The Indiana Nonprofit Survey: Does What You Know Depend on How You Draw Your Sample?

We examine technical aspects of surveying nonprofits to determine whether inferences about characteristics of the sector depend on how samples are drawn. We use a stratified sampling design drawn from a comprehensive database of Indiana nonprofits to address two key methodological questions faced by nonprofit researchers: (1) how do major nonprofit sampling sources compare in terms of survey efficiency, and (2) how do major profile features of the nonprofit sector (e.g., service field, revenue, funding mix, age) vary by sampling source and by how broadly researchers define the sector. We find significant differences in the accuracy and completeness of alternative sampling sources and in response rates associated with them. In addition, nonprofits found on multiple listings tend to respond at higher rates and to be more formalized than those found on only one sampling source. We also find that sampling sources vary in the profiles they generate for the nonprofit sector. We conclude that researchers or policy analysts who wish to generalize their findings to the full nonprofit sector should seriously consider using the list of incorporated nonprofits as a sampling frame – it is among the most efficient and produces the fewest distortions in nonprofit profile features.