Mask small numerators for sensitive data privacy

This function masks small numerator values in sensitive data to protect privacy by preventing potential identification of individuals. It applies two key rules: (1) mask any numerators in the range (0, threshold), and (2) if only one numerator is masked, also mask all occurrences of the second smallest value to prevent back-calculation.

Usage

get_masked(vec, threshold = 3, mask_label = "masked")

Arguments

vec: A numeric vector of count data (numerators) to be masked.
threshold: Numeric. The threshold value for masking. Values greater than 0 and less than threshold will be masked. Default is 3.
mask_label: Character string. The message to replace masked values with. Default is "masked".

Value

A character vector with small values replaced by the mask message. If no values need masking, returns the original vector converted to character.

Details

The function implements a two-step masking process:

Initial masking: Any values in the range (0, threshold) are masked
Secondary masking: If only one value was masked in step 1, all occurrences of the second smallest value (excluding 0) are also masked to prevent back-calculation of the original masked value

This approach prevents scenarios where having only one masked value would allow users to calculate the masked value by subtraction from known totals.

Note

Zero values are never masked as they typically represent legitimate absence of cases. The function always returns a character vector to facilitate merging with other masked datasets that may contain non-numeric mask messages.

Examples

# Basic usage - mask values less than 3
example_vec1 <- c(4, 6, 2, 4, 0)
get_masked(example_vec1)  # Returns: "4" "6" "masked" "4" "0"
#> [1] "masked" "6"      "masked" "masked" "0"     

# With different threshold
example_vec2 <- c(4, 6, 2, 4, 1)
get_masked(example_vec2, threshold = 3)  # Masks values 1 and 2, plus second smallest (4)
#> [1] "4"      "6"      "masked" "4"      "masked"

# Custom mask message
example_vec3 <- c(4, 6, 2, 4, 3)
get_masked(example_vec3, threshold = 3, mask_label = "*")
#> [1] "4" "6" "*" "4" "*"

# Demonstrating secondary masking rule
example_vec4 <- c(4, 6, 3, 4, 3)
get_masked(example_vec4, threshold = 3)  # Only value 2 would be masked initially
#> [1] "4" "6" "3" "4" "3"

# Higher threshold
example_vec5 <- c(4, 6, 3, 4, 3)
get_masked(example_vec5, threshold = 5, mask_label = "*")  # Masks 3s and 4s
#> [1] "*" "6" "*" "*" "*"

# No masking needed
example_vec6 <- c(5, 6, 7, 8, 0)
get_masked(example_vec6)  # Returns original as characters: "5" "6" "7" "8" "0"
#> [1] "5" "6" "7" "8" "0"

# Realistic workflow example
if (FALSE) { # \dontrun{
library(dplyr)
library(tibble)

# Create example demographic data
set.seed(1234)
dat <- tibble(
  unit_number = sample(8, size = 100, replace = TRUE),
  gender = sample(c("boy", "girl", "non-binary", "prefer not to say"),
                  size = 100, replace = TRUE)
)

# Apply masking to grouped counts
results <- dat |>
  group_by(unit_number, gender) |>
  summarise(numerator = n(), .groups = "drop") |>
  mutate(
    proportion = sprintf("%.1f%%", 100 * numerator / sum(numerator)),
    numerator_masked = get_masked(numerator, threshold = 3),
    proportion_masked = ifelse(numerator_masked == 'masked', 'masked', proportion)
  )
} # }