Bug Report: {p_miss_unweighted} (and possibly {p_miss}) in tbl_svysummary Displays Proportion Instead of Percentage

[ken1th]

Bug: {p_miss_unweighted} in tbl_svysummary Displays Proportion Instead of Percentage

Description

When using {p_miss_unweighted} in the missing_stat argument of tbl_svysummary, it incorrectly displays the proportion (on a 0–1 scale) instead of the percentage (on a 0–100 scale). This inconsistency can cause confusion and misinterpretation.

Reproducible Example

library(gtsummary)
library(survey)

# Example data with missing values
data <- data.frame(
  var1 = c(1, 2, NA, 4, NA, 6),
  var2 = c("A", "B", "A", "B", NA, NA),
  weights = c(1, 2, 1, 1, 2, 1)
)

# Create survey design
design <- svydesign(id = ~1, data = data, weights = ~weights)

# Create tbl_svysummary
tbl <- tbl_svysummary(
  design,
  include = c(var1, var2),
  statistic = list(all_categorical() ~ "{n} ({p}%)"),
  missing = "ifany",
  missing_stat = "{N_miss_unweighted} ({p_miss_unweighted}%)"
)

# Display table
tbl

Observed Output

For the missing rows (e.g., Unknown), the table shows:

• For var1: 2 (0.3%)
• For var2: 2 (0.3%)

The percentage value (0.3%) appears to be incorrectly derived as if it were a proportion (0.003) formatted directly as a percentage without multiplying by 100.

Expected Output

The correct missing statistics should display percentages:

• For var1: 2 (30.0%)
• For var2: 2 (30.0%)

Proposed Fix

Update the internal logic for {p_miss_unweighted} to ensure the proportion is multiplied by 100 before formatting. This will align the behavior of {p_miss_unweighted} with other placeholders like {p}, which correctly display percentages.

---

This bug leads to significant misinterpretation in summary tables. Thank you for addressing this issue!

[ddsjoberg]

Thanks for the report and the reprex! This should be fixed in the dev version of the package.

library(gtsummary)

# Example data with missing values
data <- data.frame(
  var1 = c(1, 2, NA, 4, NA, 6),
  var2 = c("A", "B", "A", "B", NA, NA),
  weights = c(1, 2, 1, 1, 2, 1)
)

# Create survey design
design <- survey::svydesign(id = ~1, data = data, weights = ~weights)

# Create tbl_svysummary
tbl_svysummary(
  design,
  include = c(var1, var2),
  statistic = list(all_categorical() ~ "{n} ({p}%)"),
  missing = "ifany",
  missing_stat = "{N_miss_unweighted} ({p_miss_unweighted}%)"
) |> 
  as_kable()

Characteristic	N = 8
var1
1	1 (20%)
2	2 (40%)
4	1 (20%)
6	1 (20%)
Unknown	2 (33%)
var2
A	2 (40%)
B	3 (60%)
Unknown	2 (33%)

^{Created on 2024-11-21 with reprex v2.1.1}