|Estimating the Muslim Population in the United States|
Estimating the Muslim Population in the United States
Dr. Tom W. Smith
by Tom W. Smith
About Dr. Tom W. Smith
NORC, University of Chicago
Tom W. Smith is a nationally recognized expert in survey research specializing in the study of social change and survey methodology. Since 1980 he has been co principal investigator of the National Data Program for the Social Sciences and director of its General Social Survey (GSS). This is the largest and longest term project supported by the Sociology program of the National Science Foundation. The mission of the GSS is to monitor changes and consistencies in American society and to develop models to explain change (or stability).
Smith has authored over 225 scholarly papers. His work in the social change area includes both wide ranging studies that integrate trends across many different topics such as Trends in Public Opinion: A Compendium of Survey Data (1989), "Liberal and Conservative Trends in the United States Since World War II," Public Opinion Quarterly (1990), and "Is There Real Opinion Change?" International Journal of Public Opinion Research (1994) and specialized studies on such matters as public attitudes toward the most important national problems, crime and punishment, ethnic and racial relations, governmental spending priorities, and sexual behavior.
He has also written on virtually every aspect of survey methods including nonresponse, question wording, nonattitudes, order and context, recall, respondent understanding, and test/retest reliability. He has organized an extensive program of data quality assurance and methodological analysis on the GSS and edits the GSS Methodological Report series which contains 87 papers.
Smith has taught at Purdue University and Northwestern University, and in Sociology and Political Science at the University of Chicago.
Smith has served on the National Academy of Sciences' Panel on Survey Measurement of Subjective Phenomena, the Board of Directors of the Roper Center for Public Opinion Research, the National Science Foundation's Committee on Data Sharing in the Social Sciences, and the Centers for Disease Control and Prevention's Subcommittee on Monitoring the AIDS Epidemic. He was awarded the 1994 Worcester Prize by the World Association for Public Opinion Research for the best article on public opinion. He currently is on the editorial boards of the Public Perspective and the International Journal for Public Opinion Research and is former editor of the Poll Trends section of Public Opinion Quarterly. A long time member of the American Association for Public Opinion Research, he has served as secretary treasurer, standards chairman, and counsellor at large.
1) reviews estimates of the Muslim population in the United States used by the mass media,
2) describes estimates that have been proposed during the last five years,
3) evaluates the few existing attempts to measure this population based on data and an explicit methodology,
4) marshals the best available information from national surveys, and
5) considers what government census and immigration statistics suggest about the Muslim population of the United States.
Latest Media Estimates
Since the September 11 terrorist attacks, the media has used estimates of the Muslim population in the United States of 5-8 million, with an average of 6.7 million or 2.4 percent of the total population.
Range of Recent Estimates
Over the last five years (1996-2001) estimates of the Muslim representation in the total population have ranged between 3 million and 9 million. Looking at 20 specific estimates, they spread out as follows:
||Number of Estimates
|Less than 4 million
|7 million +
|Average of Estimates
None of the 20 specific estimates during the last five years is based on a scientifically-sound or explicit methodology. All can probably be characterized as guesses or assertions. Nine came from Muslim organizations such as the Islamic Society of North America, the Muslim Student Association, the Council on American-Islamic Relations, the American Muslim Council, and the Harvard Islamic Society or unspecified “Muslim sources.” None of these sources gave any basis for their figures.
Six were from general reference works (Britannica - 3, World Almanac - 2, and Atlapedia Online- 1). The Britannica Book of the Year estimates were prepared by David B. Barrett and Todd M. Johnson, but they do not explain how their figures are derived in the Britannica and have not responded to requests for information on the basis of their calculations. One source citing the World Almanac claims that the World Almanac’s figures came from the Yearbook of American and Canadian Churches. While most church figures in the World Almanac do come from the Yearbook, the Muslim estimate and several others were instead “based on reliable estimates; figures from other sources may vary” (World Almanac, 1998). The Atlapedia also gave no basis for its estimate that Muslims were 2 percent of the population.
Two of the estimates purport to use figures from surveys, but neither actually does so. Mujahid (2001) claims that “public opinion survey organizations normally get about 2 percent Muslims in their samples.” However, no survey estimate available places the Muslim population above 1 percent, and as Table 1 below indicates, most place the Muslim share well below 1 percent. Similarly, Robinson (2001) claims, “Some surveys show that there are about 3.5 to 3.8 million Muslims (1.4 to 1.5 percent) in the U.S.” However, the source cited for this (Peters, 1997) gives no basis for its estimate of 3.5-3.8 million and specifically does not mention a survey as the source.
Two cited other sources—“experts” in one case and the U.S. ambassador to Pakistan in the other. No basis was given for their estimates.
Two cited no source or basis for their estimates.
Over the last twenty years, there have been only a few estimates of the Muslim population based on data and an explicit methodology.
Stone (1991) estimated the Muslim population in United States by taking the national origins of the U.S. population in 1980 according to the Census ancestry question and assuming that the same percentage of U.S. residents from these groups was Muslim as the proportion of Muslim residents in these nations, and then adding a large estimate for Muslims among African Americans. This gave an estimate of 3,300,000 in 1980. This number was then brought forward to 1986 by adding in an estimate of births plus immigrants during 1980-86 (assuming, as before, that immigrants matched the religious profile of their countries of origin). This gave a 1986 estimate of 4,000,000.
There are several problematic aspects to these procedures. First, it has been well documented that immigrant flows to the United States often differ greatly from the profile of the country of origin. For example, Russian immigrants in the late 1800s and early 1900s were heavily Jewish, when the vast majority of Russians were Russian Orthodox. Similarly, the majority of immigrants from Lebanon have been Christians, not Muslims. The estimate that 880,000 Muslims came from Eastern Europe is particularly suspect in this regard. All other estimates of the national origins of American Muslims indicate that few came from Europe (Zogby, 2000; Bagby, Perl, and Froehle, 2000; Numan, n.d.), compared to Stone’s figure that 27 percent did.
Second, many of the estimates of the Muslim population of country of origin are high, such as the figure that 18 percent of China is Muslim. Third, there seems to be little basis for assuming that 1,000,000 or 3.7 percent of African Americans in 1980 were Muslim. Lastly, the projections for 1980-1986 a) repeated the questionable country-of-origin assumption about the religion of immigrants and allowed for no emigration, b) added in births, but allowed for no deaths, and c) ignored gains or losses from conversions.
Kosmin and Lachman
The second study with an explicit methodology was the Kosmin and Lachman’s National Survey of Religious Identification (1993). From April 1989 to April 1990 they had a major survey research company, ICR, collect religious identification data on 113 rounds of their bi-weekly poll. Altogether 113,723 people were interviewed, and 295 respondents or 0.3 percent of their sample were Muslims. This yielded an adult estimate of 527,000 Muslims. They believed, however, that Muslims were undercounted in their survey and put the adjusted figure at 0.5 percent or 877,000 adult Muslims. They felt that Muslim immigrants might have been underrepresented because of language limitations and because of suspicion among those from “societies with deep religious hostilities.” They also thought that African American Muslims might represent 2 percent rather than 1 percent of Blacks due to people missed in prison and “youth on the streets,” and that some other Muslims may have mentioned particular Islamic sects that interviewers did not recognize as Muslim. If their adjusted figure is applied to the total population, using the 1990 adjusted population figures, there were then an estimated 1,264,000 Muslims.
A third study with an explicit methodology was carried out in Illinois by Ba-Yunus (1997) in 1996. Lists of Muslims were gathered from Islamic organizations outside of the Chicago metropolitan area. Then it appears that people with similar last names were added from telephone directories, and finally, respondents from these lists were contacted and a snowball sampling technique was used to locate other Muslims. For the Chicago metropolitan area a sample of names from telephone directories was taken. These procedures yielded estimates that Muslims were 0.9 percent of the population outside of the Chicago metropolitan area, 3.9 percent of the Chicago area population, and 2.9 percent of the population of Illinois.
These procedures represent a creditable approach for estimating the Muslim population in Illinois. However, there are a number of unanswered questions about the Chicago-area survey that raises concerns. First, the survey was apparently not done by an established survey-research institute. Second, the researcher had spent considerable effort in identifying Muslim names before deciding to draw a sample of names, and this may have biased the sampling of names. Third, directories only contain listed numbers, thus excluding often 40 percent or more of urban telephone households, which is why most telephone polls use random digit dialing (RDD) sampling. Fourth, the research group widely advertised the survey among the Muslim community, which may have made Muslims more likely to cooperate than non-Muslims and thereby inflated the Muslim share of respondents. Finally, no information was given on how the survey or its sponsorship was described to respondents or the wording of the questions asked. In addition, even if accepted as non-problematic, it is practically impossible to translate these estimates for Illinois into national figures.
Mosque Study Project
A fourth approach estimates the Muslim population through studying Islamic organizations. There are considerable differences in estimates of the number of mosques and Islamic centers in the United States. Eck (2001) says there are nearly 1400 mosques; the Council on American-Islamic Relations reports that there are “almost 2,000 mosques, schools, and Islamic centers.” Mujahid (2001) states that there are “3,000 small and large Islamic centers, mosques, and prayer locations.”
The most rigorous estimate was from the Mosque Study Project 2000 (Bagby, Perl, and Froehle, 2001) which combined seven lists of mosques, eliminated duplicates, and attempted to verify the existence of each place. This generated a final list of 1209 mosques in 2000. The researchers then drew a sample of 631 and were successful in obtaining information about 416 of the mosques. They found that 340 adults and children regularly participated in the average mosque, and that 1629 were “associated in any way with the religious life of the mosque.” This converts to a national estimate of 1,969,000 mosque-associated Muslims nationally.
On the one hand, these figures naturally exclude all Muslims not associated with a known mosque. On the other hand, there are several reasons to question the 1629 per mosque figure. First, denominations and congregations in general frequently overestimate their membership (Smith, 1991). This typically comes from both intentional overstatements and the failure to cull former members from the rolls. Second, the estimate of 1629 Muslims associated per mosque is very high. It is 2.4 times greater than a similar figure from 1994, 5.6 times greater than the average number attending weekly prayers, and 4.8 times the number of adults and children who regularly participate in the mosques. The mean figure is also 3.3 times the median mosque size of 500. In fact almost 15 percent of the mosque-affiliated Muslims in their sample come from two mosques, each reporting 50,000 affiliates. Third, individuals may be associated with more than one mosque and therefore be double counted. Duplication is particularly likely for the “associated in any way” figure, since the net is cast so broadly. Finally, the report acknowledges that small mosques were underrepresented in the achieved sample of 416 mosques, and this would bias upwards estimates of the size of the average mosque.
In addition to the estimate of 1,969,000 mosque-associated Muslims, Bagby, Perl, and Froehle (2001) also state that “estimates of a total Muslim population of 6-7 million in America seem reasonable in light of the figure of 2 million Muslims who associate with a mosque.” However, this assertion seems untenable. It increases the total number Muslims by three to three-and-a-half times and assumes that for every Muslim associated in any way with a mosque, there are 2 to 2.5 other Muslims with no association whatsoever.
Recent Survey Estimates
Survey estimates of the adult Muslim population living in households in the United States range from 0.2 percent to 0.6 percent (Table 1). Assuming that the same proportion applies to the total adult population as among adults living in households, this puts the total number of Muslim adults in the United States between 418,000 and 1,254,000. The Muslim share of the total population can be estimated from the GSS (General Social Survey) figures by assuming that all members of households with Muslim respondents are also Muslim and that all members of households with non-Muslim respondents are non-Muslims. Under this assumption the Muslim share of the total population is virtually the same as among adults (a tenth of a percentage point higher in 1998 and a tenth of a percentage point lower in 2000). Using 0.5 percent-0.6 percent as the estimates for the total population gives figures of 1,407,000 to 1,689,000.
Several factors have been offered as to why the survey estimates may undercount Muslims among the household population. First, a language barrier among Muslim immigrants might reduce response. Detailed language-use figures from the 2000 Census are not yet available to use as a basis for estimating Muslim non-English speakers, but studies of other groups with high levels of recent immigration find that 20 percent-25 percent of adults lack enough English to participate in a survey. Using these figures suggests that figures of 0.5 percent might be as high as 0.67 percent, or a total of 1,886,000. Second, it has been suggested that some Muslim respondents may have been lost because they mentioned sects of Islam (e.g., Sunni) that were not recognized as Muslim by interviewers. However, this possibility does not exist for the GSS, which recorded exactly what respondents stated was their religion (Muslim, Moslem, Islam, Sunni, etc.) and classified all as Muslim.
Third, it has been suggested that Muslims may be reluctant to admit their religion in surveys. This is unlikely to be a notable factor for three reasons. First, very few people decline to give their religion. Second, among the few declining to give religion, there is nothing in their ethnic profile that points to this group including any Muslims. Third, many Muslims indicate their religious identity by their overt behaviors (e.g., dress, names, daily prayers), so there seems little basis for believing that any appreciable number of Muslims in surveys deny their religion. Finally, it has been argued that as a recent immigrant group, Muslims may decline to participate in surveys in general. There is no direct evidence of this, and indirect indicators suggest this as unlikely. First, once language is taken into consideration, immigrant groups in general are not less likely to participate in surveys than non-immigrants are. Second, the best available information indicates that Muslims in the United States are not a marginalized group cut-off from the general community, but are better educated than the general population, with many employed in managerial and professional capacities (Bagby, Perl, and Froehle, 2001; Ba-Yunus, 1997; Kosmin and Lachman, 1997; Zogby Poll, 2000).
The American Freshman Study of first-year college students does find a somewhat higher number of Muslims (0.9 percent), but this is actually consistent with the lower adult figure of up to 0.7 percent, once language is adjusted for. First, no language adjustment is needed in this case, since college students presumably know enough English to answer simple survey questions. Second, the available demographic data on Muslims indicate that they are younger and better educated than the general public (Bagby, Perl, and Froehle, 2001; Ba-Yunus, 1997; Kosmin and Lachman, 1997; Zogby Poll, 2000) and thus that their share among college students would be higher than their share among the general population.
There are several reasons to give credence to these survey estimates. First, surveys are scientifically valid means of estimating the share of the population that various groups constitute and are the most frequently used method for estimating the size of other religious groups in America (Smith, 1991). Second, the surveys assembled here include estimates by different organizations (academic and commercial) using different methodologies (in-person, telephone, self-completion), and they produce very similar estimates. Third, the current estimates also agree with other survey estimates during the 1990s (Kosmin and Lachman, 1993; Smith, 1996 and 1998; “Where...,” 1991). Finally, among the surveys used are two leading surveys supported by the National Science Foundation (the GSS and ANES), using the most rigorous survey methods. This, of course, does not mean that these survey estimates are certainly correct, but they are the firmest figures that do exist and show a good deal of consistency.
Neither the Bureau of the Census in the decennial Census or in its Current Population Survey (CPS) nor the Immigration and Naturalization Service (INS) in its statistics on immigrants, visitors, and naturalized citizens records religious affiliation. As a result, the only inferences about the Muslim population that can be drawn must rely on figures on national origin or language use, linking these to religious affiliation. Such linkages are so uncertain that no reliable estimate of the Muslim population can be derived from such sources. However, Census and INS figures on national origin can be used to assess the estimates from surveys.
First, the 2000 Census ancestry question found that 0.7 percent of the population mentioned origins in countries with a majority Muslim population. Under the assumption that all were Muslims and no one from non-majority Muslim countries were Muslims as immigrants, applying this to the total population yields an estimate of 1,970,000 Muslims. To this figure, one would have to add non-immigrant Muslims, mostly African Americans. Using the Kosmin and Lachman (1993) estimate of 0.9 percent-2.0 percent of African Americans being Muslims would add 328,000-728,000, for a total of 2,298,000-2,698,000. Alternatively, using estimates that African Americans make up from 24 percent to 42 percent of all American Muslims (Bagby, Perl, Froehle, 2001; Britannica, 2001; Kosmin and Lachman, 1993; and Zogby, 2001), and assuming that the ancestry estimates are not already counting some of this group, gives total estimates of 2,592,000-3,397,000.
Second, country of origin for the foreign born can be used in a similar manner. Detailed, country-specific breakdowns do not yet exist from the 2000 Census, but the 1997 CPS indicates that about 1,153,000 or 4.5 percent of the foreign-born are from Muslim countries. If Zogby’s estimate that 78 percent of American Muslim adults are foreign-born is used, this would translate into about 1,456,000 Muslims overall.
Third, from 1987-1997 the INS counted 853,619 immigrants from countries with a majority Muslim population. This represents 7.9 percent of all immigrants during the period. There were about 24,800,000 immigrants in 2000, and applying the 7.9 percent would yield 1,959,000 Muslim immigrants. Again using Zogby’s estimate that 78 percent of American Muslim adults are foreign- born would yield a total estimate of 2,524,000.
Fourth, in 1987-1997 the INS reported 350,558 naturalized citizens from countries with a majority Muslim population. This represents 8.4 percent of all naturalizations during this period. There were about 10,622,000 naturalized citizens in 2000. Applying the 8.4 percent to this figure yields an estimate of 892,000 naturalized Muslim citizens. Using the same level for non-naturalized immigrants and the Zogby adjustment makes a total estimate of about 2,685,000.
In sum, the crude estimates from Census and INS figures suggest a Muslim population in the range of 1,456,000 to 3,397,000 or 0.5 percent-1.2 percent of the total population. These figures bracket the best survey estimate of 1,886,000.
Estimates of the Muslim population of the United States 1) vary greatly, 2) are seldom based on any credible scientific methodology, and 3) seem to have been undergoing inflation of late, rising a million from the average claim in 1996-2001 of 5.65 million to a post-September 11average figure of 6.7 million.
The best, adjusted, survey-based estimates put the adult Muslim population in 2000 at 0.67 percent or 1,401,000, and the total Muslim population at 1,886,000. Even if high-side estimates based on local surveys, figures from mosques, and ancestry and immigration statistics are given more weight than the survey-based numbers, it is hard to accept estimates that Muslims are greater than 1 percent of the population (2,090,000 adults or 2,814,000 total).
Thus, the average number being cited by the media at present (6.7 million) is 2.4 to 3.6 times greater than the best available estimates (1.9-2.8 million).
Survey Estimates of Muslim Population
|General Social Survey (GSS)
|American National Election Study
American Freshman - Sax, 2001.
American National Election Study--analysis by author. The lower figure is the percentage of all respondents who identified as Muslim. Only those who attend religious services or who say they identify with a religion even though they do not attend are asked their denomination. The higher figure uses as the base only those who were asked their religion.
Barna - Barna, 1999. This report says that 7 percent or 14,000,00 of the adult population are atheists or agnostics and that they outnumber Muslims by 14:1. This would put the Muslim share at about 0.5 percent and 1,000,000.
Gallup – “Latest...,” 2000; “Latest...” 2001; and special analysis for author by Gallup.
GSS - Davis, Smith, and Marsden, 2001 + analysis by author.
“A Brief History of Islam in the United States,” at www.colostate.edu/orgs/msa
Ahmed, Saeed, “American Under Attack: The Human Toll,” Atlanta Journal and Constitution, Sept. 12, 2001, p. 3a.
Ali, M.M., “Muslims in America: The Nation’s Fastest Growing Religion,” Washington Report on Middle East Affairs, 15 (May/June, 1996), pp. 13, 107.
Bagby, Ihsan; Perl, Paul M.; and Froehle, Bryan T., “The Mosque in America: A National Portrait,” Unpublished report of the Mosque Study Project, Council of American-Islamic Relations, April, 2001.
Barna Research Group, “Atheists and Agnostics Infiltrating Christian Churches,” October 15, 1999, at www.barna.org
Ba-Yunus, Ilyas, “Muslims of Illinois: A Demographic Report,” unpublished paper, Islamic Studies Institute, East West University, Chicago, 1997.
Cavaan, Azell Murphy, “About Islam,” Boston Herald, Sept., 26, 2001, p. 56.
Chaves, Mark; Schraeder, Peter J.; and Sprindys, Mario, “State Regulation of Religion and Muslim Religious Vitality in the Industrialized West,” Journal of Politics, 56 (Nov., 1994), pp. 1087-1097.
Council on American Islamic Relations, “About Islam and American Muslims,” at www.cair-net.org
Davis, James A.; Smith, Tom W.; and Marsden, Peter V., General Social Surveys, 1972-2000: Cumulative Codebook. Chicago: NORC, 2001.
Eck, Diana L., A New Religious America. San Francisco: Harper, 2001.
“Election 2000...,” AMC Report, 10 (Winter, 2001) at www.amconline.org
Encyclopaedia Britannica Book of the Year, 1997-2001.
Haddad, Yvonne Yazbeck, ed., The Muslims of America. New York: Oxford University Press, 1991.
Harvard Islamic Society, “Facts about Islam,” at www.digitas.harvard.edu
Hasan, Asma Gull, Americans Muslims: The New Generation. New York: Continuum, 2000.
Kosmin, Barry A. and Lachman, Seymour P., One Nation Under God: Religion in Contemporary American Society. New York: Harmony Books, 1993.
“Latest Religious Preference for U.S.,” Emerging Trends, 22 (March, 2000), p. 2.
“Latest (2000) Religious Preference for U.S.,” Emerging Trends, 23 (January, 2001), p. 4.
Linder, Eileen W., Yearbook of American and Canadian Churches, [1997-2001]. Nashville: Abingdon Press, 1997-2001.
Milbank, Dana and Wax, Emily, “Bush Visits Mosque...,” Washington Post, September 18, 2001, p. A1.
Mujahid, Abdul Malik, “Muslims in America: Profile 2001,” www.soundvision.com
“Non-Christian Religious Adherents in the United States,” at www.infoplease.com
“Not the Act of Real Muslims,” Omaha World-Herald, p. 6b.
Numan, Fareed H., “The Muslim Population in the United States: A Brief Statement,” at www.amcoline.org
Nyang, Sulayman S., Islam in the United States of America. Chicago: ABC International, 1999.
Peters, William, “Religion in America,” U.S. Society and Values, 1 (March, 1997), pp. 15-17.
Pluralism Project, “Statistics by Tradition,” www.pluralism.org/resources/tradition.php
Robinson, B.A., “Religious Makeup of the United States,” 2001 at religioustolerance.org
Sanchez, Mary, “Voices of Reason Counter Misperception of Muslims,” Kansas City Star, Sept. 13, 2001, p. A4.
Sandstrom, Karen, “Beyond Stereotypes...” Plain Dealer, September 16, 2001, p. A4.
Sax, L.J., et al., The American Freshman: National Norms for Fall 2000. Los Angeles: Higher Education Research Institute, 2001.
Smith, Tom W., “Counting Flocks and Lost Sheep: Trends in Religious Preference Since World War II,” GSS Social Change Report No. 26. Chicago: NORC, 1991.
Smith, Tom W., “Moslems,” unpublished analysis, 1996.
Smith, Tom W., “Muslims,” unpublished analysis, 1998.
Stone, Carol L., “Estimate of Muslims Living in America,” in Haddad, Yvonne Yazbeck, ed., The Muslims of America. New York: Oxford University Press, 1991.
“United States of America,” Atlapedia Online at www.atlapedia.com
“United States of America,” Middle East News and World Report at www.middleeastnews.com
U.S. Department of State, “Fact Sheet: Islam in the United States,” 2001.
U.S. Immigration and Naturalization Service, Statistical Yearbook of the Immigration and Naturalization Service, 1997. Washington, DC: GPO, 1999.
Van Biema, David, “As American As...” Time, Oct. 1, 2001, pp. 72-74.
Vozzella, Laura, “For Better and Worse...” Baltimore Sun, September 18, 2001, p. 5A.
“Where Are the Muslims in the United States?” Emerging Trends, 13 (September, 1991), pp. 4-5.
World Almanac and Book of Facts, 1998.
Zogby Poll for the American Muslim Council, 2000 at www.amcomline.org.