Can these ties be unioned?

Union is the third combination: treat several ties as one larger tie and estimate once from their pooled reports. Unlike compare_ties() and pool_ties() it cannot be done from finished estimates, because it changes what is estimated rather than combining answers — the reports have to be pooled and the visibility recomputed on the union.

Usage

ties_union_check(..., alter.key = NULL)

Arguments

...

report-level data frames, named by tie. Usually the esc.dat from each network_survival_estimator() result

alter.key

name of a column identifying an alter across ties

Value

a ties_union_check, invisibly; printed for its report

Details

It also needs something the package does not carry by default: an identity for alters that holds across ties. An alter reachable through two ties must be recognised as one alter, or their reports are counted twice and their visibility computed as though two separate people. Alter ids in this package are unique only within an ego, which is not enough.

This function checks whether a supplied key does the job, and reports what union would involve. It does not perform the union: once the key exists, the operation is to bind the report rows, deduplicate on the key, and estimate once with a tie_config() describing the union — which is usually not the structure of either part. Two cliques unioned are generally not a clique.

See also

pool_ties(), which averages estimates instead and needs no such key

Examples

if (FALSE) { # \dontrun{
ties_union_check(maternal = a$esc.dat, paternal = b$esc.dat,
                 alter.key = "person_id")
} # }