tests/testthat/test-SE.R

In metaHelper: Transforms Statistical Measures Commonly Used for Meta-Analysis

test_that("SE from SD and n", {
  expect_equal(SE_from_SD(30.6, 80), 3.42, tolerance = 0.001)
  expect_equal(SE_from_SD(1.975, 6), 0.8063, tolerance = 0.001)
})


test_that("SE pooled from SD pooled and n", {
  expect_equal(SEp_from_SDp(76.7, 30, 65), 16.9, tolerance = 0.01)
})


dat <- data.frame(CI_low = c(0.31, 1.07, 1.07), CI_up = c(5.68, 26.5, 26.5), sig_level = c(0.05, 0.05, 0.05), two_tailed = c(T, T, F))
test_that("SMD.SE from OR", {
  expect_equal(SE.SMD_from_OR.CI(0.31, 5.68), 0.41, tolerance = 0.01)
  expect_equal(SE.SMD_from_OR.CI(1.07, 26.5), 0.452, tolerance = 0.01)
  expect_equal(SE.SMD_from_OR.CI(1.07, 26.5, two_tailed = F), 0.5378, tolerance = 0.01)
  expect_equal(SE.SMD_from_OR.CI(0.18, 1.89), 0.331, tolerance = 0.01)
  expect_equal(dplyr::mutate(dat,
                      SE = SE.SMD_from_OR.CI(CI_low, CI_up, sig_level, two_tailed)) %>%
                 dplyr::pull(SE),
               c(0.41, 0.452, 0.5378), tolerance = 0.01)
})


dat <- data.frame(SMD = c(0.6, 0.5, 0.8), n1 = c(NA, 23, 55), n2 = c(NA, 45, 60), method = c("cohen", "cohen", "hedges"))
test_that("SMD.SE from SMD and sample size", {
  expect_equal(SE.SMD_from_SMD(-0.2184, 140, 140, method = "cohen"), 0.119, tolerance = 0.01)
  expect_equal(SE.SMD_from_SMD(-0.0291, 60, 60, method = "cohen"), 0.182, tolerance = 0.01)
  expect_equal(SE.SMD_from_SMD(-0.92028841, 55, 49, method = "cohen"), 0.19644, tolerance = 0.1)
  expect_equal(SE.SMD_from_SMD(0.5970, 50, 50, method = "hedges"), 0.2028, tolerance = 0.001)
  expect_equal(SE.SMD_from_SMD(0.0, 500, 500, method = "hedges"), 0.06324, tolerance = 0.001)
  expect_error(SE.SMD_from_SMD(0.5970, 50, 50, method = "hedge"))
  expect_equal(SE.SMD_from_SMD(0.6, NA, NA, "cohen"), as.double(NA))
  expect_equal(SE.SMD_from_SMD(0.8, 55, 60, "hedges"), 0.1926, tolerance = 0.001)
  expect_equal(
    dat %>%
      dplyr::mutate(SE = SE.SMD_from_SMD(SMD, n1, n2, method)) %>%
      dplyr::pull(SE),
    c(as.double(NA), 0.2598, 0.1926), tolerance = 0.001)
})


dat <- data.frame(CI_up = c(10, 10, 10),
                  CI_low = c(5, 5, 5),
                  N1 = c(NA, 10, NA),
                  N2 = c(NA, 10, NA),
                  t_dist = c(F, T, T))
test_that("Pooled standard error from CI", {
  expect_equal(SEp_from_CIp(-0.37, 5, t_dist = F), 1.37, tolerance = 0.001)
  expect_equal(SEp_from_CIp(-0.666, 5, 10, 15, t_dist = T), 1.37, tolerance = 0.001)
  expect_equal(suppressWarnings(dplyr::mutate(dat, SE = SEp_from_CIp(CI_low, CI_up, N1, N2, t_dist = t_dist))) %>% dplyr::pull(SE),
               c(1.27551, 1.189, NA), tolerance = 0.001)
})


dat <- data.frame(TE = c(17, 1.204, 1, 1), p = c(0.032, 0.034, 1, 0.99999))
test_that("SE pooled from p and treatment effect", {
  expect_equal(SEp_from_TE.p(17, 0.032), 7.94, tolerance = 0.01)
  expect_equal(SEp_from_TE.p(1.204, 0.034), 0.569, tolerance = 0.01)
  expect_equal(dat %>%
                 dplyr::mutate(SE = SEp_from_TE.p(TE, p, two_tailed = TRUE)) %>%
                 dplyr::pull(SE), c(7.94, 0.569, 79788.45, 79788.45), tolerance = 0.01
               )
})