template_match: Optimal Matching with Two Criteria.

Description

This function takes as arguments a dataset to be matched and a template, and outputs matched pairs that are closely matched, well balanced, and mimicking the user-supplied template in covariates' distributions of the given template.

Usage

template_match(
  template,
  X,
  Z,
  dataset,
  multiple = 1,
  lambda = 1,
  caliper_gscore = 1,
  k_gscore = NULL,
  penalty_gscore = Inf,
  caliper_pscore = 1,
  k_pscore = NULL,
  penalty_pscore = Inf
)

Value

This function returns a list of three objects: 1) feasible: 0/1 depending on the feasibility of the matching problem; 2) match_treated: a data frame of the matched treated units; 3) match_control: a data frame of the matched control units.

Arguments

template: A dataframe of template units.
X: A n-by-p matrix of covariates with column names.
Z: A length-n vector of treatment indicator.
dataset: Dataset to be matched.
multiple: Number of treated units matched to each template unit. Default is 1.
lambda: A tuning parameter controlling the trade-off between internal and external validity. A large lambda favors internal validity of the matched sample. A small lambda favors resemblance to the template.
caliper_gscore: Size of generalizability caliper.
k_gscore: Connect each template unit to k_gscore treated units closest in the generalizability score.
penalty_gscore: Penalty for violating the generalizability caliper. Set to Inf by default.
caliper_pscore: Size of propensity score caliper.
k_pscore: Connect each treated to k_pscore control units closest in the propensity score.
penalty_pscore: Penalty for violating the propensity score caliper. Set to Inf by default.

Details

Please refer to the vignette for reproducible examples.