CBPS.fit determines the proper routine (what kind of treatment) and calls the approporiate function. It also pre- and post-processes the data
CBPS.fit(
treat,
X,
baselineX,
diffX,
ATT,
method,
iterations,
standardize,
twostep,
sample.weights = sample.weights,
...
)
A vector of treatment assignments. Binary or multi-valued treatments should be factors. Continuous treatments should be numeric.
A covariate matrix.
Similar to baseline.formula
, but in matrix form.
Similar to diff.formula
, but in matrix form.
Default is 1, which finds the average treatment effect on the treated interpreting the second level of the treatment factor as the treatment. Set to 2 to find the ATT interpreting the first level of the treatment factor as the treatment. Set to 0 to find the average treatment effect. For non-binary treatments, only the ATE is available.
Choose "over" to fit an over-identified model that combines the propensity score and covariate balancing conditions; choose "exact" to fit a model that only contains the covariate balancing conditions.
An optional parameter for the maximum number of iterations for the optimization. Default is 1000.
Default is TRUE
, which normalizes weights to sum
to 1 within each treatment group. For continuous treatments, normalizes
weights to sum up to 1 for the entire sample. Set to FALSE
to return
Horvitz-Thompson weights.
Default is TRUE
for a two-step estimator, which will
run substantially faster than continuous-updating. Set to FALSE
to
use the continuous-updating estimator described by Imai and Ratkovic (2014).
Survey sampling weights for the observations, if applicable. When left NULL, defaults to a sampling weight of 1 for each observation.
Other parameters to be passed through to optim()
.
CBPS.fit object