Standardized mean differences — smd • ecmeta.nsclc

Compute standardized mean differences (SMDs) between the treated and control subjects. Computed as \((\mu_t - \mu_c)/\sigma\) where \(\mu_t\) is the weighted mean among the treated subjects, \(\mu_c\) is the weighted mean among the control subjects, and \(\sigma\) is the standard deviation in the unweighted sample (among the treated subjects only since we are interested in the average treatment effect for the treated).

smd(object, ...)

# S3 method for grouped_psweight_mi
smd(object, x_vars = NULL)

Arguments

object	An object of the appropriate class.
x_vars	A character vector containing the covariates used to check for balance. These must be columns in `object`. If `NULL`, then variables that were used to fit the propensity score model(s) are used.

Value

A smd object that inherits from dplyr::tibble. The following columns are always returned:

var: A variable - either a covariate or the propensity score.
method: The propensity score method.
mean_control: The (weighted) mean among the controls.
mean_treat: The (weighted) mean among the treated.
sd_treat: The standard deviation among the treated.
smd: The standardized mean difference.

In addition, a column for the group included if object is grouped and the column imp is included if object was instantiated from a propensity score fit using multiply imputed datasets.