pkgdown/mathjax-config.html

Skip to contents

Subclassification on Propensity Score

Source: R/matching_subclass.R
subclass_match.Rd

Divides units into K strata based on quantiles of the propensity score, then computes within-stratum weights for treatment effect estimation. This is a simple, transparent approach to propensity score adjustment that allows visual inspection of balance within each subclass.

Usage

subclass_match(
  formula = NULL,
  data = NULL,
  treatment = NULL,
  n_subclasses = 5L,
  ps = NULL,
  ps_model = NULL,
  estimand = "ATT"
)

Arguments

formula

Formula for propensity score model (e.g., treatment ~ age + income). Ignored if ps is provided.

data

Data frame containing all variables

treatment

Character, name of the binary treatment column (0/1)

n_subclasses

Integer, number of subclasses to create (default: 5). Cochran (1968) showed that 5 subclasses removes over 90\ a single covariate.

ps

Optional pre-computed numeric vector of propensity scores (one per row in data). If NULL, a logistic regression model is fit using formula.

ps_model

Optional pre-fitted glm object for propensity scores

estimand

Target estimand: "ATT" (default), "ATE", or "ATC"

Value

An S3 object of class c("subclass_result", "couplr_result") containing:

matched

Tibble with columns id, side, subclass, ps, weight

subclass_summary

Tibble with per-subclass statistics: counts, mean PS, and overlap status

info

List with n_subclasses, estimand, n_left, n_right, method, vars

Details

The algorithm:

  1. Estimate propensity scores via logistic regression (or use pre-computed scores)

  2. Divide the propensity score distribution into K quantile-based strata

  3. For each stratum, check overlap (both treated and control units present)

  4. Compute within-stratum weights based on the target estimand:

    • ATT: Treated units get weight 1; control units get weight n_treated_in_stratum / n_control_in_stratum

    • ATE: Both groups get weight proportional to stratum size relative to total sample

    • ATC: Control units get weight 1; treated units get weight n_control_in_stratum / n_treated_in_stratum

Examples

set.seed(42)
n <- 200
data <- data.frame(
  id = 1:n,
  age = rnorm(n, 40, 10),
  income = rnorm(n, 50000, 15000)
)
data$treatment <- rbinom(n, 1, plogis(-2 + 0.05 * data$age))
result <- subclass_match(treatment ~ age + income, data, treatment = "treatment")
print(result)