National Congregations Study Method
Also known as: Hypernetwork Congregation Sampling, Congregation Census Methodology, NCS Hypernetwork Method, Multiplicity Sampling of Congregations
The National Congregations Study (NCS) method solves a hard sampling problem: there is no complete list of all the congregations in a country, so they cannot be sampled directly. Mark Chaves and colleagues addressed this with hypernetwork (multiplicity) sampling - drawing a representative sample of individuals, asking those who attend services to name their congregation, and treating each named congregation as a sampled unit. Because a congregation is named in proportion to the number of people who attend it, this procedure automatically yields a sample of congregations with probability proportional to size, from which leaders are then interviewed. First fielded in 1998 and described in the 1999 Journal for the Scientific Study of Religion article, and repeated in later waves summarized in Chaves and Eagle's 2015 report, the NCS has become the standard way to produce nationally representative data on American congregations.
Key highlights
- Builds a representative sample of congregations without any pre-existing list of them.
- Automatically achieves probability-proportional-to-size selection through the naming process.
- Captures organization-level attributes that population surveys cannot, via key-informant interviews.
- Supports both congregation-weighted and attender-weighted inference from one design.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use the National Congregations Study method when you need nationally (or regionally) representative data on congregations as organizations but lack a complete sampling frame of them. It is the appropriate design whenever you can attach a congregation-naming module to a representative survey of individuals, and it is ideal for measuring congregation-level features - programs, worship style, staffing, social services, political activity - across the full population of congregations. It is less suitable when you must include congregations that no one in your individual sample attends (very small or declining congregations are underrepresented and the rarely attended are missed), when the population of interest is individuals rather than organizations, or when budget precludes the two-stage design of an individual survey followed by congregation interviews.
Strengths & limitations
- Builds a representative sample of congregations without any pre-existing list of them.
- Automatically achieves probability-proportional-to-size selection through the naming process.
- Captures organization-level attributes that population surveys cannot, via key-informant interviews.
- Supports both congregation-weighted and attender-weighted inference from one design.
- Congregations attended by no one in the individual sample have zero chance of selection, so very small ones are underrepresented.
- Quality depends entirely on the seed individual survey and on respondents accurately naming and locating their congregation.
- Key-informant reports may be biased or limited by the leader's knowledge of the congregation.
- The two-stage design is costly and slow, requiring both a population survey and a separate organizational survey.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What is hypernetwork sampling and why is it used for congregations?
Hypernetwork (or multiplicity) sampling reaches organizations indirectly by sampling individuals and asking them to name the organizations they belong to. It is used for congregations because no complete list of all congregations exists, so they cannot be sampled directly. By drawing a representative sample of people and collecting the congregations they attend, the NCS constructs a sample of congregations from a sample of individuals, inheriting the statistical properties of the well-understood individual sampling frame.
Why does the method automatically give probability proportional to size?
Because a congregation enters the sample by being named, and the chance that at least one of its attenders falls into the individual sample rises with the number of attenders. A congregation with a thousand worshippers is far more likely to be named than one with twenty, in direct proportion to size. This probability-proportional-to-size property is highly desirable in survey sampling, and here it arises for free from the naming mechanism rather than requiring a size-ranked list.
How do you get from the sample to statements about all congregations?
By weighting. Since large congregations are oversampled in proportion to size, analysts weight each congregation inversely by its number of attenders to estimate quantities about congregations as units - for example, what share of congregations run a particular program. Left unweighted, the same data describe the congregation of the average attender. The 2015 NCS report uses both: weighted estimates to characterize congregations and unweighted estimates to characterize the typical worshipper's experience.
Sources
- 1.Chaves, M., Konieczny, M. E., Beyerlein, K., & Barman, E. (1999). The National Congregations Study: Background, Methods, and Selected Results. Journal for the Scientific Study of Religion, 38(4), 458-476.
- 2.Chaves, M., & Eagle, A. (2015). Religious Congregations in 21st Century America. Durham, NC: Department of Sociology, Duke University (National Congregations Study).
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). National Congregations Study Method. ScholarGate. https://scholargate.app/sociology-of-religion/national-congregations-study