby fewer mental problems are expected at higher levels of religious witnessing. Similar to the previous set of models, the level of mental symptoms is expected to be similar among those with low witnessing. As witnessing levels increase, there is a divergence in mental health symptoms relative to private religiosity. Among individuals with high levels of public religiosity, fewer mental problems are expected at higher levels of witnessing. This stands in contrast to the higher levels of mental health problems expected among those with low levels of private religiosity who engage in witnessing more frequently. As a result, the gap in mental health between low and high levels of private religiosity widens at higher levels of religious witnessing. 5

Mental health relative to witnessing and private religiosity.
Discussion
While researchers have long noted connections between religiosity and mental health (Ellison and Levin 1998; Koenig et al. 2012), one aspect of this connection that has drawn limited attention is how religious witnessing is related to mental health. The present study seeks to help address this gap in the literature by examining relationships between religious witnessing, religiosity, and three mental health outcomes. We do not find a consistent and statistically significant association between frequency of witnessing and mental health. Rather, relationships between witnessing and mental health are consistently contingent on levels of religiosity. With some exceptions, more frequent witnessing among high-religiosity individuals is associated with more favorable mental health outcomes, whereas more frequent witnessing among low-religiosity individuals is associated with less favorable mental health outcomes.
Based on the notion that religious attendance provides a mutual verification context and group setting wherein religious witnessing can result in identity verification, a sense of belonging, and positive emotional experiences, we hypothesized that religious witnessing would be more protective of mental health at higher levels of public religiosity. Given the nonnormative nature of religious witnessing in many social settings, we hypothesized that religious witnessing among individuals with low public religiosity would result in identity nonverification, negative emotions, and feelings of rejection and would therefore ultimately undermine mental health. Our results support these hypotheses and suggest that having access to and participating in religious contexts and groups are important for determining the impact of religious self-expression on mental health.
Drawing from research and theory on authenticity, we argued that individuals who are high in private religiosity would experience religious witnessing as authentic self-expression and experience it as verifying a valued identity central to their global self-concept, thereby enhancing mental health. Conversely, religious witnessing for individuals low in private religiosity should produce feelings of inauthenticity. The findings reported here support this hypothesis, suggesting that the impact of this particular form of religious self-expression on mental health depends on one’s private feelings about religion and the corresponding private religious practices that are undertaken to buttress one’s personal sense of religiosity. Taken together, these two sets of findings strongly suggest that the impact of religious witnessing on mental health depends heavily on (1) the availability of and participation in mutual verification contexts and groups that are conducive to religious self-expression and (2) the private importance of religion to one’s self-concept along with private religious practices that bolster one’s personal sense of religiosity.
More broadly, our findings highlight the potential that other forms of religious self-expression may mirror the contingent relationship between religious witnessing and mental health outcomes. Indeed, although research on such contingencies is “surprisingly rare in the literature,” our results resonate with a thread of work examining how distinct aspects of religion “come together as a ‘package,” to affect mental health (Schieman et al. 2013:471, 470). Some findings show that high religiousness in one domain (e.g., prayer) combines with high religiousness in another domain (e.g., close God concepts) to amplify a positive relationship with mental health (Bradshaw and Kent 2018; Bradshaw et al. 2008; cf. Pollner 1989). Another set of findings shows that low religiousness in one domain combines with high religiousness in another domain to produce a negative relationship with a mental health outcome (Schieman 2008; Uecker et al. 2016). Finally, some research shows the presence of both patterns (Bradshaw and Kent 2018; Bradshaw et al. 2008; Ellison et al. 2014; Lim 2015). Our findings most align somewhere between the second and third categories, that is, research finding some benefits of high religiousness combinations while also consistently finding a well-being “penalty” when there is a dissonance between low religiousness in one dimension and high religiousness in another (Lim 2015:700). Important to note, in the case of the current study, the fact that religious witnessing was protective of mental health for some and detrimental to mental health for others generally masked the true (contingent) effects of religious witnessing in models not including interactions (see Table 3). Thus, future research should not only assess direct effects but also examine whether different forms of religious involvement are tied to mental health in a similarly contingent manner.
This research also contributes to identity theory by highlighting the importance of “mutual verification contexts” for mental health outcomes. Identity theory suggests that identity nonverification produces negative emotions in a given encounter, and recurrent and sustained instances of identity nonverification should undermine mental health by producing prolonged negative affective states. However, little empirical work within identity theory focuses on the structural or environmental factors that can produce such prolonged identity nonverification. Here, we argue that one important structural factor that can produce such recurrent and sustained instances of identity nonverification is a lack of access to mutual verification contexts. In this way, mutual verification contexts provide an important theoretical link between identity (non)verification in particular encounters and the long-term emotional states that could have an appreciable impact on mental health.
In addition to the identity processes and contextual factors outlined here, another potential explanation for the conditional effects of religious witnessing on mental health may center on the motivations 6 underlying such action. Intrinsically motivated individuals would be expected to engage in religious practices more frequently than those who are extrinsically motivated (Steffen, Masters, and Baldwin 2016). Given the established link between intrinsic religiosity and mental health (Steffen and Masters 2005; Steffen et al. 2016), witnessing should provide beneficial effects on mental health for intrinsically motivated individuals, which is consistent with the results of our analyses. Since intrinsically motivated individuals are driven by a desire to follow the tenets of their religious beliefs (Allport and Ross 1967), religious witnessing by these individuals is likely to be based on concern for those to whom they witness. Compassion for the nonreligious is a possible key component of the mechanisms through which the positive relationship between religious witnessing and improved mental health among the highly religious is maintained (Steffen and Masters 2005). Religious witnessing driven by such concern broadly parallels research on volunteering (Cnaan, Handy, and Wadsworth 1996). Indeed, better mental health experienced by high-religiosity individuals who participate in witnessing may be comparable to the mental health benefits of volunteering (Piliavin and Siegl 2015), suggesting that helping others may constitute a component of the link between better mental health and religious witnessing among the highly religious for those who are motivated by compassion.
While intrinsically oriented individuals are motivated by a desire to follow their religious beliefs, extrinsically oriented individuals participate in religion as a means of extracting personal benefits, including social benefits accrued through status gained from witnessing (e.g., recounting witnessing encounters to fellow adherents or enjoying the reputation of being a “soul winner”) (Dayton and Johnston 2001:23). In religious groups that place an emphasis on conversion, individuals who are successful witnesses are held in high esteem by their fellow congregants. An additional implication related to extrinsically oriented individuals who are also less religious is that participating in compassionate acts over time may build and strengthen compassionate attitudes (Steffen and Masters 2005). If this is the case, then for those who are less religious, repeated witnessing over time may eventually yield mental health benefits, thereby reducing the negative mental health effects experienced by being low in religiosity.
It is important to note that several variables were significant in a multivariate OLS regression of witnessing frequency (not shown), including sociodemographic controls, RELTRAD, public religiosity, and private religiosity. Compared to the lowest income category (< $10,000), all other income categories were negatively associated with witnessing frequency. Compared to evangelical Protestants, only religious nones had a significantly different level of witnessing. Nones had a lower level of witnessing compared to evangelicals. Public and private religiosity were both significantly and positively associated with witnessing frequency. Future research should consider the interplay between social class, witnessing, and mental health outcomes.
This study has several limitations. First, we are not able to exclude the possibility of reverse causality. Mental health might enable individuals both to maintain a high level of public and private religiosity (e.g., Whitehead 2018) and to engage in witnessing. It is also possible, particularly in the case of social anxiety, that mental health problems could be an antecedent that determines the extent that one feels comfortable engaging in social behaviors such as witnessing. 7 While future longitudinal research can assess changes in witnessing, religiosity, and mental health over time, at present we are limited by cross-sectional data. Appropriate longitudinal data could allow researchers to study the long-term effects of witnessing. Long-term witnessing might erode the negative effects experienced by low-religiosity individuals. It is also possible that mental health problems could increase after the accumulation of impacts from religious witnessing unmoored from supporting religious communities and practices. Second, the current analysis is limited by its focus on the moderating role of public and private religiosity as measured in this study. Future research should explore other potentially moderating religion effects (e.g., religious/spiritual beliefs) at the intersection of witnessing and mental health. Third, due to data limitations, we were unable to directly observe the social mechanisms theorized. Specifically, although we argue that identity verification and authenticity likely play an important role in how witnessing and religiosity combine in their associations with mental health, we are unable to directly assess these arguments. Thus, future research should attempt to gain more insight into the social mechanisms underlying this relationship. Particularly needed are rich descriptions of those who witness yet are low in other forms of religiosity.
Fourth, like biblical literalism and evangelical Protestant affiliation, religious witnessing is an element of religion found in a minority of the population, limiting its applicability (Schnabel and Bock 2017; Stroope 2011). Nevertheless, by bringing religious witnessing to the study of religion and health, we have drawn further attention to the multidimensionality of religion and underscored the importance of recognizing the interplay between diverse religious practices and mental health. Though our examination of witnessing is novel, it also is limited by measurement constraints. For some individuals, religious witnessing may take the form of praying before a public meal, leaving a gospel tract for a restaurant waiter, engaging in “friendship evangelism,” taking an evangelistic short-term mission trip, or street preaching (Numrich 2009:161; Sagan 2017; Trinitapoli and Vaisey 2009). Also, some witnessing may take place in religious settings and some in secular settings. The rich variety of witnessing and the importance it may hold for a person’s identity and emotions is illustrated in a popular Christian book (Whitney 2014:119): Some of the most rewarding times of my life have been during mission trips when I have done nothing but talk about Christ, on the streets and in homes, with one individual or group after another, all day long. Likewise in my own locale—nothing so excites me as a conversation about Christ with someone who does not know Him. But my experience is not unusual; talking to a person about Jesus can be an intensely rewarding experience for any believer.
Future research should examine a range of witnessing settings and practices among different religious traditions and how they might differentially affect mental health and other health outcomes. In addition, researchers should study how being witnessed to affects one’s emotional state. The above extensions to this line of research would enrich our grasp of the beneficial and harmful potential of religious practice for mental health and advance our understanding of the role of religion in people’s lives.
Conclusion
This research adds to the growing literature highlighting factors that can moderate the link between religious practices and health. Specifically, we find that the relationship between religious witnessing and this study’s mental health outcomes is contingent on levels of public and private religiosity. Drawing from identity theory, we suggest that this contingent relationship is due to (1) the availability (or lack thereof) of mutual verification contexts that are conducive to religious self-expression and (2) authenticity strivings linked to private religiosity and practices that bolster one’s religious identity. In general, our results highlight the importance of religious witnessing and the potential importance of identity processes as a moderating factor in the link between religious practices and mental health outcomes.
