February 1, 2008

U.S. Religious Landscape Survey: Religious Affiliation

Introduction

From the beginning of the Colonial period, religion has been a major factor in shaping the identity and values of the American people. Despite predictions that the United States would follow Europe’s path toward widespread secularization, the U.S. population remains highly religious in its beliefs and practices, and religion continues to play a prominent role in American public life.

In recent decades, much high-quality research has been done on the religious makeup of the
United States and on the way religion relates to politics and public life. Nevertheless, there are still major gaps in our knowledge of the American religious landscape. For instance, estimates of the size of religious communities in the U.S. – especially the smaller groups – are often contested, basic information on the religious beliefs and practices of many groups is lacking and there is little solid data on the demographic characteristics of many of America’s newer faiths. The increasing diversity of the American religious landscape, the remarkable dynamism of its faith communities and the pervasive presence of religion in the American public square all serve to underscore the pressing need for up-to-date, reliable information on these and other questions.

Building on our own work as well as others’ previous research, the Pew Forum on Religion &
Public Life has conducted a path-breaking survey on American religion that seeks to address many of these important issues. The U.S. Religious Landscape Survey includes reliable estimates of the size of religious groups in the United States as well as detailed information on their demographic characteristics, religious beliefs and practices, and basic social and political values. Based on interviews with a representative sample of more than 35,000 Americans, this study will serve as the baseline for similar large-scale surveys the Forum plans to conduct periodically.

There are good reasons for the relative absence of authoritative information on American religion. Most importantly, the U.S. Census Bureau has been prevented by law or administrative rules since the late 1950s from collecting even basic information on religious affiliation from the public in its decennial census or other demographic surveys (see Appendix 3 (PDF)), thus excluding religion from America’s largest and most authoritative survey instrument. Even when the Census Bureau collected such data, however, it was of very limited value for shedding light on Americans’ religious beliefs or practices. The absence of such official statistics is not unique to the U.S. (only about half of the world’s countries include questions on religion in their censuses), but the omission is particularly significant in the U.S. because it is among the most religiously dynamic and diverse
countries in the world.

Two types of studies have attempted to fill the void created by the absence of census data on
American religion. One approach has been to aggregate statistics collected by individual religious bodies. Good examples of this include the Yearbook of American and Canadian Churches, produced by the National Council of Churches, and Religious Congregations & Membership in the United States, produced by the Glenmary Research Center. Some efforts – including the World Christian Database – attempt to merge these denominational statistics with other data to produce broader estimates of the size of religious groups.

These collections are quite valuable, but they have their shortcomings. For instance, religious groups and denominations use different methods for counting members, some do not share their counts publicly and others do not collect membership statistics at all. Moreover, relatively few groups collect information on the religious beliefs and practices of their members. In addition, there is a sizeable number of Americans who are not affiliated with any particular religious group but who nonetheless have religious beliefs or engage in a variety of religious practices.

A second approach to collecting data on American religion has been to measure religion through surveys rather than head counts. One kind of survey – such as the General Social Surveys, conducted since 1972 by the National Opinion Research Center at the University of Chicago, and numerous surveys conducted by The Gallup Organization – involves administering fairly lengthy interviews to a small number of Americans on a wide range of topics, including religion. Other surveys – such as the 2001 American Religious Identification Survey, conducted by the Graduate Center of the City University of New York – go in the other direction; they ask a few questions about religion to a large sample of Americans.

These surveys are also a very valuable source of information on religion in the U.S., but they too have their limitations. On the one hand, most in-depth surveys interview relatively few people, making it difficult to analyze smaller religious groups. Large-sample surveys, on the other hand, usually ask relatively few questions on religion and thus do not delve deeply into the particular beliefs and practices of religious groups.

The U.S. Religious Landscape Survey builds on the foundation of these previous studies by
combining the advantages of more in-depth surveys based on small samples with the strengths
of shorter surveys based on large samples. We believe this combination of multiple questions on religion and large sample size fills an important niche by providing a new basis for enumerating and understanding the country’s increasingly diverse religious landscape.

No matter how rigorous, however, all surveys have their limitations, and the Landscape Survey
is no exception. We fully acknowledge these and seek to be transparent throughout our analysis. These limitations are particularly apparent when it comes to providing definitive figures on membership in religious groups.

First, the Landscape Survey, like most surveys, was conducted among people who are age 18 and
older, so it documents the religious affiliation of adults, who represent only about three-quarters of the U.S. public. Moreover, as the Landscape Survey illustrates, a significant percentage of Americans have only a vague denominational identification (that is, they tell us they are “just a Baptist” or “just a Methodist”). In fact, many Americans are simply unclear about the religious group to which they belong, ensuring a degree of ambiguity in any survey-based measure of affiliation.

Another limitation is related to the relatively high number of foreign-born individuals in the U.S. who are not fluent in English. According to recent U.S. Census figures, of the approximately 225 million adults in the U.S., more than 34 million are foreign born, and approximately half of this group – around 8 percent of the total number of U.S. adults – is not proficient in English. This number is particularly high among Hispanic immigrants, only about one-quarter of whom are fluent in English. Since many surveys are conducted only in English, the number and views of individuals who are unable to complete a survey in English will not be fully represented.

To help address this shortcoming, the Landscape Survey was conducted in both English and
Spanish, allowing for a more accurate representation of the religious affiliation of Latinos, who constitute nearly half of foreign-born adults in the U.S. It should be kept in mind, however, that even the Landscape Survey was not truly bilingual in nature. In other words, most interviewers were not able to switch between English and Spanish as necessary. Rather, English-speaking interviewers made note of the households they encountered where there was a Spanish-language barrier. These households were later called back by Spanish-speaking interviewers and asked to participate in the survey.

Although this represents a significant improvement over surveys conducted solely in English,
we know from other Pew research that even these efforts fall short of a truly bilingual approach. This has consequences for the survey findings. For example, our previous research shows that Latinos who are able to complete interviews only in Spanish are more likely to be Catholic as compared with Latinos who are fluent in English, and so the Landscape Survey may understate the proportion of Catholics among U.S. Latinos. (For a fuller discussion of the challenge of measuring religious affiliation among Latinos, see Chapter 3.)

Furthermore, although we know from U.S. Census figures that the number of people who are not
fluent in English is much lower among non-Latino immigrants, this still means that other religious groups that have a large proportion of foreign-born members will also likely be undercounted by the Landscape and other surveys. For instance, previous Pew research finds that most English only surveys estimate the Muslim share of the U.S. population to be roughly 0.5%. But Pew’s 2007 survey of Muslim Americans, which was conducted in Arabic, Urdu and Farsi in addition to English, estimates the Muslim share of the population to be higher, at 0.6%. Much of this difference is likely attributable to the more complete representation of the Muslim community yielded by conducting the survey in multiple languages. For this reason, Landscape Survey estimates of the size of various religious groups that have disproportionately large numbers of adherents who are foreign born (such as Buddhists, Hindus and members of other world religions) should be viewed as minimum estimates.

This first report based on the Landscape Survey includes basic information on religious affiliation and provides estimates of the size of religious groups that are as small as three-tenths of 1 percent of the adult population. The report describes and analyzes the relationship between religious affiliation and various demographic factors, including age, ethnicity, nativity, educational and income levels, gender, family composition and regional distribution – including for these smaller groups. The report also examines the sources of the shifts in the religious composition of the U.S., including immigration and changes in affiliation.

Groups analyzed in this report include specific denominations such as the Church of God in
Christ (a prominent historically black Pentecostal denomination) and the United Church of Christ (the largest Congregationalist denomination) as well as groups such as Jehovah’s Witnesses and Unitarians, each of whom account for between 0.5% and 1% of the total adult population. In typical surveys, such groups would be represented by just a few dozen respondents, making it impossible to draw any statistically valid conclusions about the characteristics of these groups. But the large sample size of the Landscape Survey ensures that even smaller groups than these are represented by at least 100 respondents. This provides unprecedented detail on the characteristics and views of America’s multitude of religious groups. (For definitions of these and other religious groups, we recommend the Religion Newswriters Association’s Religion Stylebook and The Associated Press’ Stylebook as starting points.)

Other findings from the Landscape Survey – specifically on Americans’ religious beliefs
and practices as well as their social and political views – will be released this spring. We will extensively probe such topics as belief in God and the afterlife, attitudes toward the authority of sacred writings, frequency of worship attendance and prayer, views on abortion, attitudes about the proper role of government and opinions on foreign affairs. Using the responses to these and other survey questions on a variety of subjects, we will examine the internal diversity that exists on these questions within the country’s various religious groups, including people who are unaffiliated with any particular religion.

Additionally, the survey findings will serve as the basis for a series of portraits that will provide an easily accessible view of the religious and demographic characteristics, beliefs and practices, and social and political views of American religious groups, including such smaller groups as Buddhists, Hindus, Jews and Muslims.

In conjunction with the release of this report, the Forum is introducing some new features on its website, www.pewforum.org. The online presentation of the findings of the Landscape Survey, available directly at religions.pewforum.org, includes interactive mapping, dynamic charts that illustrate key findings and a variety of other tools that are designed to help users delve deeper into the material. The website will be updated as subsequent analyses are released.

It is our hope that the U.S. Religious Landscape Survey will contribute to a better understanding of the role religion plays in the personal and public lives of most Americans.

Luis Lugo

Director, Pew Forum on Religion & Public Life