Making Sense of Polls About Religion

Reports about religion in America often seem to tell different stories. Gallup, for example, within the past year or so, has reported studies headlined Americans' Religious Engagement Holds at Lower Levels, More Americans See Religion Increasing Its Influence in U.S. and Drop in U.S. Religiosity Among Largest in World. A research report I coauthored with Lydia Saad earlier this year, Rise in Young Men's Religiosity Realigns Gender Gaps, prompted questions from journalists and others about how its findings compared with other reports that appeared to show different results.

In many cases, the differences don’t reflect conflicting realities, but rather significant dissimilarities in how researchers measure and report religion in America today. This, in turn, leads to confusion and can create difficulties in understanding what’s really going on.

In this review, I’ve assembled a brief "sophisticated poll watcher's guide" analyzing some of these differences in religious polling. My advice to the consumer ends up focusing on three basic questions:

1. What measure of religiosity is being used?
2. Who exactly is being studied?
3. When was the survey conducted?

Paying attention to these three methodological considerations will hopefully lead the reader to a more informed understanding of American religion.

1. What Measure of Religiosity Is Being Used?

Religion at the individual level is complex and multifaceted. Reflecting this complexity, survey researchers have, over the years, developed a wide range of ways of measuring religiosity.

These generally fall into four buckets:

• Religious identity
• Importance of religion
• Religious beliefs
• Religious behavior

And there are, in turn, a number of ways of measuring religiosity within each of these buckets. (As testimony to that fact, we at Gallup have used more than 25 different measures of religiosity over the decades!) These measures can each tap into different dimensions of religiosity. As a result, the choice of measure can substantially influence a research study’s conclusions.

• For example, a high percentage of Americans believe in God or a spiritual being. This presents a generally positive conclusion about religiosity in the U.S.
• But a fairly low percentage of adults report attending religious services weekly. That figure would present a more negative conclusion about religiosity in America.

Differences also arise when the same concept is measured using different wording or differing response scales.

• Gallup routinely measures religious service attendance as follows: “How often do you usually attend church, synagogue, mosque or temple — every week, almost every week, about once a month, seldom or never?”
• Pew Research, on the other hand, goes through a more complicated sequence, asking about religious service attendance separately for those who have identified as Muslim, Jewish or non-Muslim. The non-Muslim question is: “Aside from weddings and funerals, how often do you attend religious services — more than once a week, once a week, once or twice a month, a few times a year, seldom, or never?”

As a result, while measuring the same underlying variable (religious service attendance), Gallup and Pew may report different estimates in part because they use different question wording and response alternatives.

Even something as seemingly straightforward as calculating the percentage of Americans with no religious identity can be complicated.

• Some respondents do not choose an offered “none” category but say they “don’t know” when asked about their religion. It is unclear if “don’t know” means the person doesn’t have a religious identity or that the person may be uncertain of how to label their religious identity. The researcher has to decide whether or not to include these responses as “nones.”

Researchers ask about the importance of religion in different ways, as well.

• The Public Religion Research Institute (PRRI), for example, asks the following: “Which of the following comes closest to how you think about the place of religion in your life? ‘Religion is the most important thing in my life,’ ‘Religion is one among many important things in my life,’ ‘Religion is not as important as other things in my life,’ or ‘Religion is not important in my life?’”
• Gallup asks the same question somewhat differently: “How important would you say religion is in your own life — very important, fairly important or not very important?”
• As a result, Gallup and PRRI can report different conclusions about the importance of religion among Americans, based not on differences in the underlying population, but instead on the different question wording used by the two firms.

Plus, researchers can group responses to questions in different ways. One researcher could label “frequent church attenders” as the percentage of Americans attending religious services weekly or almost weekly. Another could label frequent church attenders as those attending weekly, almost weekly or monthly — yielding different results.

Bottom line: The sophisticated poll watcher must be careful in determining the exact measure of religiosity used in reported research, particularly when comparing research results across separate reports or time periods.

2. Who Exactly Is Being Studied?

Reports on Americans’ religiosity frequently include the analysis of subgroups. The sophisticated poll watcher has to look carefully to see exactly how the researcher defines these subgroups.

One of these involves age. Researchers in recent years have focused extensively on changes in religiosity among “young people.” But who are these young people?

• The study I coauthored with Lydia Saad focused on those aged 18 to 29.
• But other studies focused on “young people” have defined them as 18 to 24.
• Still other studies focus on Gen Z, defined most often as those born between 1997 and 2012. For surveys conducted in 2025, that group would match an 18- to 28-year-old age group definition (excluding those under age 18, usually not included in surveys). But for surveys conducted this year (2026), the Gen Z age would shift to 29 as the upper limit. And for studies conducted in 2024, the upper limit would be 27. And so on.

The obvious issue here is the need to be conscious of the specific age range included in the researcher’s age groupings, particularly when comparing and contrasting across studies reported by different organizations.

Another important but often confusing subgroup is “Protestants,” a very commonly analyzed group in religious research. But there is no universally accepted way of defining Protestants.

• Researchers often just ask respondents if they identify as Protestants. But that procedure has problems. There is evidence that many Americans (particularly younger Americans) don’t recognize the term “Protestant,” even though they may, in fact, be Protestant by the traditional definition of the term.
• To isolate the group traditionally thought of as Protestants (i.e., non-Catholic Christians), researchers need to probe more deeply than an initial self-identification. At Gallup, we ask respondents who don’t initially identify as Protestants if, among other categories, they identify as other Christians. In many of our analyses, we combine Protestants and other Christians into a broader category representing Christians who are not Catholic (the traditional definition of Protestants).

The poll watcher thus needs to be aware of exactly how the researcher in question isolates the group they label as Protestants.

There is also the issue of grouping within the larger Protestant category. The most traditional definitions use two groups: mainline Protestants and evangelical Protestants. But this is not a simple process. Confusion abounds.

• Some researchers ask respondents to name their denomination (a complicated task in and of itself) and then make decisions on which denominations are “mainline,” which ones are “evangelical” and which ones are some other classification. This task isn’t easy, particularly in an environment in which many large denominations (Presbyterians, Episcopalians, Methodists) are splintered into religiously distinct denominational subgroups.

Categorizing the subgroup of Americans who identify as Jewish in a survey is also more complex than it might seem.

• The traditional way of isolating this group has been to ask for self-identification as Jewish within a religious context: “Are you Catholic, Protestant, Jewish …?” etc.
• But studies by Pew Research show there is an additional group of Americans who don’t initially identify as Jewish in a religious context but will self-identify if asked about their heritage or their affinity. There are Jews by religion, Jews but not by religion, and those with Jewish affinity. The viewpoints and behaviors of those categorized as Jewish Americans can, thus, vary depending on how this group is defined. Recent research by AP-NORC highlights the attitudinal variations in differently defined groups of Jewish Americans.

Another important subgroup encountered in religious survey research are “evangelicals.”

• There is no standard definition of this group. Different research organizations define it in different ways. And the results can be significantly different, accordingly. This is critically important since the term “evangelicals” is widely used in current discourse.

Three years ago, I wrote a long piece called The Thorny Challenge of Defining Evangelicals (still worth reading today!). And thorny it is.

• As I noted in the article, some researchers ask respondents to self-define as “evangelical or born again.” Others simply ask if they are “evangelical.” Research shows big differences in the sizes and composition of groups defined in these two ways.
• Other researchers take a more complex path and ask a series of doctrinal and belief questions, affirmative responses to which define a person as evangelical.
• Still others use respondents’ denominational identity and assume that affiliation with certain denominations makes one an evangelical.

Each of these ways of defining “evangelical” produces a different subgroup of Americans.

This isn’t the place to adjudicate the merits of these various ways of defining evangelical identity. The main point: When sophisticated readers encounter research focused on evangelicals, they must be aware of how the term is defined. This is particularly relevant in comparing findings across research firms and when looking at trends.

More recently, we have seen a rise in discussions of the group defined as “Christian nationalists.”

• Survey researchers can (similar to the situation in terms of evangelicals) simply ask respondents if they identify as “Christian nationalists.” But not all Americans are familiar with the term. (A Pew survey in 2022 found that 54% of Americans knew nothing about the term, even though we can assume some of that 54% would qualify as Christian nationalists based on responses to other questions.)
• PRRI and other research firms define “Christian nationalists” using a series of questions, affirmative answers to which place one in the Christian nationalist bucket. The use of these types of questions to define a group is inherently subjective. Identification of a respondent as a Christian nationalist will thus vary from survey to survey depending on the choice of scale questions.
• Another recent study distinguishes between measures of Christian nationalism as a worldview and measures of Christian nationalism as a personal identity. These approaches identify distinct groups of respondents, potentially leading the reader to different conclusions about who Christian nationalists are, what they believe and how they behave.

Again, the poll consumer who encounters survey research focused on Christian nationalism must take the time to carefully examine the researchers’ definitional criteria.

Bottom line: The sophisticated reader needs to understand how any subgroup used in religious analysis is defined. Two surveys may appear to be measuring the same group, but if they define that group differently, their findings may not be directly comparable. Readers who overlook these definitions can easily draw incorrect conclusions about trends in American religion or differences between groups.

3. When Was the Survey Conducted?

As is the case for most survey measures, religious attitudes, identities and behaviors are shaped by current events, calendar seasons and social context.

• For example, religious behaviors measured just before the Christmas season can vary significantly compared with those measured in the summer.

This makes it mandatory that the consumer note exactly when those included in the research were interviewed.

Dates of interviewing are particularly important in trend analysis. The specific start and end points chosen to measure change can significantly affect an apparent data trend.

• Choosing a start date for a trend when religiosity was at an unusually high or low point can create a narrative of growth or decline, which is inconsistent with trends beginning at other points. Conclusions based on a church attendance measure beginning in the summer of 2020 (after the onset of the COVID-19 pandemic) and ending in the present would show an increase in attendance. This would differ from a trend line beginning with measures taken at the conclusion of the pandemic, or one that began with measures taken in the 1960s. Plus, since religiosity has declined in general over the long term, any measure today will probably be lower than decades ago. But that comparison could obscure a measure that is higher today than it was just a year or two ago (even though still lower over time).
• Grouping data by arbitrary intervals (e.g., five-year vs. 10-year blocks) can flatten out critical cyclical patterns. For example, comparing the last 10 years of religious service attendance with religious service attendance for 10-year periods prior to that would miss the nuances of any shifts in behavior within the 10-year periods. On the other hand, measuring religious service attendance on a six-month to six-month basis could introduce noise and false indications of change.

Bottom line: Survey dates are an important part of the interpretation of findings on religion. Without accounting for timing, the reader can risk mistaking temporary reactions or seasonal fluctuations for genuine changes in religiosity.

Key Point Summary

Survey research provides very valuable insights into religion, a critically important aspect of American society and culture. But survey research on religion can be complicated and often nuanced because religiosity is a complex, multidimensional concept. While there are many methodological points that need attention when analyzing survey research reports, sophisticated poll watchers can, in particular, look beyond headline findings and pay close attention to three essential aspects of any reported research focused on religion: (1) the specific measure of religiosity being used, (2) exactly how religious and other subgroups are defined, and (3) when the interviews were conducted.

Stay up to date with the latest insights by following @Gallup on X and on Instagram.