Apply a visibility rule to ego X alter X cell reports

Validates the rule's requires against the columns actually present, fits the rule, predicts a visibility for every row, and returns both the values and a provenance table describing how they were arrived at.

Usage

apply_visibility_rule(
  rule,
  esc.dat,
  ego.dat = NULL,
  sib.dat = NULL,
  ego.id = ".ego.id",
  frame.indicator = NULL,
  weights = NULL,
  ego.in.group = NULL,
  tie = NULL
)

Arguments

rule

a visibility_rule

esc.dat

ego X alter X cell reports, one row per report

ego.dat

ego-level data, used as the donor frame when a rule asks for donor = "egos". Needs a y.F column; if it has none and sib.dat is supplied, one is derived

sib.dat

long-form alter data, used to derive y.F when esc.dat or ego.dat lacks it

ego.id

name of the ego id column

frame.indicator

name of the 0/1 frame membership column. NULL (the default) takes it from tie, falling back to ".sib.in.F". Setting it here as well as on the tie is an error if the two disagree

weights

name of the column holding donor sampling weights

ego.in.group

is ego a member of the group ego reports about? Governs how y.F is derived. NULL (the default) takes it from tie, falling back to TRUE. Setting it here as well as on the tie is an error if the two disagree; the tie is where it belongs

tie

a tie_config() saying what kind of tie these reports are about. Required when rule assumes a tie structure — vis_from_clique() does — because applicability cannot be read off the data: on a tie that is not a clique the clique rule still returns a finite, plausible number, and it is wrong. A rule that makes no structural assumption, such as vis_from_donor(), needs no tie. When rule is a vis_coalesce() chain, tiers inapplicable to tie are dropped, and named in the returned provenance.

Value

a list with values (a tibble of vis, vis_weight, vis_rule, one row per row of esc.dat) and provenance (a vis_provenance tibble)