RESEARCH DOSSIER
Religion, Spirituality, and Mental Health Outcomes
Abstract. Religion and spirituality have been associated with diverse mental health outcomes, although the magnitude, direction, and causal interpretation depend on exposure definition, population, and measurement. This fundamental critical narrative research monograph synthesizes historical theories, psychometric instruments, cross-sectional meta-analyses, prospective cohorts, within-person longitudinal models, clinical-context studies, and recent cross-cultural evidence through 2026. Its central finding is that overall associations are commonly small, whereas religious/spiritual struggles, social context, and subgroup position can alter their sign and interpretation. Longitudinal meta-analysis and within-person panel models answer different questions: modest prospective associations cannot, by themselves, establish that changing an individual's religious involvement produces better mental health. The analysis examines depression, anxiety, well-being, suicide-related outcomes, social integration, meaning-making, religious coping, institutional resources and harms, research bias, mediation, selection, and reciprocal causation. It concludes with an evidence-based agenda for construct-specific, culturally sensitive, causally informative research and identifies separate limits for clinical interventions and digitally mediated spiritual support.
Keywords: religion and mental health; spirituality and mental health; religiosity; depression; anxiety; psychological well-being; suicide; religious coping; spiritual struggles; longitudinal studies; random-intercept cross-lagged panel model; causal inference; religious communities; mental health epidemiology; Artificial Era
The Research Object and Its Explanatory Boundaries
Religion, spirituality, and mental health outcomes constitute a research problem about the relationship between forms of sacred orientation and measurable psychological functioning over time. Religion may refer to institutional membership, communal observance, conviction, ritual practice, moral identity, or participation in inherited traditions. Spirituality may describe experienced transcendence, a relationship with the sacred, existential concern, or individually cultivated practices. Mental health outcomes include symptoms of distress, functioning, life satisfaction, psychological well-being, and clinically consequential events. Each variable has multiple operational forms. A valid scientific claim identifies which exposure and which outcome were observed, in which people and setting, over what interval, and under what design. The construct-sensitive pattern documented by Hackney & Sanders (2003) and the longitudinal synthesis of Garssen et al. (2021) make this specification indispensable.
The central question is whether particular religious and spiritual orientations predict later mental health outcomes, under which circumstances such predictions reflect causal mechanisms, and why ostensibly similar studies disagree. Association at one observation can arise from a beneficial practice, a shared social condition, stable personal disposition, prior illness, selection into communities, or reciprocal change. A person who is less depressed may have more opportunities for communal attendance; a person facing acute adversity may pray more intensely; a person who encounters religious exclusion may experience greater distress. Accordingly, this monograph examines positive and adverse associations as components of a heterogeneous evidence landscape rather than as properties inherent to all religious belief Prati (2024) Cowden et al. (2022).
This dossier is a critical narrative synthesis of verified original studies, research methods, and published reviews. Its scope is the epidemiology, prediction, interpretation, and possible causal pathways of mental health outcomes. Individual spiritual practices, the full phenomenology of mystical experience, religious coping typologies, and spiritually integrated treatment protocols belong to adjacent research objects and appear here only where they illuminate outcome evidence. No new participants, original pooled effect, formal systematic-review search, or clinical trial is reported. The evidence architecture compares observational surveys, prospective cohorts, within-person panel models, psychometric validation, clinical-context evidence, and intervention research, and explicitly separates empirical associations from causal interpretation Smith et al. (2003) Bouwhuis-Van Keulen et al. (2024).
Defining Religion, Spirituality, Religiousness, and Psychological Health
Religious identification, self-described spirituality, religious importance, public attendance, private prayer, and perceived religious support occupy distinct conceptual positions. Membership is a nominal or categorical social identification; attendance is an observed or recalled behavioral frequency; importance assesses perceived centrality; prayer frequency records practice but often omits motivation and emotional content; spiritual well-being may contain content extremely close to the outcome labeled psychological well-being. Combining these into a single index changes the causal question being asked. The review by Garssen et al. (2021) distinguished eight categories of religious or spiritual predictors, precisely because their prospective associations can differ. In the German longitudinal panel, Prati (2024) analyzed multiple religious dimensions rather than presuming equivalence.
Mental health is equally multidimensional. Depressive symptom scores, anxiety, psychological distress, self-rated happiness, global life satisfaction, psychiatric diagnosis, psychiatric service use, suicide attempts, and suicide deaths have different ascertainment mechanisms and different implications. A lower depression score does not itself establish the absence of a disorder; higher life satisfaction does not guarantee lower suicidality. Suicidal ideation, attempts, and deaths should be reported as separate outcomes, with age, follow-up time, and endpoint adjudication explicitly identified. The prospective female cohort investigated by VanderWeele et al. (2016) concerns deaths by suicide, whereas the general panel in Prati (2024) focuses on symptoms and self-reported well-being.
The theoretical distinction between mental health as an absence of distress and mental health as flourishing matters particularly in religious research. An organization may offer membership and purpose while simultaneously intensifying moral anxiety for a subgroup. A spiritual crisis can coexist with commitment to a religious tradition, and the apparent positivity of an index may depend on whether distress items are included. The investigators of the Religious and Spiritual Struggles Scale established six distinguishable struggle domains rather than labeling all religious experience beneficial or harmful Exline et al. (2014). A scientifically adequate definition therefore specifies exposure, outcome, population, measurement window, and hypothesized connecting process.
Historical Formation of the Religion–Health Research Program
The psychology of religion emerged from inquiry into lived religious experience, individual orientation, institutional participation, and the psychological functions of belief. Its later health research developed through epidemiology, psychiatry, counseling psychology, behavioral medicine, and sociology. The modern empirical agenda shifted from general judgments about the wholesomeness of religion toward operational comparisons: Which participants attend services, how do they interpret adversity, which symptom scales are administered, and how consistently are associations reproduced? Koenig et al. (2001) systematically assembled the historical and clinical literature available around the turn of the century, documenting the interdisciplinary breadth of the field while leaving study quality and inferential design as continuing challenges.
A decisive conceptual development was the move from undifferentiated religiousness to modes of orientation and coping. The research tradition on internal versus instrumental religious involvement recognized that apparently similar public behaviors can accompany different motives, social identities, or adjustment processes. Researchers subsequently developed detailed methods to capture how religious commitments become available during stress. Pargament et al. (2000) validated an RCOPE framework that recognized potentially constructive and potentially distressing ways of involving religion in coping. Park (2005) analyzed religion as a meaning-making system, with particular attention to responses after bereavement and the changing meaning of adjustment across time.
Meta-analysis broadened the evidentiary horizon while exposing construct inconsistency. Hackney & Sanders (2003) analyzed 34 studies and showed that definitions of both religion and adjustment were associated with differing effect patterns. Smith et al. (2003) assembled 147 independent investigations of religiousness and depressive symptoms, with a small overall negative correlation. Subsequent longitudinal aggregation by Garssen et al. (2021) and within-person modeling by Prati (2024) sharpened the problem of temporal prediction. The history of the field is therefore a history of increasingly differentiated measures and research designs, including an expanding recognition that a statistically significant average may be small and may conceal important variation.
Conceptual Models of Association and Causal Mechanisms
At least four explanatory families are relevant. A social-integration account proposes that collective participation supplies recurring contact, practical help, perceived belonging, and obligations of mutual care. A meaning-regulation account proposes that interpretive frameworks organize appraisals of hardship, mortality, loss, and identity. A behavioral account emphasizes habits and norms that influence sleep, substance use, routine, and help-seeking; such pathways may affect psychological adjustment directly or through health. A stress-and-struggle account addresses threat, guilt, exclusion, conflict, and perceived punishment as potential sources of distress. The literature provides evidence for components of these accounts rather than for a single mechanism universally operating across populations Park (2005) Krause (2006) Cowden et al. (2022).
These models imply different tests. If communal support explains an association, analyses should measure actual ties, reciprocity, perceived acceptance, exposure to discrimination, and changes in belonging instead of treating attendance as a direct psychological intervention. If meaning-making explains change, investigators should capture global beliefs, the discrepancy between events and those beliefs, reappraisal, and subsequent adjustment on an appropriate temporal scale. If religious struggle is a risk pathway, investigators should distinguish whether struggle preceded symptoms, symptoms preceded struggle, or both changed in response to a third event. The three-model longitudinal design tested in adults with chronic illness by Cowden et al. (2022) illustrates the value of explicitly comparing such temporal hypotheses.
Researchers must also allow selection and shared causes. Childhood socialization, family resources, disability, marital status, personality, geographic opportunity, local political climate, and cultural acceptance can influence both religious participation and psychological outcomes. Even highly adjusted longitudinal models cannot guarantee exchangeability when such influences remain unobserved. Thus an association can be compatible with several mechanism families at once; the relevant task is to assess which alternatives an actual design can distinguish. Randomized tests of psychotherapy adaptations concern clinical interventions and cannot establish the causal effects of ordinary religious belief or attendance Bouwhuis-Van Keulen et al. (2024) Prati (2024).
Operationalization and Measurement Validity
Frequency measures are attractive because they can be collected consistently and require little interpretation. Yet the same weekly-attendance category may involve weekly congregational support, family obligation, community surveillance, coercion, seasonal practice, or companionship. The content of participation should therefore be specified whenever it is invoked as a mechanism. Self-report of private prayer may respond to distress itself: a patient experiencing symptom escalation could increase prayer, producing a misleading positive association between prayer and symptoms. Likewise, membership in a religious community may include committed practitioners, cultural affiliates, and people who have disengaged from activities Prati (2024) Garssen et al. (2021).
Composite spirituality scales create a separate psychometric hazard when they contain items about meaning, peace, hope, or harmony while the outcome instrument also measures meaning, peace, or life satisfaction. Correlation may then partly reflect shared content or response style rather than a causal effect of spiritual orientation. A robust study reports item content, factor structure, internal consistency, construct coverage, and evidence for discriminant validity against the chosen mental health outcome. Exline et al. (2014) provides a complementary example in which different kinds of spiritual conflict were separated into divine, demonic, interpersonal, moral, doubt, and ultimate-meaning factors and subjected to factor-analytic tests.
Cross-cultural invariance is essential to comparisons. The term spirituality may represent individual transcendence in one social milieu, shared ritual membership in another, and inherited family duty in a third. Symptom disclosure may also vary with stigma, treatment access, or language. The adolescent cross-cultural review by Mannion et al. (2026) demonstrates the contemporary challenge of drawing general conclusions from studies with diverse religious contexts, variable outcomes, and limited longitudinal data. Before comparing group means or pooled effects, investigators should examine translation, dimensional structure, measurement invariance, differential item functioning, and the effects of participation opportunities.
Cross-Sectional Associations, Meta-Analysis, and Small Average Effects
In the influential synthesis by Smith et al. (2003), 147 independent investigations encompassing 98,975 participants yielded an overall correlation of approximately −0.096 between religiousness and depressive symptoms. Its direction corresponds to slightly lower symptoms among more religious participants on average. The magnitude conveys a modest population association, and individual studies differed in operationalization and the relationship between stressful life events and measured religiousness. The meta-analysis provides evidence that the effect is neither uniform nor reducible to a clinical intervention. Large sample size helps estimate small associations but does not resolve whether exposure and outcome were measured with compatible constructs.
The 34-study meta-analysis by Hackney & Sanders (2003) focused explicitly on whether definition choices explain apparent disagreement. Measures capturing intrinsic religious orientation, attendance, doctrinal involvement, and personal devotion may identify different motivational and social phenomena, while global adjustment, negative affect, and positive well-being register different mental health properties. When researchers combine such measures indiscriminately, a single pooled average may obscure effects with distinct signs and explanatory relevance. This is both a statistical heterogeneity problem and a conceptual validity problem. Results must be compared within well-defined exposure–outcome pairs before a broad statement about religiosity and psychological health can be supported.
A reported p-value or confidence interval establishes precision under the chosen statistical model, not necessarily substantive importance. A correlation around 0.08 accounts for less than one percent of variance in a simple bivariate linear description, although interpretation depends on measurement quality and potential subgroup effects. Because religiosity is socially distributed and mental health influenced by many determinants, even a small average may matter in a population-level context while offering weak prediction for any particular person. This mathematical observation clarifies the size of the prospective pooled result of Garssen et al. (2021); it is not an argument that a small correlation should be called a clinically meaningful treatment effect.
Longitudinal Evidence and the Garssen–Visser–Pool Synthesis
Longitudinal studies order observations in time and can estimate whether a religious exposure predicts later mental health, often adjusting for baseline health and selected confounders. Garssen et al. (2021) synthesized 48 longitudinal studies representing 59 independent samples in a random-effects meta-analysis. Across continuous and dichotomous distress, life satisfaction, well-being, and quality-of-life outcomes, the overall reported association was r = 0.08, with a 95% confidence interval from 0.06 to 0.10. The authors described this as a small positive effect. This summary should be attributed to the particular collection of prospective studies and outcome definitions that entered their synthesis.
Predictor subdivision was informative. Of eight types of religious or spiritual indicators, public religious participation and the perceived importance of religion showed significant pooled associations, reported respectively as r = 0.08 and r = 0.09 in Garssen et al. (2021). These results are compatible with social integration and subjective centrality as meaningful markers, but neither variable is equivalent to an experimentally assigned treatment. The remaining predictor categories did not establish a comparable consistent signal in that synthesis. A scientifically precise reading therefore highlights which particular indicators carried the association rather than generalizing the pooled average equally to private prayer, spiritual meaning, religious coping, and every religious institution.
The limitations concern time ordering, model dependence, and between-study comparability. A prospective predictor can still share causes with an outcome, while baseline adjustment may be incomplete or may introduce bias if a baseline measure is itself affected by earlier exposure. Intervals of months and years can miss short-term distress cycles; community transitions may be sparse; and symptom or life-satisfaction instruments may not have invariant meaning across samples. Longitudinal association is an important evidentiary advance over a single cross-section, but investigators still need designs that distinguish stable differences between people from changes occurring within a person. That next question is central to Prati (2024).
Within-Person Dynamics and the Nine-Year German Panel
A question about an individual changing religious participation is different from a question about whether two individuals who differ in typical participation also differ in mental health. Ordinary cross-lagged panels can conflate these two sources of variation. Random-intercept cross-lagged panel models explicitly separate stable between-person components from within-person deviations, permitting a more focused test of whether a person becoming more religious than usual predicts a later change in their own mental health. Prati (2024) applied this approach to 8,146 German-speaking adults from the GESIS Panel over a nine-year observational window, examining multiple religious indicators with depression, happiness, and life satisfaction.
The principal finding in Prati (2024) was that lagged within-person religious effects on the mental health indicators were almost entirely absent, despite some correlations between stable person-level components. Analyses with covariates and alternative missing-data handling did not recover a consistent protective pattern; a small number of paths pointed in different directions. The study also considered reverse temporal pathways from mental health toward religious involvement and found limited evidence. It directly challenges the interpretation that a modest between-person association necessarily means that encouraging an already-observed individual to engage more religiously will improve their subsequent mental health.
This panel remains observational and population-specific. Random-intercept modeling reduces one class of stable-confounder distortion but cannot eliminate unmeasured time-varying influences, model misspecification, nonrandom attrition, or a mismatch between the time interval of a mechanism and the spacing of measurements. In addition, ordinary participation may have different meaning in a country with distinct patterns of institutional membership and secularization than in communities where religious networks provide vital material support. The synthesis of Garssen et al. (2021) and the results of Prati (2024) can therefore coexist: they answer partially different questions. The first estimates small prospective associations across studies, while the second more directly tests a class of lagged within-person hypotheses in one national setting.
Depression: Symptom Burden, Vulnerability, and Context
The depression literature supplies one of the most extensive empirical bases for investigating religiousness and mental health. Depressive symptoms are ordinarily measured as continuous experiences of low mood, reduced interest, sleep disturbance, fatigue, or negative self-evaluation; diagnostic episodes require additional clinical assessment. The distinction matters because a person may report lower average symptom scores without demonstrating a lower incidence of major depressive episodes. The meta-analysis by Smith et al. (2003) showed a small average inverse association between religiousness and depressive symptoms, but the exposure covered diverse practices and orientations. Observational results should be described as population distributions of symptoms rather than as proof of a preventive intervention.
Several plausible paths produce such distributions. Structured attendance may sustain routines and relationships; a meaning framework may organize appraisals of setbacks; expectations of reciprocal support may reduce perceived isolation; and some norms may promote behaviors associated with adjustment. Yet depression may also change attendance, ability to participate, subjective spiritual experiences, and responses to survey questions. Where depression limits social energy or mobility, the observed absence of participation may partly reflect illness rather than its cause. Stable person-level factors such as family tradition and socioeconomic conditions further complicate attribution. The contrasting designs in Garssen et al. (2021) and Prati (2024) demonstrate why both temporal direction and within-person variation must be investigated.
Religious experience may itself be part of depressive distress in the form of felt divine abandonment, excessive guilt, anticipated punishment, or rupture of communal belonging. Such experiences are operationalized in the struggles literature rather than captured adequately by attendance alone Exline et al. (2014). Longitudinal work in adults with chronic health conditions has tested prospective associations between struggles and psychological distress Cowden et al. (2022). These findings make it inappropriate to assign a universal protective sign to religiousness. The relevant research unit is the combination of religious orientation, social ecology, stress exposure, personal interpretation, and mental health outcome.
Anxiety, Stress, and Clinical Heterogeneity
Anxiety outcomes include general apprehension, stress-related arousal, intrusive worry, panic phenomena, and clinically diagnosed anxiety disorders. Their measurement requires comparable symptom windows and attention to whether distress has been elicited by an immediate event. Religious frameworks can provide continuity, perceived support, and interpretations of uncertainty, yet other religious experiences may amplify threat through fear of transgression, supernatural punishment, or social ostracism. The RCOPE approach explicitly distinguishes different types of religious coping instead of treating prayer or religious identity as a singular anxiety intervention Pargament et al. (2000).
Interpretation of anxiety studies becomes especially difficult when coping is measured after stress has already occurred. Someone experiencing greater anxiety may seek sacred reassurance more often; the resulting positive correlation between prayer and anxiety can reflect response to distress, a distress-generating interpretation, or both. A cross-sectional survey cannot choose among these explanations. Prospective repeated measurement and matched comparisons can help, particularly where assessments record the triggering event, severity, coping content, and subsequent change. Studies of religious struggle suggest that its association with distress is an empirically distinct question from the general association between affiliation and well-being Cowden et al. (2022) Exline et al. (2014).
In clinical populations, diagnostic categories and clinical symptom scales should be kept separate from everyday moral or religious convictions. Intense devoutness is not a diagnosis, and moral conflict cannot be equated with a psychiatric syndrome. Conversely, obsessive, depressive, or trauma-related impairment can shape the content of religious interpretations. The present dossier investigates the evidence for outcome associations; differential diagnosis and treatment planning require distinct clinical evidence and professional assessment. The review of religiously or spiritually adapted psychotherapies by Bouwhuis-Van Keulen et al. (2024) illustrates how clinical intervention trials answer a narrower question than ordinary participation studies.
Suicidal Ideation, Attempts, and Mortality Endpoints
Suicide research requires unusually careful separation of outcomes. Ideation is a reported thought, an attempt is an event involving behavior, and death by suicide is a rare endpoint determined through records and investigation. Associations for one cannot simply be transferred to the other. In a prospective US Nurses' Health Study cohort, VanderWeele et al. (2016) examined 89,708 women and documented 36 deaths by suicide during follow-up. Attendance at religious services at least weekly was associated with substantially lower suicide mortality than never attending, with a reported hazard ratio of 0.16 and a 95% confidence interval of 0.06 to 0.46. Those figures describe one cohort analysis, not a universal effect or a clinical prevention estimate.
The observational result has strengths: prospective exposure, registry-based mortality information, and adjustment for several measured factors. It also has important limits. A small number of events yields unstable subgroup estimates, hazard ratios can be sensitive to residual confounding, and highly selected working women from a professional cohort do not represent all societies, religions, ages, or genders. The authors' analyses considered social integration, depressive symptoms, and alcohol use as possible mediators, but mediation in an observational design requires strong assumptions. A plausible association between attendance and suicide death cannot justify substituting religious activity for established clinical suicide assessment or care VanderWeele et al. (2016).
Heterogeneity is particularly consequential when religious institutions provide belonging to some participants and stigma or rejection to others. National survey analyses of religious attendance and suicidal outcomes report differences by sexual orientation, underscoring that the same nominal attendance exposure need not have the same associations across social positions Park & Hsieh (2023). In subgroup research, effect modification should be prespecified, exposure experiences documented, and absolute outcome probabilities reported alongside relative effects. Clinicians and policymakers should not infer that an intervention proven in a specific religious or demographic group will have the same consequences in groups whose religious setting differs in social acceptance.
Life Satisfaction, Happiness, and Positive Psychological Functioning
Positive outcomes must be examined independently of symptom reduction. Life satisfaction is an evaluative judgment about one's life, happiness often indexes subjective affect or global appraisal, and psychological well-being may encompass purpose, relationships, autonomy, and perceived growth. Religion and spirituality can provide narratives, valued roles, and communal rituals that correlate with these assessments. However, a survey measure of spiritual peace or existential meaning can overlap conceptually with positive well-being outcomes. The prospective synthesis by Garssen et al. (2021) included life satisfaction and well-being among its mental health endpoints, and the panel of Prati (2024) separately measured happiness and life satisfaction.
The long German panel is informative because stable religious involvement and life satisfaction sometimes covaried across individuals while corresponding changes in religious variables rarely predicted later individual changes in life satisfaction Prati (2024). A religious person might consistently report greater satisfaction due to enduring family circumstances, social resources, or response tendencies. That association is useful for description and social epidemiology, but it is a different proposition from asserting that an individual's increase in worship attendance will cause a subsequent increase in happiness. Investigators should ask which comparison is being made and whether the observed measure changes on the temporal scale of the hypothesized mechanism.
Religious life may also support a person's sense of meaning during a loss without causing uniformly higher positive affect. Meaning reconstruction can initially coexist with grief, anger, or existential uncertainty. Park (2005) presented religion as a meaning-making system and illustrated temporal variation in adjustment after bereavement. The coexistence of intense distress and perceived significance is not anomalous; it demonstrates that mental health outcomes are multidimensional. A research program focused on flourishing should report both positive and negative dimensions, the trajectory of adaptation, and the social conditions in which personal meaning is experienced.
Religion, Spirituality, and Mental Health in Serious Illness
Serious medical illness changes the meaning of religious and spiritual variables because exposure and outcome may both respond to diagnosis, disability, treatment burden, uncertainty, and mortality. In cancer populations, Salsman et al. (2015) performed a meta-analysis of religion/spirituality and mental health outcomes. The work underscores the importance of separating positive spiritual dimensions, distressing spiritual experiences, and the particular psychological outcome under study. People may derive comfort from meaning and social support while simultaneously confronting fear, anger, or perceived abandonment. A general-population association should not be mechanically imported into an oncology setting.
An observational relationship between spiritual well-being and quality of life may be partly constitutive when both scales contain peace or purpose-related items. It may also be influenced by symptom severity, functional limitations, socioeconomic support, or access to care. The causal target must therefore be stated: Is the question whether a stable religious orientation precedes later distress, whether changing spiritual struggle predicts symptom change, or whether a specific voluntary supportive intervention affects a clinical endpoint? The multidimensional clinical literature requires different approaches for these targets Salsman et al. (2015) Cowden et al. (2022).
Religious community support in illness can be tangible as well as emotional, including transportation, food, practical caregiving, visits, and sustained contact. A study of church-based support among older people found that such support may moderate the relationship between financial strain and self-rated health under certain sample conditions Krause (2006). Its endpoint was self-rated health, so it should be used as evidence of a possible social pathway, not as a direct estimate of treatment for depression. Studies in oncology and chronic illness are best synthesized through explicitly described mediating processes and verified clinical outcomes rather than an assumed general benefit of religion.
Religious Coping, Meaning-Making, and Stress Appraisal
Religious coping becomes relevant when people confront stressors through sacred meanings or practices. Pargament et al. (2000) developed and initially validated the RCOPE to capture a broad repertoire of methods, including benevolent reinterpretation, support seeking, relational spiritual conflict, and distressing interpretations. The presence of a religious coping response indicates that an event has been processed through religious resources, not necessarily that the resource will improve the eventual outcome. Positive and negative religious coping labels describe different appraisal patterns whose predictive meaning depends on adversity, prior belief, culture, and the mental health endpoint.
Park (2005) articulated a meaning-making perspective in which religion helps organize global assumptions and interpret specific stressful experiences. The research discussed bereavement and temporal variation in adjustment, showing why the quality of the meaning-making process cannot be inferred from simple frequency counts of religious activity. After a traumatic event, a person may seek meaning, challenge prior beliefs, experience moral conflict, or find new commitments. Mental health assessments immediately after the event may capture the process of conflict before longer-term adaptation becomes observable. Prospective designs should therefore record the stressor, initial appraisals, subsequent reassessment, and trajectories of symptoms rather than assuming that meaning is a fixed resource.
This dossier uses coping theory to explain differences in outcomes while preserving the independent identity of the adjacent Research Dossier Religious Coping with Stress and Adversity. A specific causal test would require temporal ordering between coping strategy and subsequent distress, adequate control of initial stress severity, and examination of whether coping is mediator, moderator, or response to symptoms. Because religious coping is strongly contingent on stress exposure, including it as a baseline trait without measuring the stressor can distort a model. Studies of struggle and subsequent distress provide a complementary warning that the content of an interpretation can have clinically meaningful associations Cowden et al. (2022) Exline et al. (2014).
Religious and Spiritual Struggles as Potential Risk Processes
The term religious or spiritual struggle describes experiences of conflict connected to sacred beliefs, communities, practices, or ultimate meaning. Exline et al. (2014) developed a 26-item Religious and Spiritual Struggles Scale with six factors: divine conflict, demonic conflict, interpersonal conflict, moral conflict, doubt, and ultimate-meaning conflict. These are related but distinguishable dimensions. An individual may doubt a doctrine without feeling alienated from a community, experience condemnation by a religious authority while retaining belief, or struggle with moral anxiety independent of institutional participation. Collapsing these experiences into a single measure of low religiosity would erase the relevant psychological phenomena.
The longitudinal study by Cowden et al. (2022) collected repeated observations from 302 US adults with chronic health conditions and compared possible directions linking religious/spiritual struggles and psychological distress. Its value lies in placing alternative temporal models within one empirical framework. Even when such data indicate prospective prediction, potential causal attribution still depends on timing, unmeasured changing stressors, and measurement continuity. Experiences of struggle may be both responses to prior distress and contributors to later distress, with different pathways across individuals. The clinical implication is to investigate subjective experience and impairment without assuming that religious conflict always represents psychopathology.
Religious struggle also intersects with social power. Public attendance can measure access to supportive relationships while simultaneously exposing a minority participant to stigmatizing discourse. Research with sexual-minority populations offers evidence that religious orientation and participation can have different outcome associations depending on acceptance and social location Park & Hsieh (2023). Such differences are essential to understanding divergent average associations. The neighboring dossier Religious and Spiritual Struggles: Conflict, Guilt, and Doubt studies the struggle constructs in detail; the present dossier analyzes how their presence changes conclusions about mental health outcomes.
Congregational Belonging, Social Integration, and Institutional Power
Religious communities can organize regular contact, shared rituals, intergenerational relationships, practical assistance, and an identity that persists across illness or social disruption. These properties make social integration a plausible explanation for associations between attendance and some mental health outcomes. Yet a community is not simply a standardized dose of social support. Membership may be voluntary or obligatory; belonging may coexist with surveillance; ties may provide resources while imposing social costs. An empirically serious community model records actual social ties, support received and given, perceived safety, inclusion, and the availability of comparable secular organizations Krause (2006) Garssen et al. (2021).
The church-support study by Krause (2006) suggests that religiously organized support can modify the relationship between financial strain and self-rated health in specific older US samples. This is useful evidence for stress-buffering hypotheses, although the measure was health rather than a psychiatric disorder. A social-support mechanism becomes more credible when the same social process is observed prospectively and shown to account for an exposure–outcome association without inappropriate statistical adjustment. If researchers control for social support when it actually mediates the effect of communal participation, they may inadvertently remove a mechanism of interest. If they fail to measure social support when it confounds religious participation and health, an estimated association may be misleading.
Institutional authority also shapes health associations. Coercive control, stigma, exclusion, moral condemnation, and restrictions on treatment seeking may alter mental health experiences independently of personal devotion. These processes can be examined through qualitative narratives, institutional comparisons, prospective exposure measurement, and individual outcomes. Generalizations about religion based solely on self-reported belonging are methodologically inadequate when the community experience differs sharply across members. Research should therefore examine both the distribution of community resources and the distribution of community harms Exline et al. (2014) Park & Hsieh (2023).
Mortality, Behavioral Pathways, and the Limits of Psychological Inference
Research relating religious service attendance to mortality overlaps with mental health research but answers a different endpoint question. In the US Nurses' Health Study, Li et al. (2016) examined 74,534 women who had no baseline cardiovascular disease or cancer and compared service attendance with subsequent death during follow-up. More-than-weekly attendance was associated with lower all-cause mortality than never attendance, with an adjusted hazard ratio of 0.67 and 95% confidence interval of 0.62 to 0.71. Death from any cause is not itself a psychological outcome. The study becomes relevant to the present dossier because investigators explored pathways involving depressive symptoms, social support, health behaviors, and optimism, which might connect communal participation to wider health trajectories.
Such mediation estimates require a causal interpretation that is stronger than the raw observational association. To say that a fraction of the estimated attendance–mortality relationship passes through depressive symptoms, smoking, or support assumes that confounding of exposure, mediator, and outcome has been addressed. Some health behaviors may also change over time and influence later participation. The reported proportion of association attributed to a mediator is model-specific; it is not a clinical promise that removing depressive symptoms would replicate the mortality finding. Researchers should distinguish the observed association, the hypothesized pathway, and the assumptions necessary for mediation. The analysis in Li et al. (2016) is therefore useful as a sophisticated example of longitudinal pathway evaluation.
The mortality and mental-health literatures should be read together only at the level that their measures permit. A population might experience reduced mortality risk associated with a social practice while reporting little average within-person change in happiness or depressive symptoms. Those observations could arise from behavior, structural resources, selection, or various pathways with different durations. Prati (2024) estimates lagged mental health relationships over years, while Li et al. (2016) models mortality in a particular professional cohort. Neither design invalidates the other simply because their outcome patterns differ. The proper synthesis begins by respecting their distinct endpoints and identifying common mechanisms that future studies can directly test.
Recent Minority-Context Evidence and the Diversity of Religious Coping
Recent longitudinal research tests the assumption that religious coping has a similar role wherever religious language or practice is present. Kim et al. (2026) examined 369 LGBTQ+ adults who had been raised within or belonged to Latter-day Saint contexts, studying religious coping as a moderator of the relation between loneliness and suicide attempts. The reported interactions differed across subgroups and, contrary to a universal buffering expectation, greater religious coping generally strengthened rather than weakened the loneliness–suicide-attempt relationship in the study's overall pattern. Protective associations were reported only for certain subgroups. This finding is specific to the sample, measures, modeling, and institutional histories examined, yet it supplies an important contemporary challenge to general protective claims.
Religious coping may carry mixed meanings when past participation involved exclusion, family rupture, or conflict with one's identity. A respondent endorsing coping items may be describing comforting practices, repeated attempts at reconciliation, exposure to institutional demands, or renewed engagement with painful beliefs. These are not equivalent psychological processes. The national survey studied by Park & Hsieh (2023) likewise found differential associations between religion and suicide-related measures across sexual-orientation groups. Evidence from different sampling frames should not be merged as a single effect, but their convergence on contextual variation warrants close investigation of stigma, perceived safety, and the content of coping.
Subgroup interactions are especially vulnerable to small sample cells, measurement differences, and analytic multiplicity. Future replication should assess whether minority stress, specific community teachings, degree of affiliation, family acceptance, or prior religious injury moderates the effect, and whether those variables are measured before the outcome. Researchers should describe participant agency and institutional context, because a generic instruction to use more religious coping may be inappropriate in a setting where religious beliefs or practices themselves are involved in the stressor. The relevant scientific goal is the identification of conditions under which a coping process supports adjustment, lacks an effect, or increases psychological burden Kim et al. (2026) Exline et al. (2014).
Clinical Research, Treatment Adaptation, and Ethical Interpretation
The effect of religiously adapted psychotherapy belongs to a distinct intervention literature. In a multi-level meta-analysis of randomized controlled trials, Bouwhuis-Van Keulen et al. (2024) evaluated religious and spirituality-based therapies against standard treatment conditions. Such trials can test the effectiveness of a specified therapeutic format with particular participants and comparison groups. They cannot establish the consequences of every religious institution, and they cannot determine whether involuntary religious participation would confer similar effects. Voluntary client preference, therapist competence, expectancy, therapeutic relationship, and concurrent treatments may influence trial outcomes.
The distinction is clinically important because psychological outcome evidence does not provide grounds for generalized prescriptions to adopt a religion, abandon one, pray, or refrain from spiritual practice. Careful assessment can establish whether a patient's own beliefs and social network are resources, sources of distress, both, or neither. Religious/spiritual struggles should be evaluated on their actual phenomenology and association with functioning, while suspected psychiatric disorders require established clinical assessment and appropriate professional care. Exline et al. (2014) and Cowden et al. (2022) help distinguish observed spiritual conflict from the general categories used in psychiatric care.
Further research should examine adverse outcomes and treatment discontinuation as seriously as mean symptom improvement. Trials should compare adequately specified interventions, account for therapist allegiance and cultural concordance, follow participants after treatment, and analyze whether treatment adaptations work differently for willing religious patients, the uncertain, and the nonreligious. Clinical guidance belongs to responsible professional processes; the scientific role of this dossier is to map associations, hypotheses, and evidence quality. Bouwhuis-Van Keulen et al. (2024) offers a useful experimental comparator while its treatment-specific conclusions remain bounded by the trials that were actually included.
Artificial Era: New Conditions of Spiritual Meaning and Psychological Measurement
The Artificial Era introduces digitally mediated settings in which people discuss religious belief, seek existential interpretations, enter virtual communities, or use conversational systems during distress. These environments can modify access to support, the frequency and privacy of self-disclosure, exposure to advice, and the perceived authority of interpretations. Their existence creates a legitimate research question about how digital mediation changes religious/spiritual participation and measured mental health outcomes. The empirical findings summarized in this dossier largely concern established forms of personal practice and institutional participation; they cannot by themselves establish effects of conversational artificial systems on mental health.
A rigorous prospective study of digitally mediated spiritual assistance would distinguish a human community, an automated informational resource, peer-to-peer online communication, and an artificial conversational partner. Outcomes should include psychological symptoms, perceived social support, autonomy, help seeking, excessive reliance, and the content of spiritual or moral conflict. It should measure baseline religiosity and digital literacy, identify whether participation is voluntary, and compare appropriate secular and human-supported alternatives. Such a program could extend the meaning-making and social-integration hypotheses studied by Park (2005) and Krause (2006), but the extension remains a testable proposal pending primary empirical evidence.
In this research program, Artificial Sapiens designates the publishing author's conceptual intellectual horizon and supplies no independent evidence that an artificial system possesses psychological experience. Observed human outcomes remain the target variables. Future research should test the consequences of digital and human–artificial interaction through specific, preregistered outcome measures and safeguards appropriate to the participants. This methodological separation preserves the discipline's capacity to examine new historical settings without presenting philosophical categories as established clinical findings.
Adolescence, Development, and Religious Socialization
Adolescent mental health research raises questions about the interaction between religious learning, family context, peer belonging, emerging identity, and autonomy. Religious involvement may coincide with shared purpose, prosocial relationships, and social continuity, while punitive or externally compelled engagement may intensify conflict. The exposures and outcomes vary across studies: measures of affiliation, attitudes, attendance, perceived centrality, depressive symptoms, self-esteem, and life satisfaction are not interchangeable. The 2026 cross-cultural review by Mannion et al. (2026) drew on ten studies and reported an overall pattern in which some forms of religious involvement correlated with positive adolescent functioning, alongside context-dependent neutral or adverse findings.
Developmental interpretation requires attention to who determines participation and whether young people can disagree safely. Family religiosity, community expectations, parental warmth, discrimination, and access to peers may all modify the meaning of reported involvement. A child attending services because the family requires it differs analytically from an adult choosing the same activity. Longitudinal research should track identity development, autonomy, changing social networks, and psychological symptoms as multiple evolving processes rather than defining religiosity once at baseline. The small and heterogeneous research base reviewed by Mannion et al. (2026) supports particular contextual observations while leaving universal developmental trajectories unresolved.
Generational changes also complicate cohort comparisons. Secularization, migration, online religious communities, and cultural norms can alter the social resources attached to a religious label. If a survey uses identical questions across generations but those questions refer to different practices or perceived costs, an apparent age effect may be an effect of measurement meaning. A theoretically useful developmental model must therefore include cohort, period, family and institutional context, and current life stage, as well as baseline mental health. The broader synthesis of longitudinal findings in Garssen et al. (2021) provides a point of comparison without settling these developmental questions.
Culture, Secular Contexts, and Cross-Group Heterogeneity
The same ritual practice may communicate solidarity, normalcy, minority identity, safety, dissent, or obligation depending on the surrounding cultural ecology. Cultural majority status shapes access to institutions, exposure to stigma, and the typical meaning of declaring oneself religious. Consequently, coefficients averaged over a population can obscure different associations within relevant groups. A US professional cohort, a nationally representative German panel, a study of hospitalized patients, and a cohort of adolescents from multiple nations sample different social worlds. The transportability of a result is an empirical question rather than an inference from a shared keyword such as religiosity Prati (2024) VanderWeele et al. (2016) Mannion et al. (2026).
An especially informative recent comparison comes from the meta-analysis of German-speaking studies by Zwingmann (2026). Across 39 studies the weighted average correlation between religious/spiritual indicators and mental health was 0.083, with a 95% confidence interval of 0.055 to 0.111. The authors found some differentiation by whether instruments emphasized religiosity, spirituality, or mixed content. This magnitude is close to the small prospective average in Garssen et al. (2021), but similarity of numerical results is not evidence that the same mechanism operates. Both estimates depend on source studies and measurement decisions, and neither demonstrates a general intervention effect.
Social acceptance must be assessed explicitly. In nationally representative US survey data from 2016–2019, Park & Hsieh (2023) studied 163,995 respondents and reported different associations between religious beliefs, attendance, and suicide-related outcomes across sexual orientations. The authors observed lower odds of some outcomes among heterosexual people but less favorable or adverse associations among sexual-minority groups. These are observational subgroup patterns subject to measurement and confounding limitations; they nevertheless demonstrate why a pooled average cannot stand in for the consequences of religious participation for every subgroup. Culturally responsible inference investigates mechanisms of belonging, judgment, discrimination, and safety instead of presuming identical exposure effects.
Selection, Reverse Causality, and Time-Varying Confounding
Selection processes arise when baseline characteristics influence who becomes or remains involved in religious institutions. People with stable housing, family ties, financial resources, or good health may be better able to attend regular services; people with chronic illness or mobility limitations may be excluded without changing their beliefs. Conversely, crises may precipitate increased religious engagement. This means that attendance and mental health can change together because of antecedent circumstances, illness consequences, or an actual effect of community engagement. Prospective sequencing and baseline adjustment improve analysis but do not identify which interpretation is correct unless important assumptions hold Garssen et al. (2021) Prati (2024).
Reverse causality must be studied as a distinct hypothesis. If a depressed person withdraws from a congregation, a later survey may show that low attendance preceded continuing depression even though depression caused the initial withdrawal. If anxiety increases prayer frequency in response to distress, prayer can appear to predict distress while being a response to it. Random-intercept cross-lagged models seek to distinguish stable between-person and varying within-person components, but still depend on observed intervals, adequate sample information, and appropriate model specification. The German panel examined possible paths in both directions and found limited consistent lagged relations Prati (2024).
Time-varying confounders produce another challenge when bereavement, employment loss, illness onset, discrimination, or a transition in marital status affects both religious engagement and psychological health during follow-up. Some of those variables may also be consequences of past participation. A conventional regression adjusting for every observed variable can therefore introduce bias, while failing to adjust can leave confounding. A targeted causal diagram should identify the proposed exposure, outcome, plausible shared causes, mediators, and competing explanations before a model is fitted. The literature supports constructing such diagrams from social, meaning-making, and stress hypotheses Park (2005) Cowden et al. (2022).
Longitudinal Models, Natural Experiments, and Identification
The principal quantitative strategies differ in their estimands. A cross-sectional regression estimates conditional associations at one time; an adjusted prospective cohort estimates associations between earlier exposure and later outcomes; a person fixed-effects model compares an individual's deviations over time; a random-intercept cross-lagged model distinguishes some stable between-person from within-person variation; and a randomized intervention targets the effect of an assigned practice or clinical treatment. These cannot be treated as interchangeable evidence for a single causal quantity. What matters is the intervention or exposure represented, the assignment mechanism, the period of follow-up, and the causal assumptions warranted by the design Prati (2024) Bouwhuis-Van Keulen et al. (2024).
Natural experiments could strengthen inference when religious participation changes due to relocation, institution closures, exogenous policy changes, or other events plausibly unrelated to prior mental health trajectories. However, such events may alter social support, stress, transport, and family relationships independently of religious practice, violating a clean interpretation. Strong research should test pre-event trends, compare affected and unaffected groups on relevant characteristics, analyze attrition, and conduct sensitivity analyses. A claim that community participation prevents depression requires designs able to address competing paths, not merely temporally ordered attendance and outcome surveys. The current meta-analytic literature, including Garssen et al. (2021), offers substantial longitudinal association evidence but does not replace this identification task.
Randomized psychotherapy trials illustrate another boundary. Bouwhuis-Van Keulen et al. (2024) synthesized randomized comparisons of religious/spirituality-based therapy and standard care, finding evidence that certain client-relevant adaptations can improve clinical outcomes in trial settings. This concerns structured treatment received by consenting participants, with therapist contact, expectancy, allegiance, and concurrent treatment as relevant variables. It does not answer whether mere religious affiliation will reduce depression in the general population. An integrated evidence map therefore keeps population-level religiosity, spontaneous spiritual coping, and therapeutic intervention as separate exposure types with separate causal estimands.
Interpreting Evidence Quality, Heterogeneity, and Publication Bias
The quality of evidence depends on the relationship between a claim and the design that supports it. A large cohort can estimate a small association precisely yet still have important unmeasured confounding. A randomized trial can improve causal identification for an assigned intervention while remaining unrepresentative of ordinary participants or of a natural setting. A meta-analysis can aggregate many papers yet inherit common limitations in their exposure definitions, outcome measures, and reporting. Quality appraisal must therefore consider construct validity, selection into samples, baseline symptoms, temporal ordering, attrition, missing data, measurement invariance, and publication bias rather than simply ranking studies by sample size Smith et al. (2003) Garssen et al. (2021) Bouwhuis-Van Keulen et al. (2024).
Effect heterogeneity should be decomposed into scientifically meaningful comparisons: public attendance versus subjective importance; depressive symptoms versus life satisfaction; majority versus marginalized populations; voluntary versus coerced engagement; stable involvement versus acute change; and support-rich versus conflictual community experiences. Statistical heterogeneity in a pooled estimate may correspond to real differences in mechanisms, but it may also arise from inconsistent scales or sampling procedures. The construct-dependent synthesis by Hackney & Sanders (2003), the later German-language evidence by Zwingmann (2026), and the clinical-context findings in Salsman et al. (2015) jointly show why the label religiousness cannot substitute for a precise measurement protocol.
A rigorous research program also reports null estimates, direction reversals, confidence intervals, multiplicity, and pre-specified subgroup tests. Rare outcomes deserve special caution: the suicide mortality cohort of VanderWeele et al. (2016) contained only 36 deaths, making apparently strong relative associations sensitive to event counts and follow-up design. The large German panel Prati (2024) is consequential precisely because its largely null within-person estimates challenge interpretations encouraged by positive between-person correlations. Selective attention to statistically positive studies would create an incomplete and potentially clinically misleading review.
Research Findings Through 2026: What Can Be Established
Several conclusions can be grounded at the level of observational summary. The depression-focused meta-analysis reported a small inverse correlation between religiousness and depressive symptoms Smith et al. (2003). A prospective synthesis found a small average positive relationship between selected religious/spiritual predictors and later mental health Garssen et al. (2021). A more recent German-speaking meta-analysis again estimated a small overall association, with the exact magnitude dependent on instrument categories Zwingmann (2026). In these datasets, the dominant population-level pattern is a small association whose sign and relevance vary by measures, social context, and outcome.
Other findings qualify a universal protective interpretation. A nine-year within-person panel analysis reported little evidence of consistent lagged effects of religion on mental health Prati (2024). Measures of religious/spiritual struggle capture experiences that can coincide with or precede distress under some circumstances Exline et al. (2014) Cowden et al. (2022). An observational US study found different suicide-related associations by sexual orientation Park & Hsieh (2023). Recent cross-cultural adolescent synthesis identified possible benefits alongside heterogeneous outcomes and major gaps in causal research Mannion et al. (2026). These are not isolated anomalies to be ignored; they are among the principal constraints a contemporary theory of religion and mental health must explain.
For suicide deaths, an influential US women's cohort observed a strong inverse association with frequent service attendance, although rare events, selection, and residual confounding restrict causal interpretation VanderWeele et al. (2016). For medical patients, syntheses and longitudinal studies indicate that health-related uncertainty, spiritual distress, support, and well-being must be measured separately Salsman et al. (2015) Cowden et al. (2022). For clinical interventions, trials may support specifically adapted, voluntary approaches under defined treatment conditions Bouwhuis-Van Keulen et al. (2024). These literatures form overlapping but nonidentical evidence domains; their synthesis favors an exposure-specific, outcome-specific, culturally situated account.
Research Priorities and Unresolved Questions
The strongest unresolved question is causal identification. Future studies should distinguish the effect of belonging, the effect of actual support, the effect of devotional practices, the effect of subjective sacred importance, and the effect of institutional experiences. Pre-registration of the target estimand and analysis is especially valuable when many exposures, symptom scales, and subgroups can be selected after observation. Designs need more repeated measures at intervals that match the hypothesized psychological process, together with prospective information about health transitions, religious involvement changes, and social-environmental changes. The discrepancy between Garssen et al. (2021) and Prati (2024) provides a concrete methodological agenda.
A second challenge concerns measurement and generalization. Researchers should report full instrument content, assess discriminant validity of spiritual well-being and positive mental health scales, and evaluate cross-cultural measurement equivalence before pooling countries or religious traditions. Studies should include affiliated and unaffiliated, spiritual and secular, current and former participants; they should document internal diversity within institutions. Particularly important are people whose experiences of religious engagement are ambivalent, constrained, or stigmatizing. The findings of Mannion et al. (2026) and Park & Hsieh (2023) demonstrate why population composition and institutional context are central scientific variables.
A third challenge is mechanism testing. Social connectedness, perceived support, meaning reconstruction, coping appraisal, health behaviors, perceived coercion, and spiritual conflict are plausible intermediary processes. Their causal status remains conditional until temporally specified designs can distinguish mediators from confounders, consequences, and measurement overlap. Cooperation between epidemiology, clinical psychology, social psychology, and qualitative research could connect repeated symptom measures to observations of actual community experience. A study should measure both supportive and harmful features of religious environments, since omitting either can bias theoretical interpretation Krause (2006) Exline et al. (2014) Park (2005).
Conclusion
Research on religion, spirituality, and mental health establishes a heterogeneous set of relationships among social participation, subjective orientation, meaning-making, coping, institutional experience, and measured psychological outcomes. The largest relevant syntheses generally identify small average associations, while the specific predictors and populations matter. Depression, anxiety, well-being, life satisfaction, and suicide-related endpoints require distinct definitions, and the evidence for one cannot serve as a proxy for another. Historical and recent meta-analyses show why the construct called religion or spirituality must be decomposed into actual practices, interpretations, and social experiences Hackney & Sanders (2003) Smith et al. (2003) Garssen et al. (2021) Zwingmann (2026).
The clearest methodological lesson is that correlations between people do not automatically describe the psychological effects of change within a person. The nine-year German panel indicates that stable differences may coexist with little consistent lagged within-person association Prati (2024). Positive observations in particular communities or groups, including prospective lower suicide mortality in an American women's cohort, require careful attention to selection and transportability VanderWeele et al. (2016). Adverse and uneven patterns associated with religious struggle and minority social positions require equal attention Exline et al. (2014) Cowden et al. (2022) Park & Hsieh (2023).
A scientifically adequate account is therefore conditional and mechanism-oriented. Religion and spirituality enter mental health trajectories through identifiable people, institutions, meanings, practices, resources, and conflicts. Empirical explanations must specify the exposure, outcome, population, context, duration, and research design before they make causal claims. This synthesis establishes the conceptual and evidentiary foundations for that research program while leaving intervention efficacy, clinical diagnosis, and the phenomenology of particular sacred experiences to their own scientific domains. Its central finding is the need for precise, context-sensitive explanation grounded in original measurements and appropriately designed longitudinal evidence Park (2005) Bouwhuis-Van Keulen et al. (2024).
References
- Bouwhuis-Van Keulen, A. J., Koelen, J., Eurelings-Bontekoe, L., Hoekstra-Oomen, C., & Glas, G. (2024). The evaluation of religious and spirituality-based therapy compared to standard treatment in mental health care: A multi-level meta-analysis of randomized controlled trials. Psychotherapy Research, 34(3), 339–352. https://doi.org/10.1080/10503307.2023.2241626
- Chen, Y., Koh, H. K., Kawachi, I., Botticelli, M., & VanderWeele, T. J. (2020). Religious service attendance and deaths related to drugs, alcohol, and suicide among US health care professionals. JAMA Psychiatry, 77(7), 737–744. https://doi.org/10.1001/jamapsychiatry.2020.0175
- Cowden, R. G., Pargament, K. I., Chen, Z. J., Davis, E. B., Lemke, A. W., Glowiak, K. J., Rueger, S. Y., & Worthington, E. L., Jr. (2022). Religious/spiritual struggles and psychological distress: A test of three models in a longitudinal study of adults with chronic health conditions. Journal of Clinical Psychology, 78(4), 544–558. https://doi.org/10.1002/jclp.23232
- Exline, J. J., Pargament, K. I., Grubbs, J. B., & Yali, A. M. (2014). The Religious and Spiritual Struggles Scale: Development and initial validation. Psychology of Religion and Spirituality, 6(3), 208–222. https://doi.org/10.1037/a0036465
- Garssen, B., Visser, A., & Pool, G. (2021). Does spirituality or religion positively affect mental health? Meta-analysis of longitudinal studies. The International Journal for the Psychology of Religion, 31(1), 4–20. https://doi.org/10.1080/10508619.2020.1729570
- Hackney, C. H., & Sanders, G. S. (2003). Religiosity and mental health: A meta-analysis of recent studies. Journal for the Scientific Study of Religion, 42(1), 43–55. https://doi.org/10.1111/1468-5906.t01-1-00160
- Kim, S., Lefevor, G. T., Miller, C. K., & Jankowski, P. J. (2026). Is religious coping a protective factor, and for whom? A moderation and subgroup analysis on loneliness and suicide attempts among LGBTQ+ adults raised as Latter-Day Saints. Suicide and Life-Threatening Behavior, 56(2), e70088. https://doi.org/10.1111/sltb.70088
- Koenig, H. G., McCullough, M. E., & Larson, D. B. (2001). Handbook of religion and health. Oxford University Press. https://doi.org/10.1093/oso/9780195118667.001.0001
- Krause, N. (2006). Exploring the stress-buffering effects of church-based and secular social support on self-rated health in late life. The Journals of Gerontology: Series B, 61(1), S35–S43. https://doi.org/10.1093/geronb/61.1.S35
- Li, S., Stampfer, M. J., Williams, D. R., & VanderWeele, T. J. (2016). Association of religious service attendance with mortality among women. JAMA Internal Medicine, 176(6), 777–785. https://doi.org/10.1001/jamainternmed.2016.1615
- Mannion, L., Harmon, M., & O’Brien, T. (2026). Unpacking the complex relationship between religiosity and adolescent mental health: A cross-cultural systematic literature review. Journal of Religion and Health. Advance online publication. https://doi.org/10.1007/s10943-026-02736-w
- Pargament, K. I., Koenig, H. G., & Perez, L. M. (2000). The many methods of religious coping: Development and initial validation of the RCOPE. Journal of Clinical Psychology, 56(4), 519–543. https://pubmed.ncbi.nlm.nih.gov/10775045/
- Park, C. L. (2005). Religion as a meaning-making framework in coping with life stress. Journal of Social Issues, 61(4), 707–729. https://doi.org/10.1111/j.1540-4560.2005.00428.x
- Park, K., & Hsieh, N. (2023). A national study on religiosity and suicide risk by sexual orientation. American Journal of Preventive Medicine, 64(2), 235–243. Published online October 18, 2022. https://doi.org/10.1016/j.amepre.2022.08.020
- Prati, G. (2024). Is religion beneficial for mental health? A 9-year longitudinal study. International Journal of Clinical and Health Psychology, 24(3), Article 100491. https://doi.org/10.1016/j.ijchp.2024.100491
- Salsman, J. M., Pustejovsky, J. E., Jim, H. S. L., Munoz, A. R., Merluzzi, T. V., George, L., Park, C. L., Danhauer, S. C., Sherman, A. C., Snyder, M. A., & Fitchett, G. (2015). A meta-analytic approach to examining the correlation between religion/spirituality and mental health in cancer. Cancer, 121(21), 3769–3778. https://doi.org/10.1002/cncr.29350
- Smith, T. B., McCullough, M. E., & Poll, J. (2003). Religiousness and depression: Evidence for a main effect and the moderating influence of stressful life events. Psychological Bulletin, 129(4), 614–636. https://doi.org/10.1037/0033-2909.129.4.614
- VanderWeele, T. J., Li, S., Tsai, A. C., & Kawachi, I. (2016). Association between religious service attendance and lower suicide rates among US women. JAMA Psychiatry, 73(8), 845–851. https://doi.org/10.1001/jamapsychiatry.2016.1243
- Zwingmann, C. (2026). Religiosity, spirituality and mental health: Meta-analysis of studies from the German-speaking area. Journal of Religion and Health, 65, 132–157. First published online August 12, 2025. https://doi.org/10.1007/s10943-025-02406-3