Start Date

8-5-2013 12:30 PM

End Date

8-5-2013 2:00 PM

Subjects

Multivariate analysis, Social sciences -- Research -- Data processing, Interpersonal relations -- Analysis, Social sciences -- Network analysis

Description

Reconstructability analysis (RA) is proposed as a complementary method for evaluating social network-related phenomena. Longitudinal records of social behavior expression among members of a social network are commonly represented as a set of social network analysis (SNA) connections, but might also be usefully represented as a set of associations derived through RA methods. Reconstructability Analysis identifies individuals as being associated when their behavior patterns appear coordinated–a representation that is unavailable with standard SNA. To explore the potential usefulness of RA for analyzing social behaviors, simulated behavior patterns were evaluated with both SNA and RA, and the results were compared. Several RA data formats were tested, as RA cases can be defined by (a) individual or (b) synchronous behavior expression, or by (c) pairwise or (d) group level interactions. Associations derived with each data format were compared with the connections captured in a routine SNA adjacency matrix. Highest agreement between the two methods was found when cases were defined as instances of behavior expression at the group level. In addition, RA was shown to be able to derive triadic and higher-order associations among individuals, as well as sets of individuals whose behavior patterns were positively or negatively associated. Thus, RA appears to offer several capabilities to the study of social network-related phenomena that are not available with standard SNA techniques. Reconstructability analysis holds promise for advancing research on social behaviors, and can likely complement many of the efforts that are currently being made with SNA.

Persistent Identifier

http://archives.pdx.edu/ds/psu/9487

Share

COinS
 
May 8th, 12:30 PM May 8th, 2:00 PM

Exploring the Application of Reconstructability Analysis to Behavior Expression Data from a Social Network

Reconstructability analysis (RA) is proposed as a complementary method for evaluating social network-related phenomena. Longitudinal records of social behavior expression among members of a social network are commonly represented as a set of social network analysis (SNA) connections, but might also be usefully represented as a set of associations derived through RA methods. Reconstructability Analysis identifies individuals as being associated when their behavior patterns appear coordinated–a representation that is unavailable with standard SNA. To explore the potential usefulness of RA for analyzing social behaviors, simulated behavior patterns were evaluated with both SNA and RA, and the results were compared. Several RA data formats were tested, as RA cases can be defined by (a) individual or (b) synchronous behavior expression, or by (c) pairwise or (d) group level interactions. Associations derived with each data format were compared with the connections captured in a routine SNA adjacency matrix. Highest agreement between the two methods was found when cases were defined as instances of behavior expression at the group level. In addition, RA was shown to be able to derive triadic and higher-order associations among individuals, as well as sets of individuals whose behavior patterns were positively or negatively associated. Thus, RA appears to offer several capabilities to the study of social network-related phenomena that are not available with standard SNA techniques. Reconstructability analysis holds promise for advancing research on social behaviors, and can likely complement many of the efforts that are currently being made with SNA.