LDS Church Growth, Member Activity, and Convert Retention: Review and Analysis
III-04: Response Rates
The number of randomly selected individuals who could not be reached or who refused to participate should be recorded as well as participants, as non-respondents may differ from respondents in salient ways. Every effort should be made to contact randomly selected respondents until they are reached if they are not available on the first attempt. These steps may seem burdensome, but they are necessary to ensure the study integrity. Many published studies make no mention of non-respondents, and this is a serious methodological error which compromises study results. An 80% response rate is typically required for a study to be considered good quality.
The size of the "address unknown file" outside of the United States is a serious challenge for internal church research. In nations like Chile where 37% of nominal members cannot be located, there would simply be no way for internal church researchers to reach 80% of randomly-selected members. As virtually all lost address file members who have been on the list for greater than twelve months are inactive, the characteristics of a lost address file members are obviously different from those of a random sampling of members from congregational membership rolls. Ignoring lost members may result in findings which could be relevant to found members, but which cannot be extrapolated to address unknown file members. Population-based research then comes into play, but such research suffers from an even more severe problem of an unknown numerator, as discussed in the section of that name.