Skip to contents

True visibility, read off a known reporting network

true_visibility_from_network.Rd

Counts, for each alter, how many frame-population members are connected to them — which is exactly what a visibility rule is trying to estimate. Only usable where the whole network is known, so in practice: simulation.

Usage

true_visibility_from_network(
  census,
  reporter,
  alter,
  reporter.in.frame = NULL,
  dedup = TRUE
)

Arguments

census

a data frame with one row per (reporter, alter) tie in the true network

reporter

name of the column identifying the person who could report

alter

name of the column identifying the person reported about

reporter.in.frame

name of a 0/1 or logical column saying whether the reporter is in the frame population, or NULL if every row is already a frame-member reporter

dedup

drop repeated (reporter, alter) pairs before counting

Value

a tibble with one row per alter: the alter id and vis_true

Whose frame membership

reporter.in.frame names a column describing the reporter, not the alter. This is the distinction that gets lost when the definition is written inline, and it is not a technicality: an alter's visibility is the number of people who could report them, so it depends on who is in the frame population and connected to them. Filtering on the alter's own frame status instead yields a different quantity — one that is also a plausible number, and also wrong.

If the census has already been restricted to frame-member reporters, leave reporter.in.frame as NULL and every row counts.

Duplicate ties

A census may carry the same (reporter, alter) pair more than once — full siblings get linked once through each parent, so every sibling pair appears twice. Counting rows then doubles the visibility. dedup = TRUE removes repeated pairs and reports how many it dropped; the count is worth looking at, since a large one usually means the census is built differently from how you thought.

See also

visibility_accuracy(), which scores a rule against this

Examples

census <- data.frame(
  from = c("a", "b", "c", "a"),
  to   = c("x", "x", "x", "y"),
  from_in_frame = c(1, 1, 0, 1))
# x is connected to three people, but only two of them are in the frame
true_visibility_from_network(census, "from", "to", "from_in_frame")
#> # A tibble: 2 × 2
#>   to    vis_true
#>   <chr>    <int>
#> 1 x            2
#> 2 y            1