Two-Group Comparison Table for Continuous Variables

One row per variable comparing two groups, ready for a manuscript: group sizes, mean (SD) in each group, the difference with its confidence interval, and a p-value.

Usage

tbl_two_sample_compare(
  data,
  vars,
  group_col,
  test = c("welch", "student", "wilcoxon"),
  conf_level = 0.95,
  digits = 2,
  p_digits = 3
)

Arguments

data

A data frame.

vars

Character vector of numeric columns.

group_col

Name of the grouping column; it must have exactly two non-missing values.

test

`"welch"`, `"student"` or `"wilcoxon"`.

conf_level

Confidence level for the difference (default 0.95).

digits

Decimals for means, SDs and the difference.

p_digits

Decimals for the p-value (values below the last digit print as `<0.001`).

Value

A tibble with `Variable`, `N (<group 1>)`, `Mean (SD) <group 1>`, `N (<group 2>)`, `Mean (SD) <group 2>`, `Difference [95% CI]` and `p-value` (all text), and attributes `groups` (the two levels), `test` and `numeric` (the unformatted results as a tibble).

Difference and tests

The difference is **first level minus second level** of `group_col` (the factor order, or sorted values if it is not a factor). `"welch"` (default) is Welch's t-test (unequal variances) and `"student"` the pooled-variance t-test; both give the mean difference and its confidence interval. `"wilcoxon"` is the Wilcoxon rank-sum test; its difference is the Hodges-Lehmann estimate (a shift in location, not a difference in means) with its confidence interval. The p-value is kept when the estimate or interval cannot be computed (those are then blank), and it is computed even when both groups are constant at different values. Each group's `N` counts the non-missing values of that variable in that group. A variable with fewer than 2 values in a group, or constant in both, gets `NA` for the test.

See also

[table_two_by_two()], [compare_domain_datasets()]

Examples

tbl_two_sample_compare(mtcars, c("mpg", "hp", "wt"), "am")
#> # A tibble: 3 × 7
#>   Variable `N (0)` `Mean (SD) 0`  `N (1)` `Mean (SD) 1`  `Difference [95% CI]`
#>   <chr>    <chr>   <chr>          <chr>   <chr>          <chr>                
#> 1 mpg      19      17.15 (3.83)   13      24.39 (6.17)   -7.24 [-11.28, -3.21]
#> 2 hp       19      160.26 (53.91) 13      126.85 (84.06) 33.42 [-21.88, 88.71]
#> 3 wt       19      3.77 (0.78)    13      2.41 (0.62)    1.36 [0.85, 1.86]    
#> # ℹ 1 more variable: `p-value` <chr>