| has_common_support {tidytreatment} | R Documentation |
Evaluate if observations have common support.
Description
The common support identification methods are based on Hill and Su (2013). Loosely speaker, an individuals treatment effect estimate has common support if the counter factual estimate is not too uncertain. The estimates are uncertain when the prediction is 'far away' from other observations. Removing estimates without common support can be beneficial for treat effect estimates.
Usage
has_common_support(model, treatment, method, cutoff, modeldata = NULL)
Arguments
model |
A supported Bayesian model fit that can provide fits and predictions. |
treatment |
A character string specifying the name of the treatment variable. |
method |
Method to use in determining common support. 'chisq', or 'sd'. |
cutoff |
Cutoff point to use for method. |
modeldata |
Manually provide model data for some models (e.g. from BART package) |
Details
Hill, Jennifer; Su, Yu-Sung. Ann. Appl. Stat. 7 (2013), no. 3, 1386–1420. doi:10.1214/13-AOAS630. https://projecteuclid.org/euclid.aoas/1380804800
Value
Tibble with a row for each observation and a column indicating whether common support exists.