Sample sizes for a whole effect grid

Applies cr_power() to every row of an effect grid produced by cr_effect_grid() and binds the sizing columns onto it, so the observed and the conservative sample size travel with the contrast they belong to.

Usage

cr_power_grid(
  effects,
  estimate = "cohens_d",
  ci_low = NULL,
  ci_high = NULL,
  available = NULL,
  power = 0.8,
  sig_level = 0.05,
  type = c("two.sample", "one.sample", "paired"),
  alternative = c("two.sided", "one.sided")
)

Arguments

effects

An effect grid from cr_effect_grid(), or any data frame with an estimate column and its two interval columns.

estimate

Name of the effect size column. The interval columns default to <estimate>_ci_low and <estimate>_ci_high.

ci_low, ci_high

Optional explicit names of the interval columns.

available

Units already available per arm, used to flag which contrasts are already large enough. Either a column name in effects or a numeric vector. Defaults to pmin(n_ref, n_cmp) when those columns are present.

power

Target power (default 0.8).

sig_level

Significance level (default 0.05).

type

Design passed to cr_power().

alternative

Alternative passed to cr_power().

Value

effects with the columns d_conservative, n_observed, n_conservative and basis appended, plus n_available and sufficient when availability is known.

See also

cr_power(), cr_effect_grid().

Other power: cr_conservative_effect(), cr_power(), cr_power_analysis()

Examples

set.seed(1)
units <- data.frame(
  compound = rep(c("CompoundA", "CompoundB"), each = 12),
  arm = rep(rep(c("reference", "interval_short"), each = 6), 2),
  log2_fc = c(stats::rnorm(6, 0, 0.3), stats::rnorm(6, -1.4, 0.3),
              stats::rnorm(6, 0, 0.3), stats::rnorm(6, -0.1, 0.3))
)
eff <- cr_effect_grid(units, "log2_fc", "arm", "reference",
                      by = "compound")
sizes <- cr_power_grid(eff)
sizes[, c("compound", "n_observed", "n_conservative", "basis")]
#> # A tibble: 2 × 4
#>   compound  n_observed n_conservative basis                               
#>   <chr>          <dbl>          <dbl> <chr>                               
#> 1 CompoundA          3              5 confidence bound nearer the null    
#> 2 CompoundB     864794             NA interval spans the null; not sizable