Disability Studies Quarterly
Fall 2006, Volume 26, No. 4
Copyright 2006 by the Society
for Disability Studies

A Statistical Note on the Religiosity of Persons with Disabilities

Gerry Hendershot, Ph.D.
Consultant on Disability Statistics
4437 Wells Parkway
University Park MD 20782-1126
Phone: (301) 927-1120
E-mail: ghendershot@earthlink.net


Indicators of the religiosity of persons with and without disability are compared using statistics from the National Survey of Family Growth, which represents community-dwelling persons of reproductive age in the U.S. It is found that persons with disabilities are less likely than persons without disabilities to attend religious services and less likely to report that religion is important in their daily lives.

Keywords: disability, religion, statistics, NSFG, NOD/Harris


The principal purpose of the statistical analyses of survey data reported here is to consider two questions: Are persons with disabilities in the United Stares more or less "religious" than persons without disabilities? If so, can that difference be attributed to their disability status, or is it the spurious result of confounding factors?

Statistical analyses of survey data cannot, in themselves, answer those questions definitively, because there are inherent shortcomings to survey data and to statistical analyses. Statistical analyses must be evaluated along with other kinds of evidence, such as qualitative analyses, to answer the questions posed.

Religiosity (or religiousness) is a general term used by sociologists to refer to religious beliefs, feelings, and behaviors. The concept of religiosity has multiple dimensions that can be reflected in statistical measures (Hill and Hood, 1999). Among the most frequently used indicators of religiosity are participation in religious services and the felt importance of religion in daily life.

Religiosity varies over time and across subgroups of the population. The American Religious Identity Survey (Kosmin & Mayer, 2002) interviewed a random sample of 50,000 adults in the U.S. to study differentials in religiosity according to gender, age, and race/ethnicity. Among the findings:

  • Women were more likely than men to report that they were "religious" (78% vs. 72%)
  • Older people were more likely than younger people to report that they were religious (81% among those 65 and over, 70% among those 18-34).
  • African-Americans were more likely than non-Hispanic whites and Hispanics to report that they were religious (81%, 77%, and 75%, respectively).

With few exceptions, other studies using a variety of methods and measures have found similar differentials in religiosity by gender, age, and race/ethnicity.

National statistics on the religiosity of persons with and without disabilities have been produced in a series of Surveys of Americans with Disabilities conducted by the National Organization on Disability (NOD) and Harris Interactive, Inc. The NOD/Harris surveys were conducted in 1986, 1994, 1998, 2000, and 2004. In all five years, people with disabilities were less likely than persons without disabilities to attend worship once a month or more (Harris Interactive, 2004, pp.19-24). The difference between those with and without disabilities was largest in 2000 (47% and 65%, respectively) and smallest in 1998 (54% and 57%, respectively).

The 2000 and 2004 NOD/Harris surveys (Harris Interactive, 2000; Harris Interactive 2004) reported national statistics on the felt importance of their religious faith for people with and without disabilities. In 2000, the percent who said their religion was "very important" or "somewhat important" was about the same for those with and without disabilities (87% and 84%, respectively), and in 2004 it was exactly the same (82%).

To summarize, by one indicator of religiosity, participation in religious services, persons with disabilities have been found to be less religious. With respect to another indicator of religiosity, however, the felt importance of religion, persons with disabilities appear to be as religious as persons without disabilities. This pattern of religiosity among persons with disabilities has been interpreted to mean that lower participation in worship by persons with disabilities is not caused by lack of interest, but by barriers to participation in architecture, communication, or attitudes (Rife & Thornburgh, 1996).

Source and limitations of the data

The source of data for the analyses presented below is the National Survey of Family Growth (NSFG), Cycle 6. The purpose of the NSFG is to provide reliable national data on trends and differentials in fertility and their underlying factors, such as sexual unions, reproductive health, and contraception. The NSFG has been conducted periodically by the National Center for Health Statistics since 1973. In Cycles 1-5, the NSFG sample represented only women of reproductive age, 15-44 years. Cycle 6 included both women and men in that age range. Data collection for Cycle 6 was conducted in 2002 and 2003. Micro-data files from Cycle 6 were released to the public in December 2004 (Groves, Benson, & Mosher, 2005).

The NSFG sample represents the household population of the United States. Persons in institutions, such as nursing homes and prisons, are not included. The sample design is multi-stage, stratified, and clustered to produce reliable statistics at a reasonable cost. The sample size in Cycle 6 was about 7,500 women and 5,000 men. The Cycle 6 data were collected by the Institute for Social Research, University of Michigan, under contract to NCHS. Specially trained interviewers conducted standardized, face-to-face, laptop computer-assisted interviews in the homes of sample persons. Respondents were offered $40 as an incentive to participate. The response rate was 79%. The "weights" for sample cases (the number of persons in the population represented by a sample person) include a correction for nonresponse.

For the first time in Cycle 6, the NSFG included two questions to identify persons with disabilities. The questions are those used by state surveys in the Behavioral Risk Factors Surveillance System, or BRFSS (Centers for Disease Control and Prevention, 2005), and they are recommended for use in surveys tracking progress toward disability-related health objectives for the year 2010 (Department of Health and Human Services, 2001). While they give no detail on type of disability, they have been used successfully to identify the general disability population (Centers for Disease Control and Prevention, 2000). The two questions on disability are:

  • Are you limited in any way in any activities because of physical, mental, or emotional problems?
  • Do you now have any health problem that requires you to use special equipment, such as a cane, a wheelchair, a special bed, or a special telephone?

The BRFSS classifies persons who answer "yes" to either question as having a disability. That practice will be followed in the present study. By this definition there are 1125 persons with disabilities in the NSFG sample, representing an estimated population of about 12 million persons in the 15-44 age range.

Because religion is a factor related to fertility, the NSFG has always included a series of questions about religion. This analysis uses the NSFG questions on attendance at religious services and the felt importance of religion in daily life. The questions (and answer categories) are:

  • Attendance: "About how often do you attend religious services (more than once a week, once a week, 1-3 times a month, less than once a month, never)?"
  • Importance: "Currently, how important is religion in your daily life (very important, somewhat important, or not important)?"

The NSFG maintains high levels of sample survey quality; nevertheless, it is subject to the usual types of "total survey error" (Groves, 1989). Only sampling error is directly measured in this analysis. With respect to coverage error, the NSFG does not cover persons over 45 years of age or persons in long term care facilities, groups that have relatively high rates of disability.

Data analysis and findings

All of the estimates of statistics in Tables 1 and 2 were made using computational procedures that weight the sample cases to make national population estimates, and estimate the sampling errors of population estimates (measured by the Standard Error) with appropriate corrections for the sample design parameters (Stata 7, 2002). Because the sample design is "complex" (stratified and clustered), estimates of sampling errors based on assumptions of simple random sampling (SRS) would not be correct, and would usually overstate the statistical significance of observed differences.

Table 1 (below) compares people with and without disability with respect to two measures of religiosity: frequency of attendance at religious services and the felt importance of religion in a person's life. The sample included 1,225 persons of reproductive age with disabilities and 11,340 without disabilities. When the sample cases are weighted, they produce an estimate of 12,092,000 persons of reproductive age with disabilities and 54,980,000 without disabilities (rounded to the nearest thousand). All of the estimates in Table 1 meet the conventional standard for statistical reliability (a Relative Standard Error of less than 30%).

Table 1. Sample size, estimated population size, and percent distributions of persons 15-44 years of age by two measures of religiosity, according to disability status: United States, 2002

Attendance and Importance Disability status
  Disability No disability
Sample size 1,225 11,340
Estimated population 12,092 54,980
Attendance Percent distribution
More once a week 9.7 10.6
Once a week 17.2 20.8
1-3 times a month 15.4 16.6
Less than once a month 27.0 27.6
Never 30.7 24.5
Very important 44.9 44.1
Somewhat important 26.8 32.4
Not important 28.3 23.5

* Percents may not add to 100 because of missing data.

Percentages for the two religiosity variables are italicized if the difference between estimates for persons with and without disabilities is significantly different (p < 0.05). Table 1 shows that people with disabilities are significantly more likely than persons without disabilities to say that they never attend worship (30.7% and 24.5%, respectively) and significantly more likely to say that their religion is not important in their daily lives (28.3% and 23.5%, respectively); that is, by both measures disability is associated with low religiosity.

As noted in the introduction, studies of the general population have found that gender, age, and race/ethnicity are statistically associated with religiosity. Men, young people, and African-Americans are less likely than their comparison groups to be religious. Because gender, age, and race/ethnicity-ethnicity are also related to the prevalence of disability, they may confound the relationship between disability and religiosity found in Table 1.

To test this hypothesis, a multivariate analysis was performed. Table 2 (below) shows the results of a multiple logistic regression predicting two binary indicators of low religiosity--never attending religious services (vs. sometimes attending), and regarding religion as not important (vs. somewhat or very important). In each case, the predictor variables are disability (none or any), sex, age (single years), and three race/ethnicity groups (Hispanic, nonHispanic white, and nonHispanic black). Statistically significant regression statistics are italicized.

Table 2. Results of multiple logistic regressions of sex, age, and Hispanic origin and race, and disability on two measures of low religiosity—attendance and importance

Low religiosity and Predictor variables Regression statistics
  Odds ratio Probability
Never attends worship
Hispanic 0.95 ns
White 1.19 ns
Black 0.56 <0.001
Disability 1.40 <0.001
Religion not important
Sex 1.69 <0.001
Age 0.99 <0.001
Hispanic 0.79 ns
White 1.56 <0.01
Black 0.64 <0.01
Disability 1.28 <0.05

Table 2 shows that sex and age are significant predictors of both measures of low religiosity. But the most important finding for present purposes is that disability is significantly related to both indicators of low religiosity, independently of age, sex, and race/ethnicity-ethnicity. Persons with disabilities are 40% more likely than those without disabilities to report that they never attend worship, and 28% more likely to say that religion is not important in their lives. Regardless of their gender, age, or race/ethnicity, persons with disabilities are less likely than persons without disabilities to attend religious services and to regard religion as important in their lives.


Earlier studies found consistently that persons with disabilities are less likely to attend religious services than persons without disabilities, and this study confirms that finding. That finding has been interpreted as resulting from barriers to participation: it is argued that persons with disabilities are as inherently religious as other persons, but they are less able to attend worship because of architectural, communication, or attitudinal barriers.

Available evidence on other indicators seemed to support that argument by showing that persons with disabilities were at least as likely as other persons to regard religion as important in their lives. This analysis contradicts the usual interpretation by showing that persons of reproductive age with disabilities are not only less likely to attend religious services, but also less likely to regard religion as important in their lives.

This could be interpreted to mean that the low evaluation persons with disabilities put on the importance of religion in their lives is the cause of their low attendance at religious services. Alternatively, barriers to participation may cause infrequent worship attendance, which then causes a devaluation of the importance of religion in life. The data in the NSFG do not allow for a determination of the causal sequence, if any, between low importance and low attendance, and the NSFG has no information about barriers to attendance. Evidence to evaluate these different interpretations will have to come from other sources.

It should be noted again that the NSFG sample limits the generality of any findings and interpretations because it includes only people living in the community who are of reproductive age; put differently, the sample excludes persons over age 45 years of age, and people living in prisons, nursing homes, and other long term care facilities. The excluded populations have relatively high rates of disability, especially serious disability. Their exclusion limits the generalizability of the NSFG findings regarding religion and disability to community dwelling persons of reproductive age who have, on average, less serious disabilities.


Centers for Disease Control and Prevention. Behavioral Risk Factors Surveillance System. Available at: http://www.cdc.gov/brfss/index.htm. Accessed August 30, 2005.

Centers for Disease Control and Prevention. (1998). State-specific prevalence of disability among adults–11 states and the District of Columbia, 1998. Mortality and Morbidity Weekly Report, 49, 711—714.

Department of Health and Human Services. (2000). Healthy people 2010 (Conference Edition, in Two Volumes). Washington, D.C.: Author.

Groves, R. (1989). Survey errors and survey costs. Indianapolis, IN: Wiley & Sons, Inc.

Groves, R., Benson, G., Mosher, W., et al. (2005). Plan and operation of cycle 6 of the National Survey of Family Growth. Vital and Health Statistics, Series 1, No. 42. Washington, D.C.: National Center for Health Statistics.

Hill, P., & Hood, R. (1999). Measures of Religiosity. Birmingham, Alabama: Religious Education Press.

Harris Interactive. (2000). 2000 N.O.D./Harris survey of Americans with disabilities. New York, NY: Author.

Harris Interactive. (2004). 2004 National Organization on Disability/Harris survey of Americans with disabilities. New York, NY: Author.

Kosmin B.A., & Mayer, E. (2002). American Religious Identification Survey. Available at http://www.gc.cuny.edu/faculty/research_briefs/aris/aris_index.htm. Accessed March 2, 2005.

Rife, J.M., & Thornburgh, G. (1996). From barriers to bridges: a community action guide for congregations and people with disabilities. Washington, D.C.: National Organization on Disability.

Stata. (2002). Stata 7.0 (intercooled) for Windows [computer program]. College Station (Texas): Author.