The Religion and Physical Health Connection  🔗 **Source:** [The Religion and Physical Health Connection](https://wheatley.byu.edu/religion-and-physical-health) > [!note]+ Document Notes > tags: [religion-and-health, academic-rigor, research-critique, echo-chamber, wellness-theology] > author: Loren D. Marks, Harold G. Koenig, et al. (Wheatley Institute, BYU) > source: Wheatley Institute, Brigham Young University > reference: https://wheatley.byu.edu/religion-and-physical-health > full-report: https://wheatley.byu.edu/0000019e-74aa-dce4-a59e-fcaa6b9d0001/religion-and-physical-health > date created: 2026-07-19 > > # 🏷️ Title: The Religion and Physical Health Connection (Wheatley Institute / BYU, Report 2) > > *Deep-read focused on Courtney's question: does the report's academic rigor create and reinforce an echo chamber? Verified against the full 50-page PDF, not just the landing page or secondary coverage.* > <!-- deep-read:sections start --> ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ## 📝 Summary - **What it is**: Report 2 of the Wheatley Institute's three-part *Religion and Human Flourishing* series (BYU), synthesizing physical-health findings from Koenig, VanderWeele, and Peteet's *Handbook of Religion and Health* (3rd ed., 2024, Oxford University Press). Authors: Loren Marks (BYU, lead), Harold Koenig (Duke), Shima Baughman, Jared Robbins, Paul Lambert, Elijah Nevares. Published June 4, 2026. - **Headline claim**: Across 1,069 significant-finding studies spanning 15 physical-health domains, positive associations outnumber negative by roughly 7:1 (876 positive, 124 negative, 69 mixed). Strongest ratios: cigarette smoking 90:1, substance abuse ~43:1, longevity ~15:1, cerebrovascular disease 10:1. Weakest: disease prevention/detection/compliance ~2:1 (and weakening post-2010). - **Why it matters to the rigor question**: The report is the single primary source behind the whole downstream chain (the two news write-ups and the Wired Word lesson all trace here). If its rigor choices bias the result, that bias propagates to every retelling. It does not fabricate — but three specific method choices manufacture a lopsided headline. - **Verified rigor findings (from the full PDF)**: 1. **Null findings excluded by design.** The methodology states plainly that the series "focuses specifically on studies reporting statistically significant associations... This excludes null findings." The 7:1 is a ratio *among studies that already found a significant effect* — not "religion helps 7x more than it harms" in absolute terms. 2. **Vote-counting, not meta-analysis.** The report tallies positive vs. negative significant studies per domain. No effect-size pooling, no confidence intervals, no per-study confounder adjustment at the synthesis level, no publication-bias testing (no funnel plots / trim-and-fill). 3. **No engagement with dissent.** Independent skeptical work (Garssen et al. 2020 meta-analysis; Prati 2023, which found negligible effects using unpublished longitudinal datasets) is absent. - **Bottom line**: The underlying Handbook is genuinely credentialed (Oxford UP; VanderWeele is a serious Harvard epidemiologist), and many primary studies the report cites are rigorous prospective cohorts. But the report's own synthesis layer is a box-score vote count with nulls removed and no bias testing, produced by a religiously affiliated institute that closes with a religious-freedom policy ask. That combination compresses a real-but-moderate literature into an astonishing soundbite ("7:1," "90:1") that then travels through faith-aligned media without friction. The rigor choices do not just risk an echo chamber — they build the echo at the source. ## 📖 Main Content ### 🗒️ Notes ### What the report actually did The authors did not run new studies or a fresh systematic search. They curated from the 2024 *Handbook*, which itself screened a very large corpus (tens of thousands of studies; the report cites Google Scholar decade-counts of 2,980 → 16,600 → 19,900 → 17,400 from 1985–2025) down to the most methodologically rigorous. From that curated pool, the physical-health report counts 1,069 studies with significant findings across 15 domains and reports the direction of each. This is a narrative/vote-count synthesis, explicitly framed as "a concise yet careful overview" of the Handbook rather than an independent meta-analysis. ### The null-exclusion choice (the crux) The report is transparent about excluding nulls: it argues null findings are "typical in any body of medical research" and "neither support nor undermine the central claim." That is defensible *if* the question is narrowly "when religion shows a significant association with health, which direction does it point?" It is misleading when the ratio is then presented publicly as "religion is beneficial 7x more often than harmful," because the denominator has already dropped every study that found no effect. Two different claims get conflated: direction-among-detected-effects (what was measured) vs. prevalence-of-benefit (what readers hear). ### Vote-counting vs. effect size Box-score synthesis counts studies, not magnitudes. A domain could show a 90:1 count with uniformly tiny effects, or a 2:1 count with large ones — the ratio does not say. The field's own leading methodologist, VanderWeele, has argued elsewhere that religion-health effect sizes are small to moderate at best and has developed specific causal-inference methods for longitudinal data. The report cites his rigorous primary studies (e.g., Kim & VanderWeele 2019; Chen & VanderWeele 2018) but does not adopt his cautionary framing about magnitude or causation at the synthesis level. ### Correlation, causation, and self-selection The report's language slides between "associated with" (correct) and causal implication (the mechanisms sections read as causal stories). The "threshold effect" — benefits concentrating among weekly-or-more attenders — is presented as dose-response evidence for a causal effect. But it is equally consistent with self-selection: people healthy enough, resourced enough, and socially connected enough to sustain weekly attendance differ systematically from those who cannot. Several cited primary studies are prospective cohorts that partially address this with baseline controls; the *synthesis* does not foreground reverse causation as a live alternative. ### Where the report is more careful than its retellings Credit where due, and important for a fair verdict: the report concedes disease prevention/compliance is only ~2:1 and has weakened since 2010 (including post-COVID); it flags that some stress-hormone findings are cross-sectional only; it states religious healing should complement, not replace, medical care; and it acknowledges harmful expressions of religion exist. The report is markedly more hedged than the *Deseret*/*Religion Unplugged* coverage or the Wired Word lesson. The echo chamber worsens *downstream* — each retelling strips the hedges. ### The LDS thread in the flagship evidence The marquee longevity/cancer findings originate with Enstrom's studies of ~9,844 Latter-day Saints (1998–99). The report is honest that Enstrom "largely replicated" the effect among non-LDS physically-active nonsmoking weekly attenders, and it cites newer, larger, non-denominational cohorts (Li et al. 2016 N=74,534; Idler et al. 2017; Kim & VanderWeele 2019). So the evidence is not LDS-only — but the institutional chain (BYU institute, LDS-linked flagship studies, LDS-owned *Deseret*/KSL among the press) is worth naming. ### Uncertainty / verification notes This deep-read is based on the full 50-page report PDF plus the landing page, cross-checked against external literature (Garssen 2020; Prati 2023; VanderWeele's commentaries). I did not independently audit the 2024 *Handbook*'s per-study inclusion coding, so claims about how individual studies were quality-rated rest on the report's and Handbook's own descriptions. ## 📚 References - Primary: Marks, Baughman, Koenig, Robbins, Lambert, Nevares (2026). *The Religion & Physical Health Connection*, Wheatley Institute, BYU. Full report: https://wheatley.byu.edu/0000019e-74aa-dce4-a59e-fcaa6b9d0001/religion-and-physical-health - Source corpus: Koenig, VanderWeele & Peteet, *Handbook of Religion and Health*, 3rd ed. (2024), Oxford University Press. - Secondary coverage (all faith-affiliated): *Deseret News*, *Religion Unplugged*, *Christian Post* (x2), *Christianity Daily*, *FāVS News*, *KSL*. - External counterweight (not cited by the report): Garssen et al. (2020) meta-analysis and commentaries (Int. J. Psychology of Religion, 2021); Prati (2023) on negligible effects using unpublished longitudinal datasets; VanderWeele's methodological work on causal inference for religion-health. - Author verification: report authored by Marks/Koenig et al.; VanderWeele and Peteet authored the Handbook, not this report. None of this is Courtney's work. ## 🗣️ Paraphrase A BYU institute took the most rigorous religion-health studies catalogued in the 2024 Oxford Handbook, kept only those that found a statistically significant effect, and counted how many pointed toward benefit vs. harm across 15 physical-health areas. Benefit won roughly 7 to 1 overall, and by lopsided margins for smoking and addiction. The report is careful in spots — it admits weak domains, cross-sectional limits, and that religion should not replace medical care — but its core method is a study head-count with null results removed and no test for publication bias, and it never engages the skeptical literature. So the striking ratio is partly an artifact of what was counted, and it becomes more overstated with each downstream retelling. ## 📰 Newsletter Not applicable — this is a religion-and-health research report for personal/church study, unrelated to WordPress, plugins, or DevRel work. ## ✅ Action Items #### DevRel / OSPO / Community / WordPress / The WPCC / FAIR - Not applicable to this source. #### Personal / Church (St. John's) - Treat the 7:1 and 90:1 figures as direction-among-significant-findings, not "benefit vs. harm" prevalence, if referenced in class or worship. - If used devotionally, pair with one self-selection question and one "whose voices are missing" question (see the Wired Word lesson note). ## 🧪 Hypotheses - If the 7:1 ratio were recomputed with null findings re-included, the impression of overwhelming benefit would shrink markedly even if direction stayed positive. - If the same 1,069 studies were meta-analyzed by effect size rather than vote-counted, pooled effects would land in VanderWeele's "small-to-moderate" range, not the "almost never seen in social science" framing. - The domains most driven by behavior (smoking, substance use) will survive causal scrutiny best; the indirect-pathway domains (cancer, cerebrovascular, Alzheimer's) will attenuate most under confounder adjustment. - Downstream retellings will strip hedges monotonically: report (most hedged) → news (less) → Wired Word lesson (least). Testable by coding hedge density at each layer. ### 📊 Methodology - Obtain the 2024 *Handbook* chapter-level inclusion criteria and confirm whether study *quality* was formally scored/weighted or whether all "high-quality" studies were counted equally regardless of design strength. - Re-express a sample domain (e.g., mortality, 89 vs. 6) as pooled effect sizes with confidence intervals to compare against the vote-count impression. - Run a publication-bias check (funnel plot / trim-and-fill) on a domain with enough studies (substance abuse N=271) — the report performs none. - Code the report, the two news pieces, and the Wired Word lesson for hedge density and causal-language creep across layers. ### ❓ Questions - What fraction of the originally screened studies returned null results, and how large would the denominator be if they were included? - Were effect sizes ever extracted, or only significance direction? (The PDF suggests only direction.) - For indirect-pathway domains, how much of the association survives adjustment for socioeconomic status, marital status, and baseline health? - Would matched secular high-attendance communities (choirs, unions, mutual-aid networks) show comparable benefits — isolating "belonging" from "religion"? ### 🔮 Future Research - A true effect-size meta-analysis of the same corpus, with confounder and publication-bias handling, to test whether the vote-count impression holds. - A head-to-head of religious vs. secular high-commitment community membership on the same health outcomes. - A provenance/hedge-decay study tracing this specific report through its media and curriculum retellings — a clean media-literacy case study. ### 💡 Implications - **Research use**: Cite the report only as a directional literature summary, never as evidence of effect magnitude or causation. Pair any citation with the null-exclusion caveat. - **Public health**: A real signal likely exists (especially for behavior-mediated outcomes), so faith communities may be a legitimate partner channel — but the report overstates certainty for policy purposes, and its religious-freedom recommendation is a values claim riding on contested effect sizes. - **Church/pastoral**: Useful for affirming embodiment theology; risky if members read "7:1" as a promise that faithfulness buys health — pastorally harmful for the chronically ill or disabled. - **Media literacy**: This is a clean teaching example of how method choices at a source (null exclusion + vote-counting) plus aligned distribution channels manufacture an apparent consensus. ### 🧠 Learning Styles Assessment - Reading level: report body ~ grade 12–college (formal, citation-dense); landing-page summary ~ grade 10. #### 👀 Visual learners - Present: bar charts per domain (Figure 3, Figure 4) make the vote counts scannable. - Improve: the charts show counts, not effect sizes or confidence intervals — a forest plot would represent uncertainty honestly. #### 🔊 Aural learners - Present: an embedded YouTube overview video accompanies the landing page. - Improve: none critical. #### 🗣️ Verbal learners - Present: the "salvation = wholeness" Tillich framing and per-domain mechanism narratives aid comprehension. - Improve: define "vote-count synthesis" vs. "meta-analysis" for lay readers, since the distinction is decisive. #### 🧩 Kinesthetic learners - Present: policy recommendations give concrete institutional actions. - Improve: n/a for a research report. #### 🧠 Logical learners - Present: transparent per-domain counts let a careful reader audit the aggregate. - Improve: the leap from "significant-finding direction" to "7x benefit" is not flagged; logical readers should be told nulls were removed. #### 🧑‍🤝‍🧑 Social learners - Present: heavy emphasis on community/social-connection mechanisms. - Improve: none. #### 🧘‍♀️ Solitary learners - Present: full PDF supports independent deep study. - Improve: none. #### 🌿 Naturalistic learners - Present: diet/fasting/plant-forward sections connect to food and body. - Improve: none. ### 📊 Accessibility Assessment - **Images**: the landing-page photo has descriptive alt text (good); charts have figcaptions describing the data (good). - **Structure**: report uses clear headings and a TOC; screen-reader navigable. - **PDF**: 10-point dense body text noted by the authors themselves ("1,100 oversized pages" for the Handbook); the report PDF is more readable but data-dense. - **Contrast/keyboard**: standard web page; no obvious violations detected without live rendering. - **WCAG note**: charts convey meaning via color (positive/negative/mixed) — verify non-color cues exist for color-blind readers. ### 🌍 DEIB Assessment - **Cultural breadth**: genuinely multi-tradition in examples (kosher, halal, Buddhist/Hindu vegetarianism, Ramadan fasting, zakat/tzedakah), and cites international cohorts (China, Mexico, Israel, Europe). - **Representation of harm**: acknowledges harmful/coercive religion and vaccine-refusal outbreaks, though briefly. - **Equity signal**: notably centers larger longevity benefits among African American attenders (13.7 years) and among low-income and minority populations — but be cautious: framing structural health disparities as offset by religiosity can understate socioeconomic confounders. - **Institutional standpoint**: a faith-affiliated publisher studying faith's benefits, closing with a religious-freedom policy ask — a values position that should be named, not hidden. - **Inclusive-language caution**: health-and-longevity framing can land painfully for disabled or chronically ill readers; the report's "purpose reframes pain" language risks minimizing real suffering. ### ✅ Logical Validity #### Strong reasoning - Transparent, auditable per-domain counts; the aggregate is not hidden behind a single number. - Honest concessions: weakest domain named (2:1), post-2010 decline flagged, cross-sectional limits noted, "not a cure-all" stated, complementary-to-medicine stated. - Multi-pathway mechanism models (behavioral, social, psychological, biological) are coherent and map to the outcome categories. #### Fallacies / weaknesses - **Suppressed evidence (by design)**: null findings excluded; skeptical syntheses (Garssen, Prati) unengaged. Fix: report the null count and address dissent. - **Vote-counting masquerading as strength**: a ratio of significant-finding directions is presented as decisive without effect sizes or CIs. Fix: pool effect sizes; report magnitude and uncertainty. - **Cum hoc ergo propter hoc**: associational counts narrated as causal mechanisms. Fix: label associations; foreground reverse causation. - **Self-selection framed as dose-response**: the threshold effect is read as causal evidence when it is equally a selection signature. Fix: model who sustains weekly attendance. - **Appeal to authority / ratio framing**: "almost never seen in the social sciences" (Marks) treats an extreme ratio as strength when, in a values-adjacent field with nulls removed, an extreme ratio is also a red flag for selection/publication artifacts. Fix: run bias diagnostics before celebrating the ratio. - **Institutional confirmation bias**: religiously affiliated producer + pro-religion result + religious-freedom policy ask; not disqualifying, but demands the bias diagnostics the report omits. #### Tone of voice - Measured and academic in the body; more triumphal in the summary and press framing. The report itself is the least echo-chamber-y layer; the danger is compounding downstream. ### 🛡️ DARVO Analysis Not applicable. The report makes no accusation and mounts no defense against a specific accuser; there is no deny/attack/reverse-victim structure. The nearest rhetorical move is pre-empting critics by asserting the evidence lets readers "move beyond dogmatic assertions" from "religious advocates or critics" — a both-sides framing that positions the report as neutral arbiter while it advances a directional conclusion. That is a framing choice, not DARVO. ## 🎯 Conclusion On Courtney's specific question — does the report's academic rigor create an echo chamber — the answer is a qualified yes, located precisely. The report is honest and even careful in places, and it sits on a credentialed source (the Oxford Handbook) and cites genuinely rigorous primary studies. But three synthesis-level choices — excluding null findings, vote-counting instead of measuring effect sizes, and ignoring the skeptical literature — pre-load a lopsided headline. Produced by a faith-affiliated institute, closing with a religious-freedom policy ask, and received almost entirely by faith-affiliated media, that headline then travels down to the Wired Word lesson and a Sunday-school class as manufactured consensus. The echo begins in the method, not just the retelling. Practical next step: whenever citing this report, state that the ratio counts only significant findings (nulls removed) and reflects direction, not magnitude — and pair it with the dissenting literature (Garssen 2020; Prati 2023) to restore the missing side. ## 🔄 Refactor Appendix — The Religion and Physical Health Connection ### 📚 Literature Atoms (H3) #### The 7:1 ratio counts only significant findings, with nulls removed > [!quote] ID:wheatley_phys_1 > "This excludes null findings, which are typical in any body of medical research and which neither support nor undermine the central claim." **Context & use**: - From the report's Scope & Methodological Approach. This is the single most important line for the rigor/echo-chamber question: it means "7:1" is direction-among-detected-effects, not benefit-vs-harm prevalence. - Load-bearing evidence for the "suppressed evidence (by design)" flag. **Tags**: #xref #rigor <!-- atomicity: 1 idea ✓ --> #### An extreme ratio in a values-adjacent field is a red flag, not only a strength > [!quote] ID:wheatley_phys_2 > "That kind of ratio is almost never seen in the social sciences." (Loren Marks) **Context & use**: - Marks frames the lopsided ratio as remarkable strength. With nulls removed and no publication-bias test, an extreme ratio equally signals selection/publication artifacts. - Anchor for the "appeal to authority via ratio framing" fallacy. **Tags**: #xref #rigor #aphorisms <!-- atomicity: 1 idea ✓ --> #### The benefits depend on protecting religious freedom (the policy ask) > [!quote] ID:wheatley_phys_3 > "The benefits from religious living identified in this report are dependent upon the ability of individuals and groups to freely practice and live out their specific faith commitments." **Context & use**: - The report's fifth policy recommendation. Shows the values standpoint riding on contested effect sizes — relevant to institutional-confirmation-bias analysis. **Tags**: #xref #rigor <!-- atomicity: 1 idea ✓ --> ### 🧠 Permanent Candidates (H4) #### Excluding nulls turns a modest literature into an astonishing soundbite **Claim** (one sentence, Courtney's voice): When a synthesis counts only studies that already found a significant effect and then reports the positive-to-negative ratio, it measures the direction of detected effects but is routinely read as the prevalence of benefit — a substitution that manufactures apparent consensus. **Evidence pointer**: [[#The 7:1 ratio counts only significant findings, with nulls removed]] **Scope & conditions**: Applies to any vote-count / box-score synthesis that excludes null results, especially in values-adjacent fields; does not apply to effect-size meta-analyses that model all results and test for publication bias. **Links**: - [[Media literacy]] — source-side manufacturing of consensus - [[Echo chamber]] — method-level origin, not just distribution <!-- atomicity: 1 idea ✓ --> #### The source can be more careful than every retelling of it **Claim** (one sentence, Courtney's voice): A research report can carry honest hedges (weak domains, cross-sectional limits, "not a cure-all") that are stripped monotonically as the finding passes through news coverage and into curriculum — so the echo chamber can worsen downstream even when the origin is comparatively responsible. **Evidence pointer**: report concedes 2:1 weakest domain and post-2010 decline; Wired Word lesson drops these. **Scope & conditions**: Applies to multi-layer science communication chains; the fix is to read upstream to the primary source. **Links**: - [[The Wired Word 07/19/2026 note]] — the downstream lesson that stripped the hedges <!-- atomicity: 1 idea ✓ --> ### ✓️ Fleeting Queue - [F1] Recompute one domain (mortality: 89 vs 6) as pooled effect sizes with CIs to test the vote-count impression — https://wheatley.byu.edu/0000019e-74aa-dce4-a59e-fcaa6b9d0001/religion-and-physical-health #rigor - [F2] Pull Garssen et al. (2020) and Prati (2023) for the skeptical counter-file — #xref ### → Things 3 - [ ] Before citing the Wheatley 7:1/90:1 figures anywhere, add the null-exclusion + direction-not-magnitude caveat — media-literacy discipline #things - [ ] Read Garssen (2020) + Prati (2023) to hold the skeptical side of the religion-health literature #things <!-- deep-read:sections end --> 📄 **Full text:** [[Full Document Contents/Articles/The Religion and Physical Health Connection.md|The Religion and Physical Health Connection by Wheatley Institute]] ## Highlights - **This current report, *Religion and the Physical Health Connection*, represents part two of the series. It specifically examines the relationship between religious involvement and measurable physical health outcomes.** Drawing on the most rigorous, high-quality studies that met the *Handbook*’s stringent inclusion criteria, 876 of 1,069 studies (82%) reported significant findings which identified positive correlations between religion and physical health. In contrast, 124 studies (12%) reported negative associations (for example, unhealthy religious influences), while 69 studies (6%) yielded mixed or complex results. **Taken together, among the best available evidence, positive associations outnumber negative ones by roughly 7:1.** ([View Highlight](https://read.readwise.io/read/01kxxa8h0gd61penm5r3mgq44e)) - Note: The 7:1 headline. Verified in the full PDF: this ratio counts ONLY studies with a statistically significant finding — nulls were excluded by design. So it's direction-among-detected-effects, not "benefit 7x more often than harm." Vote-count, not effect-size meta-analysis. - Tags: echo-chamber, academic-rigor - In addition, research consistently shows a threshold effect: the benefits of religion concentrate among those with sustained, high engagement. Active participation is linked to better mental and physical health—including lower cancer rates, greater longevity, and stronger marriages and families. Meaningful commitment tends to yield the greatest benefits, typically seen among weekly attenders across ages and faiths. From a health-promotion perspective, this growing body of evidence provides science-based support for active religious involvement, with wide-ranging [mental](https://wheatley.byu.edu/religion-and-mental-health), physical, and relational benefits extending across individuals, families, and communities. ([View Highlight](https://read.readwise.io/read/01kxxa8r3v9cehtcdffmtva3qg)) - Note: Threshold effect framed as dose-response causal evidence. But it's equally a self-selection signature: people healthy, resourced, and connected enough to sustain weekly attendance differ at baseline. Reverse causation not ruled out at the synthesis level. - Tags: academic-rigor, self-selection - While religion is not a cure-all, and harmful expressions do exist, the overall pattern is neither mixed nor evenly divided. Instead, across hundreds of high-quality studies, positive associations outweigh negative ones by wide margins. As societies confront rising rates of mental illness, addiction, social fragmentation, and family instability, the evidence suggests that **religion remains a consequential and often underrecognized contributor to human flourishing**. Understanding how, when, and why religion promotes well-being is therefore not only an academic concern, but a matter of broad social importance. ([View Highlight](https://read.readwise.io/read/01kxxa95nvhbdn5br365zxt5m1)) - Note: The report's own hedge ("not a cure-all," "harmful expressions do exist"). Worth noting: this hedge is present here but gets stripped in the Deseret/Religion Unplugged coverage and again in the Wired Word lesson. The echo chamber worsens downstream — the source is the most careful layer. - Tags: academic-rigor, echo-chamber - Higher religiosity is overwhelmingly associated with lower rates of cigarette smoking, with positive findings outnumbering negative ones by 90:1—the strongest ratio of any health phenomenon examined in this series. This protective effect likely arises from religious teachings that honor the body as sacred; from strong parental and community modeling; from prosocial peer networks; and via faith-based coping practices that offer healthy alternatives to addiction. ([View Highlight](https://read.readwise.io/read/01kxxa8ygk7eq4zze7cg994wbf)) - Note: The 90:1 flagship stat (92 studies, 90 positive, 1 negative). This is a behavior-mediated outcome — the most causally plausible of the 15, since many faiths explicitly forbid tobacco. Contrast with the indirect-pathway domains (cancer, cerebrovascular) that should attenuate most under confounder adjustment. - Tags: academic-rigor - The benefits from religious living identified in this report are dependent upon the ability of individuals and groups to freely practice and live out their specific faith commitments. Legal and cultural norms that protect the right of religious freedom for all and recognize the benefits that religious life and commitments bring to individuals and society as a whole will enhance the likelihood of society experiencing such benefits. ([View Highlight](https://read.readwise.io/read/01kxxa9cft01c4p7j7ehf2ntf5)) - Note: The religious-freedom policy ask — the report's values standpoint, riding on contested effect sizes. A faith-affiliated institute producing pro-religion results and closing with a policy recommendation is exactly where allegiance bias warrants the publication-bias diagnostics the report never runs. - Tags: echo-chamber, academic-rigor