Is the U.S. less religious than previous data suggested, or does current data underestimate U.S. religiosity?
Do previous estimates align with actual religious service attendance, or are U.S. adults showing up less often than they say?
A new Pew Research Center report published Jan. 14 seeks to answer such questions through a deep dive into the center’s polling methodology and data sets.
From 2007 to 2014, Pew’s U.S. religious affiliation and participation surveys were conducted via phone by live callers and published in a report called the Religious Landscape Study.
As responses declined and costs increased for such interviews over the years, Pew shifted its approach in early 2014 to a combination of live phone interviews and online self-guided surveys to which individuals responded at their convenience.
Since 2019, almost all polling is done online as part of Pew’s American Trends Panel (ATP), which replaced the Religious Landscape Study.
In an effort to compare data across time in light of the change in polling methods, Pew analyzed its data and found notable divergences in phone and online responses.
“Survey respondents tend to indicate higher levels of religiosity when answering questions from a live interviewer than when filling out a survey by themselves (either on paper or online),” the report said. “This is because, when representing themselves to another person, some people may (consciously or subconsciously) project a more highly religious image of themselves than when they are filling out a survey alone and unobserved.”
In addition to religious self-perception differences, Pew surmised that the different survey types might appeal to certain people more than others. Thus, results from a phone interview might skew one way in terms of religiosity, while an online survey might lean in another direction.
“Research shows that Americans who do not use the internet at all tend to be older than the general population, and older adults tend to be more religious, on average, than younger adults,” Pew said, concluding that “the ATP may underestimate the country’s religiosity.”
Pew’s in-depth data analysis – the National Public Opinion Reference Survey (NPORS) – was conducted during 2020. It involved reaching out to a representative sample of U.S. adults by offering the survey via both an online and a paper-and-pencil mail-in option.
The NPORS led to two key conclusions:
First, “the ATP may have overstated the share of religious ‘nones’ in the U.S. population.”
The primary cause is likely that the people who do not respond to online survey requests are “presumably more highly religious than the rest of the U.S. public.”
Second, data from the NPORS and the ATP surveys were closely aligned regarding religious service attendance.
This indicates that in live phone interviews people wanted to present themselves as more faithful to their professed faith tradition than they actually are.
It isn’t, necessarily, an intentional choice to mislead interviewers that leads to the skewed results, but from a distinction in the mind of respondents between their identity and their behavior.
As Pew explained, “respondents who do not actually go to religious services every week may nevertheless think of themselves as regular worship attenders, and so describe themselves in the context of a survey interview as weekly religious attenders.”
Adjusting (weighting) the 2020 ATP survey in light of NPORS data, Pew reports that 65% of the U.S. self-identified as Christian (5 points above the unweighted ATP 2020 data), 28% as agnostic, atheist or nothing in particular (4 points below) and 6% as Jewish, Muslim, Buddhist or Hindu (1 point below).
When it comes to religious service attendance, the adjusted ATP 2020 survey finds that 65% of U.S. adults attend services “a few times a year or less” (2 points below the unweighted data) and 27% attend “at least once a week” (2 points above) and 8% “once or twice a month” (1 point above).
The full report is available here.