Aggregate Bootstrap Uncertainty (median central estimate, CI bounds, SE)

Aggregate Bootstrap Uncertainty (median central estimate, CI bounds, SE)

Usage

aggregate_uncertainty(
  boot_dt,
  prob_col,
  extra_keys = character(0),
  min_boot = NULL
)

Arguments

boot_dt

The combined data.table from all bootstrap iterations

prob_col

The name of the probability column (e.g. "p_quit")

extra_keys

Additional grouping columns beyond the standard demographic keys. The smoking trends are grouped by smk.state as well, because each age/year/sex/IMD cell carries three probabilities rather than one.

min_boot

The number of iterations each group must have. Groups with fewer are an error, not something to average over quietly.