10th Biennial ACSPRI Social Science Methodology Conference in conjunction with the Australian Social Network Analysis Conference (ASNAC)

There is a significant body of egonet (ego-centric) network research using the signature methodology of name generators and follow-up (name interpreter) questions to elicit alter (others) and ego-alter relationship characteristics sometimes supplemented by the collection of a roster of alter-alter relations (as perceived by the ego). Egonet network structures only emerge with the alter-alter relations. Such a ‘structural’ egonet dataset is amenable to the array of analytic procedures used in ‘whole network’ (WN), sociocentric studies. A practical question for social researchers is what these WN measures mean in an egonet setting and whether they are useful enough to warrant collection. The methodological issue is how to use the WN output for comparative analysis of multiple egonets.
There is a significant body of personal network/ egonet literature, both substantive research (e.g. Bidart et al. Living in Networks, 2020) and methodological issues (e.g. Bellotti, Qualitative Networks, 2015; McCarty et al. Conducting Personal Network Research, 2019; Perry et al. Egocentric Network Analysis, 2018). Vacca (“Structure in personal networks”, Network Science, June 2020) directly addresses issues around structural egonet data, defining a typology of egonet structures.
We approached these questions by exploring WN measures on a databank of 15 egonet datasets from a purposive sampling of emerging adults collected in Brisbane in 2010-2013. We carried out our analyses on datasets with the ego node, and ego-alter relations included (EAA) and the ego node and ego-alter relations excluded leaving only alter-alter ties (AA). The WN analytics were done with UCINET and we used NetDraw to produce standardised visualisations of the 15 egonets.
With average path length in the AA datasets we found 10 cases where average distance was 1.00, i.e. all the clusters were complete components. We suggest such cluster closure can be an important feature of egonets. To work around this aspect of our data we took the average path length in the ego included (EAA) datasets normalized as the proportion of the average path length of ego’s star formation. We suggest this provides a meaningful, standardized metric of cohesion controlling for egonet size.
Component analysis is crucial for egonet analyses. The few component-based procedures in UCINET are highly influenced by the number of singletons. We recommend that the singletons in the egonets be a measure indicating the amount of interactive social activity, as does Vacca.
There are many possible configurations of different sized components within an egonet and this diversity will be evident in the degree distribution of the AA egonets. As with average path length we propose a metric that places the observed average degree between the extreme variation for the number of nodes in the data set and the minimum. This again produces standardised metric controlling for size.
We discuss how these metrics may enrich qualitative insights in personal network research. We also draw attention to the possibilities for regression analyses, QCA and Vacca’s typology.