------------------------------------------------------------------------------- help for onemode -------------------------------------------------------------------------------

One Mode Network Projection

onemode infilename , [ id alpha(real) matrix(string) edgelist(string) progress ]

Description

onemode creates valued, normalized, and signed one-mode network projections from two-mode data contained in infilename. (NOTE: This command does not affect the data in memory.)

Options id specifies that the first column of infilename contains actor IDs. alpha(real) specifies the alpha-level used for determining the statistical significance of edge weights in the construction of the signed projection. The default is 0.05. matrix(string) specifies the prefix of the comma-delimited output files to save the valued (string_valued.csv), normalized (string_norm.csv), and signed (string_signed.csv) projections as matrices. edgelist(string) specifies the name of the Stata data file to save the valued, normalized, and signed projections as edgelists. progress displays an (approximate) progress meter. [For this command to produce output, matrix, edgelist, or both must be specified.]

Input infilename must be a comma-delimited file in which rows represent "actors," and columns represent "events," such that cell Aij = 1 if actor i participated in event j, and otherwise is 0. The first column may optionally contain alphanumeric actor IDs.

Output Valued projection - * Aij = the number of events in which both actors i and j participate. * Aii = the number of events in which actor i participates.

Normalized projections - * Bonacich - Aij is normalized using the equation described by Bonacich (1972:179). * Pearson - Aij is normalized as the Pearson correlation between i's and j's attendance profiles. * In both cases - Aii = 0.

Signed projection - * Aij = 1 if actors i and j participate in statistically significantly more events together than would be expected if they each participated in the same total number of events randomly. * Aij = -1 if actors i and j participate in statistically significantly fewer events than would be expected if they each participated in the same total number of events randomly. * Aii = 0.

References Neal, Z. P. In press. Brute Force and Sorting Processes: Two Perspectives on World City Network Formation, Urban Studies. (CLICK FOR PDF) Neal, Z. P. 2012. A Parametric Method for Dichotomizing Unipartite Projections. Presented at the annual meeting of the International Network for Social Network Analysis (Sunbelt), March 13-18, Redondo Beach, CA. (CLICK FOR PDF) Bonacich, P. 1972. Technique for Analyzing Overlapping Memberships, Sociological Methodology 4: 176-185.

Author Zachary Neal Department of Sociology Michigan Sate University zpneal@msu.edu