Faith in the figures

Since 11 September, the USA has decided that it needs to ask something that was previously hidden under the shiny veneer of multiculturalism: just how many Muslims are there in America?

The public policy issues raised by this question encompass immigration and border control, racial profiling, and so on – and when the issues are so sensitive it helps to have firm facts and figures to use as impartial guides for reaching decisions.

But as the nation evaluated the pros and cons of introducing profiling for Muslims, it hit a roadblock – neither the Census Bureau nor the Immigration and Naturalisation Service are allowed to collect any information on religious faith. So we have no official idea of how many Muslims reside in America. To establish a credible figure, we have to wade through a swamp of conflicting data.

The media have allowed many different experts to promote their own figures: ranging from one million to seven million. The variations arise from the different approaches: estimates from individual surveys, indirect calculations based on data from places of worship, and proxy measures like ancestry and country of origin. Each has drawbacks – but some methods are better than others.

The most prominent example using ancestry data is a 1991 study that put the Muslim population at four million. Using US Census Supplemental Survey data on ancestry, this study assumed that the number of Muslims in this country of a specific ancestry (particularly Arab) matched the proportion in their countries of origin. The study then adjusted for African-American Muslims and immigration and birth rates between the data set (1980) and the date of the study was being written.

But estimates from ancestry data usually fail to account for deaths, emigration, or conversions to or from Islam. Also, immigrants often differ markedly from the general population in their countries of origin. So Russian immigrants at the turn of the century were predominantly Jewish, not Russian Orthodox.

So if we can’t pinpoint a reliable estimate of Muslims using ancestry, can we rely on statistics from places of worship? Here, the most prominent example is the 2000 Mosque Study Project (MSP) (1), conducted on behalf of several Muslim groups – which produced the estimate of six to seven million Muslims in America, the figure most frequently cited in the press.

The MSP surveyed individual mosques, finding 340 adults and children participated at the average mosque and another 1629 were ‘associated in any way’ with the average mosque’s activities, yielding a figure of two million Muslims. The authors then adjusted the estimate to six to seven million overall to take into account family members and unaffiliated Muslims.

With impressive candour, the MSP’s lead researcher, Professor Ihsan Bagby of Shaw University in North Carolina, conceded to the Associated Press that the number was a mere ‘guestimation’. Mosques (just like any other body) could have inflated their rolls by counting ‘members’ who were no longer active. Given the loose definition of just who is ‘associated in any way’ with a mosque’s religious life, it is likely that some individuals were ‘associated’ with more than one mosque, resulting in their being counted more than once.

Polling might provide better data. Survey researcher Tom Smith of the National Opinion Research Centre (NORC) at the University of Chicago was commissioned by the American Jewish Committee to come up with a better estimate of the number of Muslims in America. In his report, released in October 2001, Smith drew his estimate from NORC’s most recent General Social Survey (GSS), which found 1.4million adult Muslims. To estimate the number of Muslim adults and children, Smith took the GSS data and made two assumptions: Every Muslim respondent represented a Muslim home and every non-Muslim respondent represented a non-Muslim home. His process yielded an estimate of 1.7million.

Case closed? Not quite. Other researchers’ estimates differ still. The City University of New York (CUNY) recently released their 2001 American Religious Identification Survey (ARIS) (2), which asked members of 50,000 American households to identify their religious affiliation, if any, and that of their spouse or partner. ARIS estimated 1.1million adult Muslims, which CUNY adjusted to an overall estimate of 1.8million adults and children.

Being drawn from wider, scientifically representative samples, the estimates of religious affiliation provided by ARIS and the GSS both seem reasonable. While a precise figure remains elusive, ‘two million Muslims, give or take a few hundred thousand’ appears to be the correct range for America.

Let this be a lesson. We should be careful about putting too much faith in statistics measuring religious affiliation – personal beliefs that are not as immediately apparent as something like the colour of our skin.

Howard Fienberg is senior analyst and Iain Murray is director of research with the Statistical Assessment Service (STATS), a non-profit, non-partisan research organisation in Washington, DC.
Read on:

spiked-issue: After 11 September

(1) The Mosque Study Project at The Council on American Islamic Relations (CAIR)

(2) See the American Religious Identification Survey 2001 and American Religious Identification Survey 2001: Profile of the US Muslim population