The psychometric properties of Aggregated Relational Data and NSUM-estimated network size

Despite the widespread use of aggregated relational data (ARD, i.e., responses to questions of the form “How many people do you know with [trait X]?”) for estimating the size and composition of social networks, their psychometric properties are not well established. This paper addresses this gap by assessing (1) the test-retest reliability of a large set of ARD questions and of individuals’ NSUM-estimated acquaintanceship network size based on their responses to these questions and (2) the convergent validity of the estimated network size. The test was conducted with a heterogeneous quota sample of 50 people residing in Barcelona (Spain) in 2023. Respondents were interviewed twice over 10–15 days, responding to a series of A.R.D. questions on both occasions. We also performed qualitative debriefing to understand how individuals interpreted and responded to the questions. Test-retest reliability showed that respondents’ ranking was excellently reproduced, but the precise values were not, with respondents giving lower answers on the retest than the test. Especially the network size of people with large networks fluctuated considerably. We found moderate correlations for convergent validity. We discuss the implications of these and other results for ARD item choice and using NSUM instruments.