Calculate propensity score weights

Usage

wt_ate(.propensity, ...)

# S3 method for numeric
wt_ate(
  .propensity,
  .exposure,
  .sigma = NULL,
  exposure_type = c("auto", "binary", "categorical", "continuous"),
  .treated = NULL,
  .untreated = NULL,
  stabilize = FALSE,
  stabilization_score = NULL,
  ...
)

wt_att(.propensity, ...)

# S3 method for numeric
wt_att(
  .propensity,
  .exposure,
  exposure_type = c("auto", "binary", "categorical", "continuous"),
  .treated = NULL,
  .untreated = NULL,
  ...
)

wt_atu(.propensity, ...)

# S3 method for numeric
wt_atu(
  .propensity,
  .exposure,
  exposure_type = c("auto", "binary", "categorical", "continuous"),
  .treated = NULL,
  .untreated = NULL,
  ...
)

wt_atm(.propensity, ...)

# S3 method for numeric
wt_atm(
  .propensity,
  .exposure,
  exposure_type = c("auto", "binary", "categorical", "continuous"),
  .treated = NULL,
  .untreated = NULL,
  ...
)

wt_ato(.propensity, ...)

# S3 method for numeric
wt_ato(
  .propensity,
  .exposure,
  exposure_type = c("auto", "binary", "categorical", "continuous"),
  .treated = NULL,
  .untreated = NULL,
  ...
)

Arguments

.propensity: Either a vector of the predicted value of .exposure or a data.frame where each column is the predicted probability of a level of .exposure.
...: Passed to other functions Not currently used.
.exposure: The exposure for which .propensity is calculated.
.sigma: If exposure_type is continuous, a vector of observation-level standard errors passed to dnorm(). For an lm model this is influence(model)$sigma. For data frames produced by broom's augment(), this is the .sigma column.
exposure_type: The type of exposure. By default, automatically detected based on .exposure.
.treated: The treatment level of the exposure. Automatically detected by default.
.untreated: The control level of the exposure. Automatically detected by default.
stabilize: Logical. Stabilize the weights? By default, stabilizes with the mean of .exposure.
stabilization_score: if stabilize is TRUE, optionally include a score by which to stabilize the score, e.g. the predicted values from a regression model with no predictors.

Value

A vector of propensity score weights

Examples

propensity_scores <- c(.1, .3, .4, .3)
x <- c(0, 0, 1, 0)
wt_ate(propensity_scores, .exposure = x)
#> ℹ Treating `.exposure` as binary
#> [1] 1.111111 1.428571 2.500000 1.428571