adjust() functions - remove eval(parse()) and switch to matrix multiplication for better efficiencymultibias_plot() to visualize sensitivity analysis resultsmultibias_adjust() the function now
incorporates uncertainty of the effect estimates from the validation data
by sampling from each estimate's mean and SE. Now, when using validation data,
the confidence intervals from multibias bootstrapped results will represent
two sources of uncertainty: random error and systematic error.bias_params
as an input for multibias_adjust()multibias_adjust() now has built in bootstrappingsummary() method to data_observedbias_params class to handle bias parameter inputs to
multibias_adjust()adjust() functions with a single multibias_adjust()
function. Users now specify the biases they want to adjust for in the
data_observed object. Bias adjustment formulas are now found in the
bias_params documentation.bias input of
data_observedevans data; now only used in vignettepkgdown web page: www.paulbrendel.com/multibiasdata_validation as an input for bias
adjustment:
adjust_om_sel.Radjust_uc_sel.Radjust_uc_em.Radjust_uc_om.Radjust_uc_em_sel.Radjust_uc_om_sel.Rdata_validation as an input for bias
adjustment:
adjust_em_om.Radjust_em_sel.Radjust_em.R and adjust_om.Rdata_observed and data_validationdata_observed to represent observed causal dataadjust functions now take data_observed as inputdata_validation to represent causal data that can be used
as validaiton data for bias adjustmentdata_validation as an input for bias
adjustment:
adjust_uc.Radjust_em.Radjust_om.Radjust_sel.Remc changed to emomc changed to omadjust_multinom_uc_em_sel into adjust_uc_em_seladjust_multinom_uc_om_sel into adjust_uc_om_seladjust_uc_em_sel.Radjust_uc_om_sel.Radjust_multinom_emc_omc into adjust_emc_omcadjust_multinom_uc_emc into adjust_uc_emcadjust_multinom_uc_omc into adjust_uc_omcadjust_emc_sel (exposure must be binary)adjust_omc_sel (outcome must be binary)adjust_uc_emc (exposure must be binary)adjust_uc_omc (outcome must be binary)adjust_multinom_uc_emc (exposure must be binary)adjust_multinom_uc_omc (outcome must be binary)df_omc_seldf_omc_sel_sourceadjust_ucadjust_emc (exposure must be binary)adjust_omc (outcome must be binary)adjust_seladjust_uc_seldf_uc_omcdf_uc_omc_sourcedf_uc_emcdf_uc_emc_sourcedf_uc and df_uc_source now both have continuous and
binary exposures and outcomes.adjust_uc_omc_sel &
adjust_multinom_uc_omc_sel.df_uc_omc_sel and df_uc_omc_sel_source.df_uc_seldf_uc_sel_sourceadjust_emc_omc & adjust_multinom_emc_omc.df_emc_omc and df_emc_omc_source.df_emc_seldf_emc_sel_sourceadjust_omc_sel.df_omc_sel and df_omc_sel_source.df_ucdf_uc_sourcedf_emcdf_emc_sourcedf_omcdf_omc_sourcedf_seldf_sel_sourceadjust_omc that appears when using three confoundersadjust_uc_omc and adjust_multinom_uc_omc.df_uc_omc and df_uc_omc_source.adjust_omc.df_emcdf_emc_sourcedf_omcdf_omc_sourcedf_seldf_sel_sourcedf_ucdf_uc_sourceadjust_sel had been weighing with the probability of selection
instead of the inverse probability of selection.