| Abstract: |
Measuring ego-centered networks is an especially time-expensive endeavor.
However, to be included in multi-thematic surveys, parsimony is needed instead
of following a maximizing strategy. Therefore, one pressing questions is,
whether and which type of parsimonious data collection produces what losses of
information. We address these questions by comparing three different, but
closely related operationalizations of ego-centered networks in three
different surveys: the German Socio-Economic Panel Study as a multi-thematic
large-scale survey, IDUN as a small study designed as a single-purpose
instrument to measure egocentered networks in great detail, and the
¿Minipanel¿ as something in between these two. Different setup parameters are
the number and type of name generators and descriptors of alteri and a
numerical limitation of naming ties. We look specifically whether there are
effects on the size of the networks, the composition of networks, and the
sociostructural differences with regard to availability of positive aspects of
social networks, especially different types of social support. Additionally,
we examine the differences between effects of network variables on perceived
satisfaction in the surveys. Among others, and above the highly expectable
differences in network size, we find an effect of different
operationalizations on the network composition. Compared to IDUN, GSOEP
produces a considerable overestimation of family ties and a considerable
underestimation of relations stemming from school, vocational training and
leisure time. We show that specific strategies of parsimony not only lead to
losses of information and distortions of network characteristics but can lead
in addition to different conclusions about the availability of social capital. |