Visibility as reported by the respondent

Ego is asked, about each alter, how many frame-population members that alter is connected to. The answer is the alter's visibility, read straight off the questionnaire instead of derived from a roster or borrowed from donors.

Usage

vis_from_report(
  report.var,
  counts.ego = TRUE,
  counts.self = FALSE,
  on_missing = c("error", "na"),
  on_impossible = c("error", "floor", "na"),
  label = NULL
)

Arguments

report.var: name of the column holding ego's reported count for each alter
counts.ego: did the respondent count themselves? When FALSE, one is added back
counts.self: did the answer count the alter themselves? When TRUE, one is subtracted for an alter who is in the frame population, since an alter cannot report themselves
on_missing: what to do about an alter with no reported value: "error" or "na" (leave unresolved, so vis_coalesce() can try the next tier)
on_impossible: what to do about a reported visibility below one: "error", "floor" (raise it to one) or "na". An alter who was reported was, by construction, visible to at least one person, so a zero is a data problem rather than a small number
label: optional short name for provenance

Value

a visibility_rule

Why this is the honest option for some ties

Every other rule here recovers visibility from something the survey happened to collect for another purpose. vis_from_clique() exploits the tie's structure, vis_from_donor() and vis_from_model() substitute the respondents' own. All three are attempts to get at a quantity nobody was asked about.

For an "unbounded" tie there is no way round that. Neighbours and acquaintances have no bounded group to count, so there is no roster to derive from and nothing that makes the respondents a good stand-in for the alters. Asking is the only route to the quantity, and this rule is what turns the answer into an estimate.

It is a survey-design choice as much as an analysis one: it costs questionnaire time, and it can only be used if somebody decided in advance to ask. Where the question was asked, though, it beats an approximation derived from something else — and unlike the other rules, it puts the uncertainty somewhere visible, in reporting error rather than in an assumption.

Two things the answer may or may not include

The wording of the question decides both, and the package cannot tell from the numbers which was meant.

counts.ego — did the respondent count themselves among the alter's connections? "How many people like you does X know?" usually includes them; "how many other people like you" does not, and then one has to be added back.
counts.self — did the answer count the alter? A question phrased about a group ("how many people are in X's household?") does; one phrased about connections does not. When it does, this rule behaves like vis_from_group_size(), which is the better constructor to reach for.

Examples