Comparative Methods for Delineating Social Networks

Abstract

By centering on the assumption that clear conceptualization precedes appropriate measurement, four methods for defining and enumerating personal networks are detailed. Global networks are defined in terms of the domain from which all other personal networks are derived. The three additional types, including significant other, exchange and interactive networks, are conceptually unique and largely non-overlapping in their memberships. The network types reviewed here do not exhaust all of the methods available for sampling personal networks, but they do represent methods with favorable psychometric properties and, most importantly, clear conceptual foundations.

Living Arrangements and Social Networks of Older Adults

“…In the original study, Kahn, Wethington, and Ingersoll-Dayton (1987) found an average network size of 8.9. In six other studies using the affective method, with samples not specifically of the aged, the mean network size varied from 3.0 to 6.7 (Milardo, 1992). The average number we found was lower than the number Van Tilburg (1992) found in his study of retiring men (mean 20.0), using 20 exchange questions.…”

“…In the study by Mugfold and Kendig (1986), the exchange procedure resulted in an average of 6.6 for Australian elderly. In five other studies, with samples not specifically of the aged, using the exchange method, mean network sizes between 10.1 and 21.8 were reported (Milardo, 1992).…”

Partner naming and forgetting: Recall of network members

“…Network name generators may ask about behavior-based relationships or about affective relationships (Campbell & Lee 1991;Milardo 1992;van der Poel 1993). In their study using a two-year time frame, Brewer and colleagues (1999) found higher levels of forgetting for drug injection partners than for sex partners.…”

Friendship, cliquishness, and the emergence of cooperation

“…First, previous research has ignored the potential effects of errors, which have been shown to be critically important in the canonical model (Bendor et al, 1996). Second, and more importantly, previous approaches assume that players maintain complexly integrated memories of past interactions with every other individual in a group, making the models implausible for human populations of more than 5-10 individuals (Dunbar, 1998;Milardo, 1992). Finally, the existing simulations have been uniformly limited to fewer than 60 players (often using only 20), restricting the generalizability of findings to more realistic populations and raising concerns about the effects of random drift (Leimar and Hammerstein, 2001).…”

“…Third, we restrict individuals' memories to one number for K different partners. For much of the analysis below, K is set to three, as this is cognitively plausible for many primates and is in a range observed for the number of close, supportive ties observed in humans (Dunbar, 1998;Milardo, 1992;Sugawara, 1984). We also briefly examine the effect of changing K, varying it from one to eight.…”