Click, collect, compare: evaluating a nonprobability web survey for family demography

Abstract

High-quality population data are expensive to collect and limited in substantive scope, constraining timely and innovative research. OBJECTIVEWe evaluate nonprobability data from the KINMATRIX survey by comparing it with external data on kinship structure and relationships to kin. METHODSKINMATRIX is a web-based quota survey of over 12,000 younger adults (aged 25-35) in 9 European countries and the United States.Respondents reported egocentric kinship networks, yielding more than 250,000 ego-kin dyads.We compare indicators in three domains -family network size (number of living kin), family relationships (emotional closeness, geographical distance, coresidence, and contact frequency), and family complexity (parental separation and number of half-siblings) -with estimates from probability-based surveys, demographic projections based on official vital statistics, and national registers. RESULTSKINMATRIX estimates of family network size tend to be lower than demographic projections based on official data, with smaller gaps for aunts and uncles and larger gaps for siblings and cousins.Cross-national differences in the number of parents, siblings, aunts, uncles, and cousins are broadly consistent across data sources, whereas patterns differ for grandparents.Regarding family relationships,