HOW THE INDEX IS BUILT

RMBI Methodology.

RMBI combines seven dimensions, standardizes unlike measures, and shows the path to the final index. The evidence base includes 13 research study/data families: 6 core scoring studies and 7 supporting research families, with U.S. Religion Census administrative data used separately for Institutional Reach.

7 DIMENSIONS13 RESEARCH FAMILIES0–100 STANDARDRAW EVIDENCE AUDIT

THE FRAMEWORK

Seven Different Views of Member Benefit

Seven dimensions capture personal, family, community, institutional, charitable, structural and engagement outcomes.

VARIABLES USED

Volunteer Service

3 variables

Pew 2014

Volunteered in past seven days

Pew 2023-24

Volunteered in past 12 months

NSYR Wave 1

Verified study composite

VARIABLES USED

Life / Youth Outcomes

15 variables

Pew 2014

Very happy
Very or pretty happy
Very satisfied with family life
Satisfied with family life
Very satisfied with health
Satisfied with health
College graduate or higher
Family income $100,000 or more

Pew 2023-24

Very happy
Very or pretty happy
Excellent or very good health
Excellent or very good family life
College graduate or higher
Family income $100,000 or more

NSYR Wave 1

Verified study composite

VARIABLES USED

Family Stability

7 variables

Pew 2014

Currently married
Married or living with partner
Same broad religion as spouse/partner

Pew 2023-24

Currently married
Married or living with partner
Same religion as spouse or partner

NSYR Wave 1

Verified study composite

VARIABLES USED

Member Resources

6 domains

Member-access resource domains

Basic Needs / Welfare
Education
Employment / Self-Reliance
Family / Counseling
Youth / Children
Health / Recovery

VARIABLES USED

Charitable Giving

3 variables

Philanthropy Panel Study 2001

Religious giving dollars · standardized
Total charitable giving dollars · standardized
Religious giving as share of income · standardized

2023 PSID · Supporting corroboration

Used as corroborating evidence for the Religiously Unaffiliated charitable-giving treatment

VARIABLES USED

Institutional Reach

4 variables

2020 U.S. Religion Census / RCMS

National adherent rate
Population access
Geographic breadth
Congregational availability

VARIABLES USED

Religious Engagement

16 variables

Pew 2014

Attends weekly or more
Attends monthly or more
Religion is very important
Member of local house of worship
Prays at least daily
Reads scripture at least weekly
Weekly prayer/scripture/religious education group

Pew 2023-24

Attends religious services weekly or more
Member of a house of worship
Religion is very important
Prays at least daily
Reads scripture at least weekly
Participates weekly in prayer, scripture study or religious education group

CES 2024

Religion is very important
Weekly+ attendance
Daily+ prayer

TRIANGULATION

What Strong and Weak Triangulation Look Like

Triangulation asks whether independent studies point toward the same underlying insight. The more independent evidence that points in the same direction, the more confidence we have that the directional story is valid.

STUDY 1Independent evidence source✓

STUDY 2Independent evidence source✓

STUDY 3Independent evidence source✓

STUDY 4Independent evidence source✓

STUDY 5Independent evidence source✓

DESIRED INSIGHT / OUTCOMEHappiness

Confidence increases when independent evidence points toward the same outcome.

HOW TO READ IT

TRIANGULATION STRENGTHSTRONG4 of 5 studies point in the same direction

Four independent studies support the same directional insight. One study differs, but the majority of evidence still points toward the same conclusion.

ANIMATION EXPLANATIONStart with weak evidence

When most independent studies do not point toward the same insight, confidence is low.

FROM DATA TO INDEX

The Calculation Pipeline

Different studies report percents, dollars, rates, survey responses and institutional measures. RMBI converts those inputs into comparable components before building the final index.

Raw Evidence

Start with the observed survey, census, financial or institutional measure.

Standardized Variable

Put unlike metrics onto a comparable 0–100 scale using the rule appropriate to that variable.

Study / Source Composite

Combine qualifying variables within the study so longer questionnaires do not automatically dominate.

Dimension Score

Combine qualifying source evidence into one score for the research dimension.

Overall RMBI

Average the available qualifying dimension scores. N/A dimensions are excluded, not scored as zero.

ROBUSTNESS & VALIDATION

We Didn't Just Calculate the Index. We Tried to Break It.

RMBI was stress-tested to determine whether the results depend on a particular weighting choice, one evidence source, missing-data treatment, normalization choice, measurement assumptions, or website transcription. These tests evaluate robustness and auditability. They are not claims of causal proof or predictive accuracy.

13 RESEARCH STUDY / DATA FAMILIES 6 CORE SCORING STUDIES 7 SUPPORTING RESEARCH FAMILIES U.S. RELIGION CENSUS · ADMINISTRATIVE DATA

WEIGHT SENSITIVITY

100,000 Alternative Weight Scenarios

Positive weights across the seven dimensions are repeatedly changed and renormalized to test whether the overall ranking depends on the baseline equal-weight decision.

ROBUSTNESS TEST

Interpretation: 99.88% is ranking stability under alternative dimension weights. It is not a claim of 99.88% predictive accuracy and it does not measure sampling error in the underlying studies.

STANDARDIZATION

One Scale. Many Measures.

Choose a method to see how RMBI makes evidence comparable.

DEFAULT METHOD

Min-Max Normalization

0–100 SCALE

Comparable percentages and surveys preserve their relative position on a 0–100 scale.

RAW INPUT42 · 57 · 72

→

TRANSFORM(x − min) ÷ (max − min) × 100

→

RMBI OUTPUT

050100

WHY THIS METHOD?Preserves ranking while making percentages comparable.

A CRITICAL DISTINCTION

0.0 Is Not the Same as N/A

RMBI separates a bottom-of-scale observed result from a group with no qualifying evidence.

0.0

Observed, but lowest in the comparison range

A 0.0 score does not mean 0%. It means an observed value standardized to the bottom of the comparison range. Hovering reveals the raw result.

N/A

No qualifying score

If no defensible observation exists, RMBI uses N/A. Missing evidence is never converted to zero and is excluded from the denominator.

MEMBER RESOURCES

Availability Alone Is Not Enough

Member Resources measures tangible support an ordinary member can reasonably access, separate from religious behavior.

Member Resources = Access × Affordability

A resource receives credit only when meaningful member access is demonstrated. Free or subsidized services receive more credit than market-rate services. Institutional ownership alone is not enough.

The six equal domains are Basic Needs / Welfare, Education, Employment / Self-Reliance, Family / Counseling, Youth / Children, and Health / Recovery.

1.00 · FREE / MEANS-TESTED FREE

Full affordability multiplier.

.90 · HEAVILY SUBSIDIZED

Nominal fee or substantial institutional subsidy.

.80 · MEMBER DISCOUNT

Meaningful affiliation-based economic advantage.

.70 · NORMAL FEE + ROUTINE AID

Normal charge with documented routine aid or waiver availability.

.60 · MARKET RATE

Resource exists but membership provides little price advantage.

0 · NO DEMONSTRATED MEMBER BENEFIT

Institutional existence alone is not sufficient.

WEIGHTING

Preventing One Dataset From Dominating

RMBI separates variable-level from study-level evidence so longer surveys do not gain extra weight.

Within a study

Qualifying variables are combined into a source composite so questionnaire length does not determine study weight.

Variable 1equal contribution

Variable 2equal contribution

Variable 3equal contribution

Study Compositeone source score

Across qualifying sources

Dimension scores combine qualifying source composites, using equal weighting where multiple sources qualify. The overall score is the equal-weight mean of available qualifying dimensions.

Study / Source Aqualifying composite

Study / Source Bqualifying composite

Study / Source Cqualifying composite

Dimension Scorecombined evidence

EVIDENCE QUALITY

Not All Evidence Is Treated the Same

Evidence grades show support strength.

Direct / primary analyzed evidence

Direct RMBI analysis or an aligned validated composite.

Strong documented evidence

Published or primary evidence coded through a transparent rule.

Documented proxy / aggregation

Broad-family or denomination evidence with an explicit crosswalk.

Not used in the final model

Unsupported imputations are excluded and treated as N/A.

AUDIT TRAIL

Transparent score.

Users can trace each score backward through the complete evidence chain.

Overall RMBI→Dimension Score→Study Composite→Standardized Variable→Raw Evidence

On the Details page, variable scores can be compared and hovered to reveal raw evidence.

BOUNDARIES

What RMBI Does and Does Not Claim

Comparative, not causal

Observed differences may reflect demographic, social, economic or other factors.

Triangulation increases confidence

Converging evidence strengthens an interpretation but does not prove causation.

Not every dimension needs multiple studies

Where one authoritative source is appropriate, RMBI uses it rather than adding weaker evidence.

Scores are relative

A 0–100 RMBI score is comparative, not a moral, theological or percentage grade.

Contrary evidence remains eligible

Mixed or contradictory qualifying evidence remains in the model.

N/A is not negative

N/A means no qualifying score, not poor performance.