propensity_matching {CausalModels}

Propensity Matching

Description

'propensity_matching' uses either stratification or standardization to model an outcome conditional on the propensity scores. In stratification, the model will break the propensity scores into groups and output a glht model based off a contrast matrix which estimates the change in average causal effect within groups of propensity scores. In standardization, the model will output a standardization model that conditions on the propensity strata rather than the covariates. The model can also predict the expected outcome.

Usage

propensity_matching(
  data,
  f = NA,
  simple = pkg.env$simple,
  p.scores = NA,
  p.simple = pkg.env$simple,
  type = "strata",
  grp.width = 0.1,
  quant = TRUE,
  ...
)

Arguments

data

a data frame containing the variables in the model. This should be the same data used in init_params.

f

(optional) an object of class "formula" that overrides the default parameter

simple

a boolean indicator to build default formula with interactions. If true, interactions will be excluded. If false, interactions will be included. By default, simple is set to false.

p.scores

(optional) use calculated propensity scores for matching. Otherwise, propensity scores will be automatically modeled.

p.simple

a boolean indicator to build default formula with interactions for the propensity models. If true, interactions will be excluded. If false, interactions will be included. By default, simple is set to false.

type

a string representing the type of propensity model to be used. By default, the function will stratify. Standardization with propensity scores may also be used. The value given for type must be in c("strata", "stdm").

grp.width

a decimal value to specify the range to stratify the propensity scores. If option quant is set to true, this will represent the spread of percentiles. If false, it will represent the spread of raw values of propensity scores. Must be a decimal between 0 and 1. By default, this is set to 0.1. This option is ignored for standardization.

quant

a boolean indicator to specify the type of stratification. If true (default), the model will stratify by percentiles. If false, the scores will be grouped by a range of their raw values. This option is ignored for standardization.

...

additional arguments that may be passed to the underlying propensity_scores function.

Value

propensity_matching returns an object of class "propensity_matching"

The functions print, summary, and predict can be used to interact with the underlying glht or standardization model.

An object of class "propensity_matching" is a list containing the following:

call

the matched call.

formula

the formula used in the model.

model

either the underlying glht or standardization model.

p.scores

the estimated propensity scores

ATE

a data frame containing the ATE, SE, and 95% CI of the ATE.

ATE.summary

either a data frame containing the glht or standardization summary.

Examples

library(causaldata)
library(multcomp)

data(nhefs)
nhefs.nmv <- nhefs[which(!is.na(nhefs$wt82)), ]
nhefs.nmv$qsmk <- as.factor(nhefs.nmv$qsmk)

confounders <- c(
  "sex", "race", "age", "education", "smokeintensity",
  "smokeyrs", "exercise", "active", "wt71"
)

init_params(wt82_71, qsmk,
  covariates = confounders,
  data = nhefs.nmv
)

pm.model <- propensity_matching(nhefs.nmv)
pm.model$ATE.summary
summary(pm.model)
head(data.frame(preds = predict(pm.model)))