Decompose Constraint


1. EdgeList Data
×
EdgeList Data
Required.

A comma—separated values (.csv) file with 3 fields: ego_id, alter_id, zij. Column names must be exact. ego_id and alter_id can be any data type (e.g., int, string). zij must be numeric (e.g., int, float).

As per Burt, zij values are raw edge weights. If the graph is unweighted, set zij = 1 where ties exist. The edgelist need not include weight = 0 ties.

The program presumes “,” as a separator and ”.” as a decimal.

A sample edgelist .csv is available at: https://github.com/jchandlerj/generalConstraintDecomp. Files must follow this formatting to load properly.

 
2. Node Data (optional)
×

Node Data
Optional.
A comma—separated values (.csv) file with 2 fields: ego_id, conc. Column names must be exact.

conc values are subnetwork concentration ratios, corresponding to Burt’s Oj values . conc values must be passed as decimals (e.g., 0.65 would represent a 65% 4-firm concentration ratio).

The program presumes “,” as a separator and ”.” as a decimal.

 
3. Cull Self-Loops?
×

Cull Self-Loops?
Self-loops can affect constraint in two ways: as zij values summed in the denominator that normalizes all pij values (total activity) and as one of the pij values (pii).

If this option is enabled, self-loops are culled before computing total activity. If the option is not enabled, zij values for self-loops are included in the ‘total activity’ used to compute pij values. Including self-loop zijs increases the ‘total activity’ denominator and reduces pij values to ego’s alters.

In all cases, self-loop pijs are eliminated prior to computing constraint. Thus, including self-loops will tend to reduce constraint values.


 
4.