Visibility from a group size the caller supplies

vis_from_clique() works out the group by counting ego's roster. Sometimes the group whose size sets an alter's visibility is not the roster the alter was reported on, and only the caller can say what it is. This rule takes that count from a column instead of deriving it.

Usage

vis_from_group_size(
  size.var,
  counts.ego = TRUE,
  subtract.self = TRUE,
  ego.in.group = TRUE,
  label = NULL
)

Arguments

size.var

name of the column giving the number of frame-population members in the alter's group

counts.ego

does size.var already count ego? Rosters that carry the respondent as a row do, in which case the column is the package's yprime.F rather than y.F. When FALSE and ego belongs to the group, one is added

subtract.self

subtract one for an alter who is themselves in the frame population, since they cannot report themselves. TRUE unless you have a reason

ego.in.group

is ego a member of the group being sized? Only consulted when counts.ego is FALSE. May be declared on the tie_config() instead

label

optional short name for this basis, used in provenance. Worth setting when comparing several bases, so the output says which produced which estimate

Value

a visibility_rule

When the group is not the roster

Three cases, all real:

Pooled ties

Cousins are reported through the maternal and the paternal side separately, but an alter's visibility depends on everyone who could have reported them — both sides, plus ego's own siblings. That pooled group is not any single roster.

Nested ties

"Pooled cousins, excluding ego's siblings" is the pooled group minus the sibship. A difference of two counts, which no roster holds directly.

Borrowed groups

A parent's visibility is the number of their children on the frame — a fact about the sibship, not about the parent roster the parent was reported on. This is the mechanism the parents case needs.

What this rule does and does not assume

It makes no structural assumption, so it needs no tie and is valid for any structure. That is not because it is safe in the way vis_from_donor() is safe — it is because the caller has taken responsibility for the part vis_from_clique() would have derived. The package can check the arithmetic; it cannot check that the column counts the right people.

What it still does for you is the frame split, which is easy to get wrong and matters more than it looks. An alter who is themselves in the frame population cannot report themselves, so their visibility is one lower than the group total. Since a death is always off-frame while exposure is a mixture, that asymmetry is the only route by which visibility reaches a rate. Set subtract.self = FALSE only deliberately.

See also

vis_from_clique(), which derives the group instead of taking it

Examples

# a roster that carries the respondent as a row, so the count includes ego
vis_from_group_size("n_in_cousinship_and_F")
#> <visibility_rule: group_size(n_in_cousinship_and_F)>
#>   requires:     n_in_cousinship_and_F, .sib.in.F
#>   is_estimated: FALSE  (fit once; frozen across bootstrap replicates)
#>   parameters:
#>     size.var = n_in_cousinship_and_F
#>     counts.ego = TRUE
#>     subtract.self = TRUE
#>     ego.in.group = TRUE
#>     label = NULL
#>   assumptions:
#>     - visibility is taken from the supplied column 'n_in_cousinship_and_F', which the caller has computed; the package does not check that it counts the right people
#>     - that column already counts ego
#>     - an alter in the frame population does not count themselves

# the same, named for provenance
vis_from_group_size("n_in_pooled_and_F", label = "pooled cousins")
#> <visibility_rule: pooled cousins>
#>   requires:     n_in_pooled_and_F, .sib.in.F
#>   is_estimated: FALSE  (fit once; frozen across bootstrap replicates)
#>   parameters:
#>     size.var = n_in_pooled_and_F
#>     counts.ego = TRUE
#>     subtract.self = TRUE
#>     ego.in.group = TRUE
#>     label = pooled cousins
#>   assumptions:
#>     - visibility is taken from the supplied column 'n_in_pooled_and_F', which the caller has computed; the package does not check that it counts the right people
#>     - that column already counts ego
#>     - an alter in the frame population does not count themselves