by Caroline Soler
White evangelicals have emerged as a dominant force in conservative politics and an increasingly large share of Trump’s base, which means accurately measuring this group is more critical than ever. Yet, researchers still lack consensus on the fundamental question of how to define evangelicals. Two main competing approaches exist: self-identification and a denominational approach. This choice is not trivial, as these approaches can yield different pictures of what evangelicalism looks like in America, potentially affecting research conclusions about this group.
The denominational approach categorizes evangelicals based on their church affiliations, racial background, and church attendance patterns. The approach was developed in 2000 using the RELTRAD method, which was then refined by Ryan Burge in 2016 after he discovered that the original strategy was undercounting some evangelical groups. Pew went on to improve this strategy even further, creating a classification scheme that includes slightly more denominations than Burge’s strategy. However, each of these methods relies on classifying individuals as evangelical based on the churches they claim to be members of.
By contrast, the self-identification approach simply asks respondents directly if they identify as born-again and/or evangelical. Through this approach, one need not be a member of a church at all to be classified as evangelical.1
While most surveys ask only the self-identification question for efficiency, the Cooperative Election Study (CES) includes detailed denominational measures alongside the self-identification question, allowing for a direct comparison of both approaches.
Using the cumulative CES file from 2008 to 2024, as well as the 2020-2024 panel survey that studied the same respondents across three survey waves, I can examine how these approaches differ in their validity, providing insight into which approach identifies evangelicals more accurately. I focus on the Pew methodology for the denominational approach, as it casts the widest net for identifying evangelicals across denominations.
Self-identification consistently identifies more evangelicals than the denominational approach: about one-third of Americans compared to slightly less than one-quarter. The gap between these approaches is shockingly stable from year to year, suggesting that this difference is systematic rather than random or contextual. Examining the overlap between these categorizations reveals that many Americans are captured by only one approach — 75% of denominational evangelicals self-identify as evangelical. In contrast, only 48% of self-identified evangelicals fit into the denominational group. This limited overlap suggests the approaches may be measuring fundamentally different dimensions of evangelical identity.
So, is one of these approaches capturing “true” evangelicalism? I interrogate this question across multiple measures of validity: face validity (whether each approach appears to measure what it claims to measure), construct validity (whether each approach predicts what evangelicals should believe and do), and reliability (whether each approach produces consistent results over time).
Face validity
To assess face validity, I examine the extent to which each variable measures what it is supposed to measure on its face — in this case, evangelicalism. However, evangelicalism itself is not clearly defined — not only have researchers not come to a consensus on what this identity looks like in data, but evangelicals themselves lack clear definitional boundaries. According to the National Association of Evangelicals, evangelicals are those who “take the Bible seriously and believe in Jesus Christ as Savior and Lord” — a description that appears to encompass many Christians across denominational lines, and without any meaningful distinction. Most scholars agree that evangelicals are Protestants who are highly committed to their religion. Thus, to assess face validity, I will examine the religious composition of these two evangelical groups and the varying levels of religiosity within these groups.
The denominational group is entirely Protestant by construction, as I restricted evangelical classification to those who self-identified as Protestant. The self-identification group is fairly religiously diverse; only 68% identify as Protestant, 12% “something else”, 9% are Catholic, and interestingly, 9% identify as “Nothing in particular.” If we accept the standard assumption that evangelicals should predominantly be Protestant, then the denominational approach appears to have much better face validity. Yet, it is possible that the denominational definition of evangelicalism may be wrong.
Given the rise of movements such as “evangelical Catholicism,” these Catholics and other non-Protestant respondents may reflect a broader evangelical identity that transcends conventional Protestant classifications.
This possibility is reinforced by examining the racial composition of each group. While denominational evangelicals are 76% white and 9% Black, self-identified evangelicals are 69% white and 18% Black. This difference reflects varying rates of evangelical identification across racial groups: 45% of Black Americans identify as evangelical compared to just 27% of white Americans. This suggests that the denominational approach may be missing key subgroups of the evangelical population, potentially undermining its face validity.
Furthermore, evangelicals are characterized by their profound religious commitment, including focus on Jesus Christ, bible obedience, and active evangelism. Although the CES does not explicitly ask about these behaviors, it does include multiple questions related to the strength of one’s religious identity, including church attendance and the importance of one’s religion.
Comparing each of these measures across the self-identified and denominational groups reveals similar levels of religiosity. In terms of church attendance, both groups show nearly identical patterns of church attendance, with 51% attending once a week or more, compared to 50% of denominational evangelicals — a negligible difference. Similarly, 75% of self-identified evangelicals say that religion is “Very important”, in comparison to 70% of denominational evangelicals. Hence, it appears that self-identified evangelicals may be more committed to their faith than those identified denominationally.
Self-identified evangelicals have actively affirmed their evangelical identity through the “born-again” question, demonstrating that this religious identity is central to how they perceive themselves. In contrast, the denominational group includes many who explicitly rejected the evangelical label when asked directly. These individuals may attend evangelical churches for reasons beyond religious commitment, such as family tradition or community ties, which may be why they demonstrate a lower level of spiritual commitment than the self-identified group. Despite these minor differences, both groups are highly religious compared to the rest of America — only 40% of Americans say religion is important. Thus, both measures seem to have sufficient face validity.
Construct validity
Construct validity assesses whether each measure correlates with theoretically related concepts — in this case, whether evangelical identification predicts the conservative political attitudes typically associated with evangelicals. Voting behavior and abortion preferences are highly similar between these groups. For abortion, I look at the question, “Always allow a woman to obtain an abortion as a matter of choice,” as it directly contrasts the traditional evangelical views on the sanctity of life.
In terms of vote choice, self-identified evangelicals are slightly more Democratic than denominational evangelicals. Both groups have followed the same trend since 2008, becoming increasingly conservative. Self-identified evangelicals, for instance, went from voting 38% Democratic in 2008 to 26% in 2024. Denominational evangelicals shifted from voting 29% Democratic in 2008 to 21% in 2024. This downward trend is as expected, given that evangelicals have garnered lots of attention for their strong support of Republicans. Today, across either measure, only around one-fourth of evangelicals are voting for Democrats in the House. Americans in general have followed similar patterns, shifting slightly more Republican since 2008, from 53% Democratic to 48% in 2024.
Denominational evangelicals showed 34% support for always allowing abortion in 2016, 29% in 2020, and 32% in 2024. Self-identified evangelicals followed nearly identical trends at 37%, 34%, and 36% respectively. The remarkably similar patterns — both the timing and magnitude of the shifts — demonstrate that both measures are capturing populations with essentially the same underlying attitudes. The consistent pattern, where self-ID evangelicals are slightly more liberal, mirrors what we observe in voting behavior; however, the differences are minimal enough to confirm that both measures identify similar evangelical communities. The liberalization between 2020 and 2024 on abortion likely reflects substantive attitude shifts on abortion following the Dobbs decision and subsequent restrictions on women’s healthcare access.
Evangelicals across both approaches are significantly more conservative than the general American population. Furthermore, both groups exhibit the same patterns year after year, confirming that both measures possess high construct validity.
Reliability
Finally, I examine the reliability of these approaches by assessing the consistency of each approach over time, ensuring that they yield similar results across multiple cycles. If one of these approaches has higher reliability, scholars would likely favor it as the more effective way to classify evangelicals. This can be done using the 2020 to 2024 CES recontact survey, which studied the same 6,175 respondents over three election cycles, allowing us to evaluate reliability within the same population.
The consistency of these measures between 2020 and 2024 is illustrated in the figure below, which shows how respondents maintained or changed their evangelical classification across the two cycles. Among self-identified evangelicals, 23% remained evangelical in both waves, and 71% were consistently non-evangelical, leaving only 5% who switched between categories. The denominational approach shows similar but slightly less stable patterns: 17% remained evangelical across both waves, and 74% stayed consistently non-evangelical, with 9% switching categories.
To measure this reliability statistically, I look into the intra-class correlation coefficient (ICC) of each variable across the three survey waves. The ICC measures how consistently respondents are classified by each approach temporally — essentially, if someone is identified as evangelical in 2020, what is the probability that they’ll be identified as evangelical again in 2022 and 2024?
In survey research, ICCs above .75 are generally considered acceptable for making meaningful comparisons over time. The self-identified approach has an ICC of 0.88, while the denominational approach reaches 0.78 — both exceeding this benchmark. These strong statistical measures reinforce what the consistency patterns already suggested: evangelical identity, however measured, is stable. With over 90% of respondents maintaining consistent classifications across survey waves and ICC scores well above acceptable thresholds, both approaches demonstrate high reliability. Either approach provides sufficient reliability for tracking evangelical identification and attitudes.
Conclusion
Both the denominational and self-identification approaches demonstrate sufficient validity and reliability for measuring evangelicals, though they capture meaningfully different populations. The denominational approach identifies a more Protestant, racially homogeneous group that aligns with traditional assumptions about evangelical identity. The self-identification approach captures a more diverse population, including significant numbers of Black evangelicals and some non-Protestants, while demonstrating slightly higher religious commitment and reliability.
Rather than declaring one approach superior, researchers should choose based on their specific goals. For those focused on traditional white Protestant evangelicalism or requiring precise denominational classification, the denominational approach remains valuable. For research prioritizing evangelical identity as individuals understand it, seeking greater racial diversity, or operating under resource constraints, self-identification offers a valid alternative. The parallel behavior across both groups suggests that either approach will capture populations that respond to events in characteristically evangelical ways.
- However, even within the self-identification approach, there is significant debate over the wording of the question. The CES uses the double-barrelled approach from Pew, asking respondents if they would describe themselves “as a ‘born-again’ or evangelical Christian, or not?”, whereas other surveys, like the GSS, ask respondents if they “have been ‘born again’ or have had a ‘born again’ experience — that is, a turning point in your life when you committed yourself to Christ?” Obviously, these questions are not probing the same response. Gallup found that 27% of Americans identify as born-again, but only 12% identify as evangelical, so what exactly are we measuring when we study and discuss evangelicals? ↩︎