AI note
Gender, diversity, and movement leadership: what the “18%” finding actually implies
Executive summary
The candid internal statistic is not a footnote; it is a credibility variable. If Movemental’s curated graph is built from network-visible leaders and verification flows through institutional artifacts (publishing, conference stages, organizational titles), then an 82% / 18% gender skew is less likely to be a random sampling accident and more likely to be a structural mirror of who has historically been credited, funded, and invited.
That does not automatically mean the scoring model is “wrong” in a narrow statistical sense. Public benchmarks in U.S. Christianity are themselves uneven: women remain underrepresented in many senior pastoral and top denominational roles, and the Evangelical Christian Publishers Association baseline survey (limited participation, wide margin of error) still shows men disproportionately occupying executive/senior manager roles in religion publishing. In other words, 18% can be “in line with” parts of the landscape while still being misaligned with Movemental’s brand promise if the promise is epistemic breadth and movement authenticity rather than reproducing existing visibility curves.
The honest product question is dual:
- Pipeline: Are there fewer women in the kinds of roles the gates measure?
- Selection / legibility bias: Are women present in movement leadership but less legible to criteria tuned on male-typical pathways (sole authorship, headline titles, “visible impact” proxies that track money and media)?
Movemental should assume both are nonzero until proven otherwise. The organizational sociology literature on gendered organizations (Acker) and extensive work on credit, citation, and invitation gaps in knowledge fields gives strong reasons to treat “neutral-looking criteria” as disparate-impact suspects until stratified.
Network science adds a second layer of risk: homophily and preferential attachment mean early composition becomes self-reinforcing, especially under a cap (the prompt notes a 100-leader cap). A small curated system is not “the market”; it is a designed public whose demographics will be read as intentional.
1. Read the headline as a signal about credibility, not virtue
Movemental’s thesis depends on curated credibility and bounded network verification. In that frame, composition is not cosmetic. Users and partners will reasonably ask: Whose discernment trained the system? Whose elders and peers count as “verification”? Which movement forms count as “impact”?
If the endorsement graph is skewed male, the platform risks two opposite failures at once:
- Epistemic: Missing non-redundant information about where the Spirit is forming movements (Hong & Page’s formal results on diverse toolkits are not a sociology transplant, but they gesture at why monochrome heuristics underperform in complex search problems).
- Political / reputational: Being interpreted as “the founder’s network, formalized”—which is often homogeneous along gender lines (prompt Q15).
Neither failure requires bad intent. Intent is not the variable. Structure is.
2. Is 18% “expected” given Christianity’s public statistics?
2.1 Congregations and top roles
Secondary summaries of National Congregations Study / Hartford-type statistics frequently land near low double digits for congregations with a woman as senior or sole ordained leader, with large denominational variance (higher in many mainline settings; lower in many conservative evangelical settings). Pew’s older congregational snapshot (2012 wave, as reported in 2014) included an ~11% figure for congregations led by women—useful historically, but do not treat it as current without revisiting the latest wave.
Pew’s 2016 overview of top leadership across major bodies underscores a harsher pattern at the apex: even where ordination is allowed, women have rarely held the single top seat. That matters if Movemental’s pipeline implicitly treats denominational-scale visibility as a trust shortcut.
2.2 Publishing and “movement thought leadership”
Christian publishing is both a visibility engine and a gatekeeping stack. The ECPA baseline survey (2022; reported by Publishers Weekly in 2023) is the best industry-shaped comparator this pass located. Among participating publishers, religion publishing skewed more male than general publishing at the workforce level, and executive/senior management roles skewed heavily male relative to women in the same tiers.
Critically, the survey’s participation rate was low (~26% of member publishers) with a reported ±20% margin of error. So: directionally informative, not precision-grounded.
Trade press also documents a market expansion of women-authored religious books—so authorial voice can be rising even while editorial power remains skewed. That split is exactly where selection bias hides: more books by women does not automatically mean more women acquiring, positioning, and introducing peers into elite networks.
Bottom line: Depending on how Movemental defines “movement leader,” 18% is not automatically an outlier relative to parts of U.S. Protestant leadership and evangelical publishing power structures. The more uncomfortable possibility is the opposite: the number is “normal” for a biased system—which is worse for a product that claims to curate better than the default.
3. Pipeline vs selection bias: how to decompose without self-exonerating
3.1 Pipeline (supply)
There are fewer women in certain publicly scored roles: senior pastorates in some traditions, C-suite publishing leadership, marquee conference headliners. If Movemental’s gates map those roles tightly, the pipeline supplies skewed inputs.
3.2 Selection and legibility (rules + measurement)
Even with a fair-sounding rubric, criteria can inherit bias when:
- “Visible movement impact” tracks funding, staff size, media, and stage time—all gender-skewed in many ecosystems.
- “Published work” tracks imprints and marketing muscle, not only insight.
- mDNA foundation is measured through courses/books/networks that historically platformed men as default teachers.
Joan Acker’s theory of gendered organizations is the clean conceptual frame: organizations present as neutral while embedding a male-coded “ideal worker” template. Movemental is not a traditional firm, but curation is an organizational technology with the same hazard: what counts as “serious” is culturally trained.
3.3 Practical decomposition Movemental can run internally
Without theatrics, run three audits in parallel:
- Longlist vs shortlist: Compare gender composition at ingest vs post-gate to see where the cliff appears.
- Tradition-stratified rates: Mainline / evangelical / charismatic / global networks may differ sharply; a single 18% can conceal good and bad stories.
- Pathway typology: Classify leaders by how they became legible (publisher-backed author vs practitioner founder vs scholar vs activist vs unsalaried movement catalyst). If women cluster in categories the rubric underweights, that is selection bias, not “absence.”
4. Network effects: why a cap makes composition strategic, not decorative
In small curated systems, early roster choices shape who recommends whom, who appears in “people like you” loops, and whose co-signatures become training data for what “movement credibility” looks like.
Homophily literature and preferential attachment models predict self-reinforcement: the graph’s skew becomes the prior for the next invitation cycle. That is dangerous under a 100-leader cap because exclusion is structurally loud—absence is interpreted as judgment.
Hong & Page’s diversity in problem-solving result is often invoked too casually in DEI conversations, but a narrower, safer use is legitimate: if Movemental’s verification problem is search under uncertainty—who is real, who is performing expertise—then redundant priors (everyone sharing the same blind spots) are a security and quality liability, not only a fairness issue.
5. Comparable platforms: analogies, not benchmarks
Public reporting on TED speaker demographics shows long-run improvement in women’s representation on TED.com alongside persistent gaps in other dimensions; older reporting cited very low shares of women among shortlisted TEDx talks recommended “upward.” Treat this as a warning about nomination funnels, not as a target quota.
Movemental should not claim precise parity with TED, Masterclass, YPO, or GLN without primary desk research; those organizations do not publish a tidy, comparable “verified leader graph gender %.” The honest comms line is narrower: curated networks everywhere inherit founder and gatekeeper demographics unless they engineer against it.
6. Theological notes (keep them functional, not preachy)
APEST / mDNA, read plainly, does not entail gender hierarchy in the text of the frameworks themselves. The platform risk is cultural, not a syllogism: imagery, exemplar lists, and “apostolic” tropes often track masculine archetypes in practice, which shapes who feels named and who gets introduced.
Egalitarian movement leaders (the prompt names Felicity Dale, Jo Saxton, Tina Teng-Henson as examples) are useful positive controls: if the scoring pipeline struggles to surface leaders like them at proportionate strength, the issue is likely legibility and graph seeding, not theology on paper.
7. What Movemental can do (without diluting standards)
Principle: Diversity work here is not lowering the bar. It is widening the aperture for legitimate bar-clearers whose impact has been under-credited by default institutions.
Concrete moves:
- Publish methodology for how gender and diversity metrics are tracked internally (privacy-respecting), with a commitment to annual reporting even pre-launch.
- Separate “exemplar graph” from “verification graph” if needed: keep strict gates, but actively recruit shadow validators and cohort mentors outside the founder’s first-degree network.
- Rebuild the longlist with explicit pathway quotas (not outcome quotas): e.g. minimum depth searches in non-publisher-led movements, majority-world networks, women-led church planting communities, and practitioner founders without traditional imprints.
- Criteria audit for disparate impact: For each gate (
research-016,research-017, publishing/affiliation proxies), document which populations are statistically less likely to satisfy the proxy and add alternative evidence types (e.g. verified cohort outcomes, audited field metrics, peer review by geographically diverse councils). - Cap-aware sequencing: If the first 100 is unavoidably skewed, say so plainly and publish a public roadmap for rebalancing as the graph grows—silence reads as endorsement.
8. Risks of launching with a heavily gendered network
- Brand: “Curated credibility” becomes “credibility for a certain kind of leader.”
- Product: Verification misses real movement at the margins; the graph is gamed by those who already know how to look “legible.”
- Ecosystem: Women leaders (and men who care about this) may defer participation if they infer the network is not built to see their impact.
- Narrative vulnerability: Competitors or critics can reduce Movemental to graph homophily with a slick UI.
9. Closing stance
The 18% figure is best treated as a diagnostic, not a moral verdict on any individual leader already in the graph. It says: the bounded network is currently shaped like the historical platforms that conferred visibility. If Movemental wants to be more than that, the response is not defensive benchmarking (“we’re similar to X”) but structural engineering: different longlists, alternative legibility paths, transparent metrics, and cap-aware sequencing so the credibility signal does not quietly become gendered by default.
The companion raw file lists primary links, caveats, and follow-up empirical tasks (docs/build/research/raw/19-gender-diversity-raw.md).