Choosing a Sampling Strategy

Logo for series on Church Demographics and Congregational Websites

Special thanks to Dr. Marty Smith, a Senior Research Analyst at ELCA Research and Evaluation. His help was invaluable in deciding on a sampling strategy and collecting the congregational demographic data, and I really appreciated his help all the way through this phase of the project.

As I discussed in the last article, I needed a good mix of small, medium, and mega-sized congregations, and I needed them across rural, suburban, and urban settings. Dr. Smith and I decided to go for a stratified sample.1 To do this, he filtered the 10,000+ ELCA congregations to give just the ones with websites, then sorted by congregational setting (rural, suburban, etc.) and then within those categories by average worship attendance. Then he took every 23rd record all the way down the list to give me 101 congregations.

For each of those congregations, he provided “Congregation Form A” data for 2003-2007 (the most recent year available at the time of data collection in late 2008). This data includes, for each year:

  • When the congregation was founded
  • Members received, and how they were received (baptism, transfer, etc.)
  • Members removed, and how they were removed (death, transfer, etc.)
  • Average attendance
  • Size of the Sunday School
  • Demographics of the membership by race
  • URL of the church website

Here’s a summary of the congregation data:

  • The congregations sampled were an average of 100 years old this year.
  • Of the more than 10,000 congregations in the ELCA, only about 2,300 have websites registered with the churchwide office.
  • Of those, about 415 (18%) are broken or otherwise unusable.

Graph of ELCA Congregations with Websites, by Setting, with percentage broken

The disparity in resources between rural and suburban congregations is evident in this table, even without delving into the quality of the individual sites:

  • Congregations in rural farming communities make up 23% of our churches, but only one third of them have websites – and half of those are broken.
  • Small towns and rural areas have disproportionately more broken websites.
  • Medium- to large-sized cities are disproportionately better represented, having more websites and more functioning sites.

This afternoon, I’ll briefly discuss the factors that threw 18% of my congregations out of the study: broken websites.

Series Navigation<< Choosing the Demographic DataSeparating the Wheat from the Chaff >>
  1. A method that samples different subgroups in a population separately to make sure everyone is well represented. Read more on Wikipedia. []