Sample-size functions for target power — n_required

Determines the smallest sample size that achieves a requested target power for common hypothesis tests.

Usage

n_required_from_power(
  power_at_n,
  target_power,
  n_min = 1L,
  n_max = 1000000L,
  digits = 4,
  quiet = FALSE
)

n_required_z_mu(
  mu_a,
  mu0,
  sigma,
  alpha = 0.05,
  beta_target = 0.1,
  alternative = c("two.sided", "less", "greater"),
  n_min = 1L,
  n_max = 1000000L,
  n_ratio = 1,
  digits = 4,
  quiet = FALSE
)

n_required_t_mu(
  mu_a,
  mu0,
  sigma_true,
  alpha = 0.05,
  beta_target = 0.1,
  alternative = c("two.sided", "less", "greater"),
  n_min = 2L,
  n_max = 1000000L,
  paired = FALSE,
  method = c("welch", "pooled"),
  n_ratio = 1,
  digits = 4,
  quiet = FALSE
)

n_required_p_z(
  p_a,
  p0,
  alpha = 0.05,
  beta_target = 0.1,
  alternative = c("two.sided", "less", "greater"),
  n_min = 5L,
  n_max = 10000000L,
  pooled = NULL,
  continuity = FALSE,
  n_ratio = 1,
  digits = 4,
  quiet = FALSE
)

n_required_var_chisq(
  sigma_a,
  sigma0,
  alpha = 0.05,
  beta_target = 0.1,
  alternative = c("two.sided", "less", "greater"),
  n_min = 2L,
  n_max = 1000000L,
  digits = 4,
  quiet = FALSE
)

n_required_var_ratio_F(
  sigma_a,
  ratio0 = 1,
  alpha = 0.05,
  beta_target = 0.1,
  alternative = c("two.sided", "less", "greater"),
  n_min = 2L,
  n_max = 1000000L,
  n_ratio = 1,
  digits = 4,
  quiet = FALSE
)

Arguments

power_at_n: Function returning power for a proposed integer sample size.
target_power: Target power in (0, 1).
n_min: Minimum integer sample size considered by the search.
n_max: Maximum integer sample size considered by the search.
digits: Integer number of decimal places used only for printed output.
quiet: Logical; if TRUE, suppress printed output.
mu_a: True mean value or vector of true means under the alternative.
mu0: Null-hypothesis mean value or mean-difference value.
sigma: Known population standard deviation input.
alpha: Significance level in (0, 1).
beta_target: Target type II error in (0, 1).
alternative: Alternative hypothesis direction: "two.sided", "less", or "greater".
n_ratio: Allocation ratio n2 / n1 for two-sample designs.
sigma_true: True population standard deviation input used in t-test planning.
paired: Logical; for n_required_t_mu(), whether the supplied summaries are for paired differences.
method: Method for the two-sample sigma-unknown t-test planning calculation.
p_a: True proportion value or vector of true proportions under the alternative.
p0: Null-hypothesis proportion value or difference in proportions.
pooled: Logical; for two-sample proportion planning, whether to use pooled standard errors.
continuity: Logical; whether to apply the continuity correction where supported.
sigma_a: True standard deviation value or vector under the alternative.
sigma0: Null-hypothesis standard deviation for the chi-square variance test.
ratio0: Null-hypothesis variance ratio for the F test.

Details

These functions return classed result objects with unrounded stored values and optional printed summaries controlled by digits and quiet.

Functions

n_required_from_power(): Generic integer sample-size solver based on a power function.
n_required_z_mu(): Required sample size for z tests of means with known sigma.
n_required_t_mu(): Required sample size for t tests of means.
n_required_p_z(): Required sample size for z tests of proportions.
n_required_var_chisq(): Required sample size for one-sample chi-square tests of variance.
n_required_var_ratio_F(): Required sample size for two-sample F tests of a variance ratio.