What methods and surveys (e.g., Pew, PRRI, ARDA) report denominational change and how reliable are 2025 estimates?
This fact-check may be outdated. Consider refreshing it to get the most current information.
Executive summary
Three major types of sources—large national surveys (Pew Research Center’s Religious Landscape Study and American Trends Panel), specialized national polls (PRRI’s American Values and landscape studies), and archival/administrative compilations (ARDA and decadal denominational censuses)—are the foundation for recent estimates of denominational change; each brings different sampling frames, question wordings and aggregation rules that materially affect 2025 estimates and how confidently those numbers can be interpreted [1] [2] [3] [4]. The 2023–24 Pew Religious Landscape Study (RLS) and related 2025 switching analyses offer the most comprehensive, high‑precision portrait to date but carry standard survey caveats—question wording, coding choices and modest margins of error for subgroups—while PRRI and ARDA provide complementary perspectives that emphasize different definitions, historical baselines and local membership data that can lead to divergent readings of “decline” or “stabilization” [5] [6] [7] [8] [9].
1. What the big national surveys measure and how they do it
Pew’s Religious Landscape Study (RLS) surveys more than 35,000 U.S. adults across all 50 states and combines a multiwave American Trends Panel (ATP) approach with address‑based sampling to produce household‑level representativeness; in 2023–24 it asked standard “What is your present religion?” and childhood religion items and then used branching questions to code Protestants into evangelical, mainline or historically Black Protestant traditions based on denominational affiliation [1] [6] [2]. The ATP-based switching analyses explicitly calculate “religious switching” from paired questions about current and childhood religion and report sampling details and margins of error (for example, ARDA notes a ±2.51 percentage point margin on weighted ATP estimates and Pew reports overall RLS margins as small as ±0.8 points for large samples) [9] [10] [2].
2. PRRI and complementary national snapshots: definitions matter
PRRI’s American Values and religious‑change reports use large national samples too (examples cited from 2016, 2022 and 2023 work) and emphasize “religious churning,” retention and local state-level shifts; PRRI sometimes defines groups differently (for instance, defining evangelicals by denomination in some reports) and highlights trends such as one in four Americans reporting prior affiliation changes—an analytic framing that can accentuate mobility compared with Pew’s tradition‑based presentation [3] [8] [11].
3. Administrative and archival sources: ARDA and denominational censuses
ARDA aggregates hundreds of surveys and congregational membership reports and links to decadal denominational censuses (the Association of Statisticians of American Religious Bodies) to map organizational membership trends; these administrative counts can capture institutional metrics (membership rolls, congregational counts) that surveys miss, but they also suffer from uneven coverage across traditions—especially non‑centralized groups like non‑denominational churches, Muslim and some Jewish communities—so their denominator and coverage choices shape perceived change [4] [12].
4. Why 2025 estimates can differ: sampling, wording and coding
Differences in estimated denominational change stem from three technical choices: sampling frame and mode (panel ATP vs single‑wave national surveys vs telephone/address frames), question wording and branching that assign Protestant respondents to traditions, and post‑survey coding rules that group denominations into families (Pew’s rule that Southern Baptists are classified as evangelical and United Methodists as mainline regardless of self‑described “born‑again” status is a clear example) [2] [6] [1].
5. Reliability: where the 2025 numbers are strong and where they’re fragile
Strengths include very large sample sizes in Pew’s RLS that reduce overall margin of error and allow state-level estimates (RLS n≈36,908 and overall margins cited as low as ±0.8 points) and transparent methodology documents; weaknesses include inevitable biases from question wording, nonresponse and the complexity of coding mixed or non‑specific Protestant identities, plus wider margins for subgroup or denominational family estimates that make small year‑to‑year shifts uncertain [1] [10] [2] [9]. Reporters and analysts should treat broad national patterns (large declines or stabilization signals) as more robust than small percentage swings for particular denominations.
6. Alternative readings, incentives and how to read future estimates
Different research organizations emphasize different narratives—Pew framing a possible leveling off of Christian decline (based on its RLS) while PRRI underscores ongoing “religious churning” and state‑level shifts—partly reflecting methodological choices and partly reflecting institutional missions and audiences; researchers should triangulate RLS, PRRI and ARDA administrative trends, pay attention to question wording appendices and margins for subgroups, and avoid overinterpreting small changes until replicated across datasets [5] [8] [4].