Node Matching

Most complete and partial network research designs have the aim of compiling a composite picture of the relationships among their study population. Capturing these complete network data allows for analyses of structural features of those relationships that extend beyond those available from ego network data. If the alter nominations were produced in any way other than fixed choice responses (and even in some cases when using those), researchers will need to actively link the multiple occurrences of the same node from multiple sources within the resulting data. This matching addresses both when a respondent is named as an alter among others’ tie reports, or when two separate interviewees identify the same alter, who may or may not themselves be among the respondents to the study.

In the simplest case, when the researchers know every node in the population, and nominations are provided that match those available on the population’s roster of names, matching nominations to other study participants or across nominations is relatively straightforward. However, we know that the identifiers that people use to describe their friends do not always match those available to the research team—e.g., people may use nicknames that differ across contexts (Helleringer et al. 2011; Young et al. 2016). To assist with overcoming this limitation, scholars have increasingly relied on multiple sources of readily known information about one’s peers (e.g., race, gender, contact information) as a means to improve the cross-validation necessary to accurately link nominations into a complete network representation (Dombrowski et al. 2015). Additionally, we know that in many study populations, even the most complete of population boundaries will not contain all of the important relationships among members of that population. If we take two of the commonly analyzed school-based samples with multiple waves of complete network data in Add Health as an example, for one (commonly referred to as “Jefferson”) the school boundary not only delineates the population of interest, but also contains the vast majority of friendship nominations made by students in the school. Contrastingly, in the other large school (commonly referred to as “Sunshine”), a large proportion of the nominated friends are outside of the school (Moody 2001).