Two-sample t-test from summary statistics, across replicates

Vectorised equivalent of BSDA::tsum.test() with unequal variances, which the empirical Bayes power priors and the test-then-pool methods call once per replicate. x is the source study, whose summary statistics are fixed; y is the target study, which varies by replicate.

Usage

summary_t_test_p_value(
  mean_x,
  sd_x,
  n_x,
  mean_y,
  sd_y,
  n_y,
  mu = 0,
  alternative = "two.sided"
)

Arguments

mean_x, sd_x, n_x

Mean, standard deviation and per-arm sample size of the first sample.

mean_y, sd_y, n_y

Mean, standard deviation and per-arm sample size of the second sample. mean_y and sd_y may be vectors.

mu

Difference in means under the null hypothesis.

alternative

One of "two.sided", "less" or "greater".

Value

Vector of p-values.