5.1. Limitations
We have several limitations. The Lubben Social Network Scale does
not measure network density or activation, and is confined to the structure of social support networks comprised of friends and family,
resulting in an inability to understand the type of social network and extent of its support (Cattell, 2001). While the scales we chose to measure
our social resource constructs are limited, the scales we chose were designed a priori to measure the construct each purports to measure and
each scale has undergone extensive psychometric evaluation in other
studies. Further, these scales are commonly used in the health literature
and the Lubben Social Network Scale has been included in the PhenX
toolkit by the U.S. National Institutes of Health to help pragmatically facilitate common data elements and cross-sample and cross-study analyses. In the future, investigators may want to consider an approach
similar to factor analysis or item response theory to further disentangle
these constructs and methods of measurement. A framework like the
PhenX toolkit may be the most likely way to accomplish these analyses
(Hamilton et al., 2011). Additionally, our cross-sectional data were only
collected at one site and may not be generalizable to other populations
of people living with HIV. However, our single site design allowed us to
consider the unique geographic attributes of the sample and their implications for assessment and use of social resources in health research. Future, multi-site studies examining social resources should analyze data
both at individual sites and pooled across sites to further explore the
role of social geography in relation to health outcomes. Finally, our behavioral data were collected using self-report measures (e.g. diaries,
3-day visual analog scale of medication adherence). This may have led
to an overestimation of both physical activity and medication adherence
in our sample.