Skip to contents

Standardize Variables (Z-Scores)

Source: R/standardize.R
std.Rd

Standardizes variables by centering on the mean and dividing by a measure of spread. Supports multiple standardization methods including robust alternatives.

When used on a grouped data frame (via dplyr::group_by()), standardization is performed within each group.

Usage

std(data, ..., method = "sd", weights = NULL, suffix = NULL, na.rm = TRUE)

Arguments

data

A data frame or numeric vector.

...

Variables to standardize (tidyselect). Only used when data is a data frame.

method

Standardization method:

"sd"

Standard z-score: (x - mean) / sd (default)

"2sd"

Gelman's 2-SD method: (x - mean) / (2 * sd). Useful in regression with binary predictors.

"mad"

Robust: (x - median) / mad. Resistant to outliers.

"gmd"

Gini's Mean Difference: (x - mean) / gmd. A robust alternative.

weights

Optional survey weights (unquoted column name or numeric vector). When provided, weighted mean and weighted SD are used for standardization. Only supported for methods "sd" and "2sd".

suffix

A character string appended to column names (e.g., "_z"). If NULL (default), the original columns are overwritten.

na.rm

Remove missing values before computing mean and SD? Default: TRUE.

Value

If data is a vector, a standardized numeric vector. If data is a data frame, the modified data frame (invisibly).

Details

Standardization Methods

  • sd (default): Standard z-transformation. Mean = 0, SD = 1.

  • 2sd: Divides by 2 standard deviations (Gelman, 2008). This makes standardized continuous predictors comparable to binary predictors in regression.

  • mad: Uses the Median Absolute Deviation instead of SD. Robust against outliers.

  • gmd: Uses Gini's Mean Difference — a robust spread measure based on all pairwise absolute differences.

Weighted Standardization

When weights is provided, the weighted mean and weighted standard deviation (using SPSS frequency weight formula) are used. This is only supported for methods "sd" and "2sd". The robust methods "mad" and "gmd" do not support weights.

Group-By Standardization

When data is grouped (via group_by()), standardization is performed separately within each group. This is useful for within-group comparisons. Weights are also subsetted per group.

See also

center() for mean-centering without scaling, pomps() for rescaling to 0-100

Other transform: center()

Examples

library(dplyr)
data(survey_data)

# Standard z-scores
data <- std(survey_data, age, income, suffix = "_z")

# Gelman 2-SD standardization (for regression)
data <- std(survey_data, income, age, method = "2sd",
            suffix = "_z")

# Robust standardization
data <- std(survey_data, income, method = "mad", suffix = "_z")

# Weighted standardization
data <- std(survey_data, income, age,
            weights = sampling_weight, suffix = "_z")

# Group-wise standardization
data <- survey_data %>%
  group_by(region) %>%
  std(income, suffix = "_z")