Skip to contents

Publication-Ready 2x2 Contingency Table with Tests and Risk Measures

Source: R/tbl_contingency_2x2.R
tbl_contingency_2x2.Rd

Cross-tabulates two binary variables as `n (percent)` cells with marginal totals, an exact and/or chi-squared test, the odds ratio and relative risk with confidence intervals, and the standard table footnote. For the institutional layout with a conditional-MLE odds ratio and mid-p handling use [table_two_by_two()].

Usage

tbl_contingency_2x2(
  data,
  row_var,
  col_var,
  percent = c("row", "col", "total"),
  test = c("fisher", "chisq", "both"),
  include_risk = TRUE,
  include_total = TRUE,
  digits = 1,
  p_digits = 3,
  note = TRUE,
  conf.level = 0.95
)

Arguments

data

A data frame.

row_var, col_var

Names of the row and column variables. Each must have exactly two observed levels after rows with a missing value in either are dropped.

percent

Denominator of the cell percentages: `"row"`, `"col"` or `"total"`.

test

`"fisher"`, `"chisq"` (Pearson, no continuity correction) or `"both"`.

include_risk

Add the odds ratio and relative risk.

include_total

Add the marginal totals.

digits

Decimals of the percentages.

p_digits

Decimals of the p-values (smaller values print as `<0.001`).

note

Build the footnote text.

conf.level

Confidence level of the risk intervals.

Value

A list with `table` (tibble of formatted cells), `counts` (the 2x2 `table`), `tests` (tibble: `test`, `p_value`, `p_formatted`), `risk` (tibble: `measure`, `estimate`, `lower`, `upper`; `NULL` if `include_risk = FALSE`), `note` (string or `NULL`) and `n` (rows used).

Risk measures

The first level of `row_var` is the exposed group and the first level of `col_var` is the outcome. The odds ratio is `(a * d) / (b * c)` and the relative risk is `(a / (a + b)) / (c / (c + d))`, with Wald (log-scale) intervals. If any cell is zero, 0.5 is added to every cell for these two measures only (the counts and percentages are unchanged). Factor levels set the order; otherwise levels are sorted.

See also

[table_two_by_two()], [tbl_table_one()]

Examples

res <- tbl_contingency_2x2(mtcars, "am", "vs", test = "both")
res$table
#> # A tibble: 3 × 4
#>   am    `0`        `1`        Total      
#>   <chr> <chr>      <chr>      <chr>      
#> 1 0     12 (63.2%) 7 (36.8%)  19 (100.0%)
#> 2 1     6 (46.2%)  7 (53.8%)  13 (100.0%)
#> 3 Total 18 (56.2%) 14 (43.8%) 32 (100.0%)
res$risk
#> # A tibble: 2 × 4
#>   measure       estimate lower upper
#>   <chr>            <dbl> <dbl> <dbl>
#> 1 Odds ratio        2    0.476  8.40
#> 2 Relative risk     1.37 0.693  2.70
res$note
#> [1] "n = 32. Cells are n (row %). Fisher's exact test, p = 0.473; Pearson's chi-squared test, p = 0.341."