From Belief To Commitment: the Community Service Activities and Finances of Religious Congregations in the United States, 1992

Data Archive > U.S. Surveys > Religious Groups > Congregations/Other Organizations > Others > Summary


The 1993 Edition "From Belief to Commitment" offers the following description of the project summary.

In 1992, INDEPENDENT SECTOR conducted a national survey of the activities and finances of religious congregations in order to provide information about religious organizations as part of a larger national survey of the activities and finances of private, nonprofit, charitable organizations in the United States. This survey was also designed to update a larger survey of the activities and finances of congregations conducted in 1987. The purpose of these surveys is to provide information about an important set of institutions and their impact on the quality of life in their communities and on individual giving and volunteering more generally. . . Specific objectives of the survey were to find answers to the following questions:
1. What are the size and membership composition of congregations?
2. Where are the congregations located by region of the country and by urban, suburban, or rural areas?
3. What are the congregations' programs in religion, education, health, human services, international activities, community development, civil rights, arts and culture, and the environment?
4. What are the total revenues of the congregations, and what proportions of these revenues come from individual giving or other sources of funds?
5. What are the expenditures of congregations? How much money do they spend on operations and programs, and how much do they use for other purposes?
6. How many people from the congregation volunteer to perform various activities, and how many hours per month do they volunteer?
7. How many programs, such as services to the elderly, do congregations operate directly, and how many programs do they support indirectly through contributions or through the voluntary service of members of the congregation?

With these questions we hoped to gain an initial understanding of the range of activities of congregations and their participation in the larger community. We also wanted to estimate nationally the size, scope, source and purpose of revenues and expenditures of congregations, and the ways these activities and expenditures relate to total philanthropy in the United States (p. xi-xii).

Data File
Cases: 727
Variables: 377
Weight Variable: WGT
Strata weight used to weight the sample to the universe (see Sampling Procedures for further discussion).
Data Collection
Date Collected: January - November 1992
Funded By
The Lilly Endowment, Inc.
Collection Procedures
The initial mailing was followed by an additional two mailings and telephone follow-up if necessary.
Sampling Procedures
The 1993 Edition "From Belief to Commitment" offers the following description of the sampling procedures.

This study of churches, congregations, synagogues, temples and other places of formal religious worship (herein after called churches) has as its universe the 257,648 such organizations which are listed in the Yellow Pages published by the nation's telephone companies. While this universe is nearly 100,000 fewer than the generally accepted number of 350,000 churches, those without telephone listings were unavailable for inclusion in the sample and therefore cannot be included in the study universe.

The actual study sample consisted of five (5) stratified random shadow samples of 1,003 each, drawn for NCCS by Survey Sampling, Inc., of Fairfield, Connecticut. The strata were the nine census regions divided into urban (SMSA) and non-urban (non-SMSA) areas, leading to 18 sampling strata. The organizations within each strata were sorted by the first three digits of their zip codes, and the 250th church was selected as the first entry for the sample one. The 249th was selected as the first church for sample two; the 251st for sample three; the 248th for sample four; and, the 252nd for sample five. The 500th church was then selected for the second entry for sample one, and the minus 1 (number 499), plus 1 (number 501), minus 2 (number 498), plus 2 (number 502) technique was used to select the second observations for the other four samples. This process continued until the end of the data was reached, resulting in the five samples used in the study. Each sample can be considered independent of the others, and each can be said to represent the universe.

Shadow sampling is a survey technique used in situations in which researchers anticipate either a low response rate from a primary sample and/or cannot afford the extensive follow-up required to bring a primary sample response rate up to acceptable levels (at least 60 percent response). Therefore, each church in sample one was matched with -- shadowed by -- four other churches, one from each of the four shadow samples. Under this technique, we were able to substitute the response from a shadow sample church for its non-respondent 'parent' church without harming the overall sampling framework. That is, if church 1 from sample one did not respond and church 1 from sample two did respond, then the survey from the sample two church was used as a substitute for the missing survey from the sample one church. This design is essentially sampling with replacement.

Our initial mailing to the 1,003 churches in sample one was followed by an additional two mailings and attempts at telephone follow-up. The second sample was then sent its initial survey request, with its own set of follow-up letters and calls. Then the third sample was mailed. Mailings to the fourth and fifth samples were only done for those churches out of the original 1,003 for which we had no response. A total, then, of 3,779 organizations were asked to participate in the study, with 728 completed surveys returned. We subsequently disregarded one survey because it represented an association of 35 congregations and ministries rather than an individual congregation. The remaining 727 returns were used as the frame of analysis. It should be noted that churches were compensated $50.00 for the time it took to complete the survey.

These completed surveys covered 615 of the original 1,003 sampling units. Using the shadow sampling technique, we then performed the following manipulations on the 727 surveys:
If a return was the only one for an original sampling unit, it was used with a unit weight of 1.00;
If two surveys were returned for a given sampling unit, each was used with a unit weight of .50;
If three surveys were returned from a given sampling unit, each was used with a unit weight of .33;
If four surveys were returned from a given sampling unit, each was used with a unit weight of .25;

This weighting technique allowed us to use all the returned surveys in the final research database. The sum of the weights over the 727 returned surveys is 615. Hence, using the shadow sampling technique, we obtained a 61.5 percent response rate for the original 1,003 in the primary sample.

A second weighting was then performed to weight the sample to the universe. This weight is called the strata weight because it takes the returns in each of the 18 sampling strata and weights them up to the strata total number of churches. Under perfect conditions, the strata weight would have been 250 because the sample took every 250th church from the population into the sample. The lowest strata weight was 96.547; the highest was 548.257. The resulting ratio of highest strata weight to lowest was 6:1, an acceptable range.

The results of these two weighting procedures provided the research data base which mirrored the population in terms of both the number of organizations in each sampling strata and the total number of organizations.

Finally, it should be noted that no attempt was made to make either the sample or the research database reflect the population according to denominational proportions. There were two reasons for this. First, it was decided that nothing in this study would be reported at the denominational level, as between-denomination comparisons were deemed inappropriate for a national study. Second, the universe data base used by Survey Sampling, Inc. had nearly 25 percent of the population categorized into a 'Denomination Unknown' category, a percentage that would not have allowed us to sample based on denomination even if we had wanted to. We did track the returned surveys by denomination, and we did correct the Survey Sampling, Inc. denomination when possible. But we did not weight the database by denomination (p. 111-112).
Principal Investigators
Virginia A. Hodgkinson and Independent Sector
Related Publications
Independent Sector. (1992). From Belief to Commitment: The Community Service Activities and Finances of Religious Congregations in the United States, 1993 Edition. Washington, DC: Independent Sector.