tests/testthat/test-descriptives.R

In MethodsConsultants/tibbletest: Clean Descriptives Tables with Significance Tests by Group

treatment_tbl <- tibble::tribble(
       ~Variable,   ~Label,          ~candy,    ~`ice cream`,         ~`P Value`,
        "gender", "female",  "124 (48.25%)",  "110 (45.27%)",   0.56317088298012,
        "gender",   "male",  "133 (51.75%)",  "133 (54.73%)",   0.56317088298012,
           "age",       "", "42.26 (22.62)", "42.39 (21.61)",  0.946208619349777,
  "sugar_factor",       "",    "0.46 (0.3)",   "0.52 (0.29)", 0.0128504785023786,
     "happiness",  "happy",  "185 (76.76%)",   "181 (78.7%)",  0.694415226643762,
     "happiness",    "sad",   "56 (23.24%)",    "49 (21.3%)",  0.694415226643762,
         "happy",     "no",   "56 (23.24%)",    "49 (21.3%)",  0.855356296516553,
         "happy",    "yes",    "47 (19.5%)",   "44 (19.13%)",  0.855356296516553,
         "happy",    "Yes",  "138 (57.26%)",  "137 (59.57%)",  0.855356296516553
) %>%
  dplyr::arrange(Variable, Label)

cont_tbl <- tibble::tribble(
       ~Variable,          ~candy,    ~`ice cream`,         ~`P Value`,
           "age", "42.26 (22.62)", "42.39 (21.61)",  0.946208619349777,
  "sugar_factor",    "0.46 (0.3)",   "0.52 (0.29)", 0.0128504785023786
)

cat_tbl <- tibble::tribble(
    ~Variable,  ~Label,         ~candy,  ~`ice cream`,        ~`P Value`,
  "happiness", "happy", "185 (76.76%)", "181 (78.7%)", 0.694415226643762,
  "happiness",   "sad",  "56 (23.24%)",  "49 (21.3%)", 0.694415226643762
) %>%
  dplyr::arrange(Variable, Label)

simple_tbl <- tibble::tribble(
       ~Variable,   ~Label,    ~Statistics,
       "age",       "",  "42.3 (22.1)",
       "sugar_factor",       "",   "0.49 (0.3)",
        "gender", "female", "235 (46.91%)",
        "gender",   "male", "266 (53.09%)",
     "happiness",  "happy", "367 (77.75%)",
     "happiness",    "sad", "105 (22.25%)",
         "happy",     "no", "105 (22.25%)",
         "happy",    "yes",  "91 (19.28%)",
         "happy",    "Yes", "276 (58.47%)"
) %>%
  dplyr::arrange(Variable, Label)

test_that("descriptives produces correct output", {

  treat_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("gender", "age", "sugar_factor", "happiness", "happy")
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    treat_out %>% select(-`P Value`),
    treatment_tbl %>% select(-`P Value`)
  )

  expect_equal(
    treat_out %>% pull(`P Value`),
    treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(treat_out, "tbl_test")

  treat_out2 <- example_dat %>%
    dplyr::select(-treat2, -weight, -no_weight) %>%
    descriptives(treatment = "treat") %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    treat_out2 %>% select(-`P Value`),
    treatment_tbl %>% select(-`P Value`)
  )

  expect_equal(
    treat_out2 %>% pull(`P Value`),
    treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(treat_out2, "tbl_test")

  treat_out3 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(gender, age, sugar_factor, happiness, happy)
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    treat_out3 %>% select(-`P Value`),
    treatment_tbl %>% select(-`P Value`)
  )
  expect_equal(
    treat_out3 %>% pull(`P Value`),
    treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(treat_out3, "tbl_test")


  cat_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = "happiness"
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    cat_out %>% select(-`P Value`),
    cat_tbl %>% select(-`P Value`)
  )

  expect_equal(
    cat_out %>% pull(`P Value`),
    cat_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(cat_out, "tbl_test")

  cat_out2 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = happiness
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    cat_out2 %>% select(-`P Value`),
    cat_tbl %>% select(-`P Value`)
  )

  expect_equal(
    cat_out2 %>% pull(`P Value`),
    cat_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(cat_out2, "tbl_test")

  cont_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("age", "sugar_factor")
    )

  expect_equivalent(
    cont_out %>% select(-`P Value`),
    cont_tbl %>% select(-`P Value`)
  )

  expect_equal(
    cont_out %>% pull(`P Value`),
    cont_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(cont_out, "tbl_test")

  cont_out2 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(age, sugar_factor)
    )

  expect_equivalent(
    cont_out2 %>% select(-`P Value`),
    cont_tbl %>% select(-`P Value`)
  )

  expect_equal(
    cont_out2 %>% pull(`P Value`),
    cont_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )

  expect_s3_class(cont_out2, "tbl_test")

  simple_out <- example_dat %>%
    descriptives(variables = c("age", "sugar_factor", "gender", "happiness", "happy")) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(simple_out, simple_tbl)

  expect_s3_class(simple_out, "tbl_test")

  simple_out2 <- example_dat %>%
    descriptives(variables = c(age, sugar_factor, gender, happiness, happy)) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(simple_out2, simple_tbl)

  expect_s3_class(simple_out2, "tbl_test")

})

base_t_test <- t.test(
  age ~ treat,
  data = example_dat
)$p.value

base_chi <- chisq.test(
  example_dat$happiness,
  example_dat$treat
)$p.value

test_that("p-values are accurate", {

  anova_output <- example_dat %>%
    p_anova(var = "age", treatment = "treat", weight_var = "no_weight")
  expect_equal(base_t_test, anova_output, tolerance = 0.0001)

  chi_output <- example_dat %>%
    p_chi_fisher(var = "happiness", treatment = "treat", weight_var = "no_weight")
  expect_equal(base_chi, chi_output, tolerance = 0.0001)

})

no_weight_treatment_tbl <- tibble::tribble(
       ~Variable,   ~Label,          ~candy,    ~`ice cream`,         ~`P Value`,
        "gender", "female",        "48.25%",        "45.27%",   0.56317088298012,
        "gender",   "male",        "51.75%",        "54.73%",   0.56317088298012,
           "age",       "", "42.26 (22.62)", "42.39 (21.61)",  0.946208619349777,
  "sugar_factor",       "",    "0.46 (0.3)",   "0.52 (0.29)", 0.0128504785023786,
     "happiness",  "happy",        "76.76%",         "78.7%",  0.694415226643762,
     "happiness",    "sad",        "23.24%",         "21.3%",  0.694415226643762,
         "happy",     "no",        "23.24%",         "21.3%",  0.855356296516553,
         "happy",    "yes",         "19.5%",        "19.13%",  0.855356296516553,
         "happy",    "Yes",        "57.26%",        "59.57%",  0.855356296516553,
) %>%
  dplyr::arrange(Variable, Label)

weight_treatment_tbl <- tibble::tribble(
       ~Variable,   ~Label,          ~candy,    ~`ice cream`,          ~`P Value`,
        "gender", "female",        "48.65%",        "44.97%",   0.430901201183401,
        "gender",   "male",        "51.35%",        "55.03%",   0.430901201183401,
           "age",       "", "42.17 (22.55)", "42.43 (21.93)",   0.901907124081075,
  "sugar_factor",       "",    "0.45 (0.3)",   "0.52 (0.28)", 0.00920038384810278,
     "happiness",  "happy",         "77.1%",        "77.38%",    0.94558985790769,
     "happiness",    "sad",         "22.9%",        "22.62%",    0.94558985790769,
         "happy",     "no",         "22.9%",        "22.62%",   0.988678719678186,
         "happy",    "yes",        "19.22%",         "18.8%",   0.988678719678186,
         "happy",    "Yes",        "57.89%",        "58.58%",   0.988678719678186
) %>%
  dplyr::arrange(Variable, Label)

test_that("descriptives produces correct weighted tables", {

  no_weight_treat_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("gender", "age", "sugar_factor", "happiness", "happy"),
      weights = "no_weight"
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    no_weight_treat_out %>% select(-`P Value`),
    no_weight_treatment_tbl %>% select(-`P Value`)
  )
  expect_equivalent(
    no_weight_treat_out %>% pull(`P Value`),
    no_weight_treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(no_weight_treat_out, "tbl_test")

  no_weight_treat_out2 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(gender, age, sugar_factor, happiness, happy),
      weights = no_weight
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    no_weight_treat_out2 %>% select(-`P Value`),
    no_weight_treatment_tbl %>% select(-`P Value`)
  )
  expect_equivalent(
    no_weight_treat_out2 %>% pull(`P Value`),
    no_weight_treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(no_weight_treat_out2, "tbl_test")


  weight_treat_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("gender", "age", "sugar_factor", "happiness", "happy"),
      weights = "weight"
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    weight_treat_out %>% select(-`P Value`),
    weight_treatment_tbl %>% select(-`P Value`)
  )
  expect_equal(
    weight_treat_out %>% pull(`P Value`),
    weight_treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(weight_treat_out, "tbl_test")

  weight_treat_out2 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(gender, age, sugar_factor, happiness, happy),
      weights = weight
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    weight_treat_out2 %>% select(-`P Value`),
    weight_treatment_tbl %>% select(-`P Value`)
  )
  expect_equal(
    weight_treat_out2 %>% pull(`P Value`),
    weight_treatment_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(weight_treat_out2, "tbl_test")

})

nonparametric_tbl <- tibble::tribble(
       ~Variable, ~Label,              ~candy,        ~`ice cream`,         ~`P Value`,
         "happy",   "no",       "56 (23.24%)",        "49 (21.3%)",  0.855356296516553,
         "happy",  "yes",        "47 (19.5%)",       "44 (19.13%)",  0.855356296516553,
         "happy",  "Yes",      "138 (57.26%)",      "137 (59.57%)",  0.855356296516553,
           "age",     "",     "42.26 (22.62)",     "42.39 (21.61)",  0.946208619349777,
  "sugar_factor",     "", "0.44 [0.18, 0.72]", "0.52 [0.28, 0.76]", 0.0116654898090404
) %>%
  dplyr::arrange(Variable, Label)

nonparametric_weight_tbl <- tibble::tribble(
       ~Variable,              ~candy,       ~`ice cream`,          ~`P Value`,
           "age",       "42 [22, 62]",      "41 [26, 61]",    0.84408577110025,
  "sugar_factor", "0.44 [0.18, 0.72]", "0.51 [0.3, 0.76]", 0.00831771381361891
)

test_that("descriptives produces correct non-parametric tables", {

  nonparametric_out <- example_dat %>%
    select(treat, happy, age, sugar_factor) %>%
    descriptives(
      treatment = "treat",
      nonparametric = "sugar_factor"
    ) %>%
    dplyr::arrange(Variable, Label)

  nonparametric_out2 <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("happy", "age", "sugar_factor"),
      nonparametric = "sugar_factor"
    ) %>%
    dplyr::arrange(Variable, Label)

  nonparametric_out3 <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("happy", "age"),
      nonparametric = "sugar_factor"
    ) %>%
    dplyr::arrange(Variable, Label)

  nonparametric_out4 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(happy, age),
      nonparametric = sugar_factor
    ) %>%
    dplyr::arrange(Variable, Label)

  expect_equivalent(
    nonparametric_out %>% select(-`P Value`),
    nonparametric_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_out %>% pull(`P Value`),
    nonparametric_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_out, "tbl_test")

  expect_equivalent(
    nonparametric_out2 %>% select(-`P Value`),
    nonparametric_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_out2 %>% pull(`P Value`),
    nonparametric_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_out2, "tbl_test")

  expect_equivalent(
    nonparametric_out3 %>% select(-`P Value`),
    nonparametric_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_out3 %>% pull(`P Value`),
    nonparametric_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_out3, "tbl_test")

  expect_equivalent(
    nonparametric_out4 %>% select(-`P Value`),
    nonparametric_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_out4 %>% pull(`P Value`),
    nonparametric_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_out4, "tbl_test")

  nonparametric_weight_out <- example_dat %>%
    descriptives(
      treatment = "treat",
      variables = c("age", "sugar_factor"),
      nonparametric = c("age", "sugar_factor"),
      weights = "weight"
    )

  expect_equivalent(
    nonparametric_weight_out %>% select(-`P Value`),
    nonparametric_weight_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_weight_out %>% pull(`P Value`),
    nonparametric_weight_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_weight_out, "tbl_test")

  nonparametric_weight_out2 <- example_dat %>%
    descriptives(
      treatment = treat,
      variables = c(age, sugar_factor),
      nonparametric = c(age, sugar_factor),
      weights = weight
    )

  expect_equivalent(
    nonparametric_weight_out2 %>% select(-`P Value`),
    nonparametric_weight_tbl %>% select(-`P Value`)
  )
  expect_equal(
    nonparametric_weight_out2 %>% pull(`P Value`),
    nonparametric_weight_tbl %>% pull(`P Value`),
    tolerance = 0.0001
  )
  expect_s3_class(nonparametric_weight_out2, "tbl_test")

})