Estimate death rates from sibling history data

Usage

sibling_estimator(
  sib.dat,
  ego.id,
  sib.id = "sibid",
  sib.frame.indicator,
  sib.sex = "sex",
  cell.config,
  weights,
  boot.weights = NULL,
  return.boot = FALSE,
  visibility = networkreporting::vis_from_clique(),
  tie = networkreporting::tie_config("clique", name = "siblings"),
  discretize.exp = FALSE
)

Arguments

sib.dat

The long-form sibling history dataset (likely produced by prep_dhs_sib_histories)

ego.id

String with the name of the column of sib.dat that has the ID of the survey respondent

sib.id

String with the name of the column of sib.dat that has the sibling ID. Defaults to 'sibid', which is the column created by prep_dhs_sib_histories and prep_nrsim_sib_histories.

sib.frame.indicator

String with the name of the column in sib.dat containing a 0/1 coded variable indicating whether or not each sib is in the frame population

sib.sex

String with the name fo the column of sib.dat that has the sibling's sex

cell.config

An object containing the configuration of cells; see TODO for more information

weights

String with the name of the column of sib.dat that has the sampling weight

boot.weights

Optional dataframe with bootstrap resampled weights. See Details for more info.

return.boot

If TRUE, and if boot.weights is specified, then return each bootstrap estimate

visibility

A visibility rule saying how each reported sibling's visibility is derived. Defaults to networkreporting::vis_from_clique(), the exact rule this function has always applied, so the default changes nothing. See networkreporting::vis_from_donor() and networkreporting::vis_coalesce() for the approximating rules that non-clique ties need.

tie

What kind of tie the reports are about, as a networkreporting::tie_config(). Defaults to tie_config("clique", name = "siblings"), which is what siblings are, so the default changes nothing.

   **Set this if you are using this function for a tie that is not a
   clique.** Whether a roster is a clique is a fact about how it was
   built, not something the data reveals: applied to a roster that is
   not one, the clique rule still returns a plausible number. On a
   socsim roster pooling maternal and paternal cousins it overstates
   visibility by 1.089x for off-frame alters against 1.061x for
   on-frame ones, and since a death is always off-frame while exposure
   is a mixture, that differential biases the rate rather than
   cancelling out of it.

   Worth knowing which way the surprise runs: maternal cousins *alone*
   are a clique, since everyone sharing a maternal grandmother forms an
   equivalence class, and the rule is exact for them. It is the union of
   the two lines that is not.

discretize.exp

Boolean for whether or not expsoure should be discretized. Not yet implemented.

Value

a list with two entries: asdr.ind, individual visibility asdr estimates; and asdr.agg, aggregate visibility asdr estimates

Details

If you want estimated sampling variances, you can pass in a data frame boot.weights. boot.weights is assumed to have a column that is named whatever the ego.id is, and then a series of columns named boot_weight_1, ..., boot_weight_M.

Estimate death rates from sibling history data

Usage

Arguments

Value

Details

See also