Elusive Numbers: U.S. Population by Religion

My print column this week examines the limited amount of data available about the population of religious groups in the U.S. The Census Bureau doesn’t collect such data, barred from compelling responses because of a 1976 law passed by Congress.
People planning for the future in ZIP code 10007, where the proposed Cordoba House Islamic center in lower Manhattan would be built, know many other things about the neighborhood’s inhabitants, thanks to the 2000 U.S. Census. There were 3,522 inhabitants, 375 of whom identified as Asian and five as American Indian or Alaska native. Eight families and 134 individuals lived below the poverty line. Similar information will be available soon from this year’s census. But because of the law, no one knows how many Muslims live in the area.
In the late 1950s and again in 1976, Americans debated the possibility of including such a question in the 1980 census at public meetings organized by the Census Bureau, according to a history prepared by the Pew Forum. But the Census director at the time, Vincent P. Barabba, decided to exclude such a question, out of fear that including it would lead to a controversy over the separation of church and state, which could impede the bureau’s work. Later that year Congress passed a law prohibiting mandatory questions about religion.
“It’s a tough tradeoff,” said Barabba, who is now retired. “You want to collect everything that’s of interest in society. [However,] you want to be careful about which ones you ask.” Describing his position at the time, he said the key question was, “What’s the value of the information, and what’s the cost relative to how it might affect other information?”
Several researchers say they wish the census would ask the question. However, Ibrahim Hooper, spokesman for the Council on American-Islamic Relations, opposes the idea. “It would be a bit difficult now to institute a religious question, particularly for Muslims, because there would be so much suspicion that it would be used to target them,” Hooper said. “I’m sure people would be afraid the government would at some point use that data for some kind of surveillance or, in the worst-case scenario, to round them up in the same way Japanese-Americans were rounded up after the outbreak of World War Two.” (In a 2007 survey of American Muslims by the Pew Forum on Religion & Public Life, 54% of Muslims said they believe the government singles out Muslims for extra surveillance.)
Steven M. Cohen, a sociologist at Hebrew Union College who has worked on estimates of the Jewish population, would like to see the U.S. Census ask about religion, but said some Jewish groups’ fears of anti-Semitism helped pass the 1976 law that bans such a question. “It’s the Jewish agencies that have been among the most vociferous in opposing asking about religion on the U.S. census,” Cohen said. “It’s a misapplication of a tragic history to the American circumstance.”
Censuses in Canada and England ask about religion. The results are used, among other things, to monitor discrimination, diversity, and the need for faith-based schools, Peter Frayne, spokesman for Statistics Canada, wrote in an email. He added, “Statistics Canada has conducted extensive consultations for every census, including on the usefulness of information on religion. In particular, the results from the 1991, 2001 and 2011 Census consultations concluded a strong support to collect this information.” Canada is moving the question, along with many others, to a voluntary survey, in a controversial move ahead of next year’s census.
England & Wales introduced a census question on religion in 2001. It is the only voluntary question on the census, according to Glen Watson, director of the census. “For England & Wales, most people were content to provide the information — for example in 2001 about 7% of people who filled in a Census form left the religion question blank,” Watson wrote in an email. “This compared (for example) to 3% who left the ethnic group question blank. Our testing in preparation for 2011 has confirmed that the majority of people are happy to provide the information.” Of local and regional governments who participated in a consultation about the 2011 census, 91% said they require information on religion — to, among other things, “promote legal obligations to prevent discrimination and promote equality.”
More than 70 countries asked about religion on their national censuses conducted in the decade up to 2004, according to Stefan Schweinfest, chief of the statistical services branch of the United Nations statistics division. Of the 62 censuses reviewed between 2005 and 2014, 32 asked a question on religion, according to Erlinda Go, officer-in-charge of the demographic and social statistics branch of the U.N. statistics division.
The U.N. also conducts or oversees some demographic surveys. It offers guidelines for whether to ask about religion, and how: “Owing to the sensitive nature of a question on religion, special care may be required to demonstrate to respondents that appropriate data protection and disclosure control measures are in place. It is important that the responding public be informed of the potential uses and needs for this information.”
U.S. surveys of religion have been conscious of the sensitivity of the question. The American Religious Identification Survey, conducted by researchers at Trinity College, includes an open-ended question about religion in a much broader survey, so as not to heighten fears among respondents who are private about their faith. Other questions on the survey might include, “What do you think about the safety of cars?”
Pew’s 2007 survey of American Muslims took care to begin interviews with “neutral or innocuous questions,” according to the survey report, such as newspaper subscription, because of Muslim Americans’ concerns about government surveillance.
The latest iteration of these surveys arrived at figures of roughly 1.3 million to 1.4 million Muslim adults in the U.S. But Ihsan Bagby, associate professor of Islamic studies at the University of Kentucky, said the telephone surveys may undercount Muslims, who may fear disclosing their religion because of potential anti-Muslim prejudice. “I just feel in the present time that there are more Muslims than other groups who will give a false answer, or no answer at all,” Bagby said.
Bagby counted Muslims differently in 2000, compiling a list of mosques and then surveying them for attendance counts. He arrived at a total of two million worshippers, which when combined with less actively religious Muslims would imply a much larger total population than the phone surveys suggested. Bagby is updating his own mosque study from 2000. He starts by collecting a list of all U.S. mosques, no easy feat since there is no central registry. New York State has 230 of approximately 1,900 in the country, and California is second with 128, according to Bagby. Greater New York City has one in 10 of the mosques in the U.S.
Some researchers have criticized Bagby’s approach because mosques might overstate attendance to boost their importance. “That’s always a concern of a researcher,” Bagby said, but he noted that independent counts by his research group in Detroit found differences of just a few percentage points with mosque estimates.
Bagby’s study has been used as the basis for estimates by some Muslim groups of a national population of seven million (an estimate repeated by President Obama last year in Cairo). Such larger estimates formed the basis for a widely cited estimate of the New York City Muslim population by Columbia University researchers a decade ago of 600,000 — later revised to 700,000. “These are estimates and extrapolations,” said Louis Cristillo, an anthropologist and lecturer at Columbia’s Teachers College who worked on the count. “We felt pretty comfortable with the estimate we came up with.”
But some religion researchers say such estimates for Muslims are inflated. “Both the Muslim boosters and Muslim enemies have always exaggerated the numbers,” said Barry Kosmin, a sociologist at Trinity College and a principal investigator with the American Religious Identification Survey. Some of the supporters of larger numbers cite estimates of national origin and ethnicity. But Kosmin pointed out that extrapolations from national origin can misstate the total, because immigrants — or descendants of immigrants — from Muslim-majority countries are less likely to be Muslim when they arrive or to identify as Muslim years or generations later.
Similar uncertainty has surrounded estimates of the U.S. Jewish population. A survey of American Jews a decade ago, the National Jewish Population Survey, counted 5.2 million American Jews. Brandeis University researchers re-analyzed the data and found that because of groups that were underrepresented in the survey, the true population likely was about a million higher. “I agree that the number is higher than reported in study,” said Cohen, who worked on the initial survey.
He’d like to update the number but hasn’t been able to find funding. “There is no active plan to do that,” Cohen said. “It’s a vision and a dream in my heart.” … It will help us create better policies, and better awareness of ourselves.”
His research into counting Jews shows how challenging it can be to define religion. For the National Jewish Population Survey, researchers defined the category of Jews as “people who identify as Jews, and have some legitimacy about them that could be recognized by most other Jews,” Cohen said. One decision was to consider a single Jewish parent the same whether that parent was the mother or father, though rabbis consider Judaism to pass through the mother. (Cohen also worked on a 1982 study that found 1.1 million Jews in the five boroughs of New York City.)
One of the Brandeis researchers, Leonard Saxe, has helped devise a method to combine multiple religion surveys and glean from them more reliable numbers. This helps to correct for a problem even in large national surveys — where there simply aren’t enough respondents from smaller religious groups, such as Jews or Muslims. Trying to estimate their population from such surveys “would be ridiculous,” said Saxe, professor of Jewish community research and social policy at Brandeis. “It would be useless.” By combining surveys, “you can begin to accumulate pretty reliable data,” Saxe said.
One thing many researchers can agree on is that the Census Bureau should ask about religion, to produce even more reliable data. “I don’t understand the logic of why religion is not on the census,” Bagby said. “Maybe it is time. Our first identity is as Americans. But we also identify with religion.”
“Religion is very central — it helps define who people will marry and where they will be comfortable living,” said Brian Grim, senior researcher for the Pew Forum on Religion & Public Life.
Estimates of world population counts also are imprecise, though they’ve gotten better as more countries conduct national censuses, as part of a United Nations push for such surveys. “It’s a lot easier than it was 20 or 30 years ago,” said Grim, of Pew. “More censuses over that time period have included religion. And there are all kinds of data sources that weren’t available before — demographic surveys, and other population surveys.”
Some trends are working in the opposite direction. Austria had, until 2001, asked about religion in its census every 10 years. But now, like several other European countries, it is moving toward a registration-based census, which involves drawing on existing databases rather than interviewing every resident nationwide, according to Teresa Bernhart, spokeswoman for the Austrian embassy in Washington, D.C. None of these databases include religion, so Austria may no longer have counts of its population by religion.
But Daniel A. Madigan, professor of theology at Georgetown University, questions the use of such estimates for comparing religions by, say, growth rates. “Religion can grow because of conversion, because countries where it is prevalent have large population increases, because of immigration,” or for other reasons, Madigan said. High growth rates for a religion “usually are presumed to be an indicator that it is most popular — that people are leaving other religions to join it. But figures don’t necessarily indicate that.”