Repository files navigation

Propensity score methods are broadly employed with observational data asa tool to achieve covariate balance, but how to implement them incomplex surveys is less studied – in particular, when the survey weightsdepend on the group variable under comparison.

In this package, we focus on the specific case when sample selectiondepends the comparison groups of interest. We implement identificationformulas to properly estimate theaverage controlled difference (ACD),or under stronger assumptions, thepopulation average treatment effect(ATE) in outcomes between groups, with appropriate weighting for bothcovariate imbalance and generalizability.

This packages also contains the code necessary to reproduce themotivating data analysis in“What’s the weight? Estimating controlledoutcome differences in complex surveys for health disparitiesresearch.” This analysis focuses on data from the National Health andNutrition Examination Survey (NHANES), investigating the interplay ofrace and social determinants of health when our interest lies inestimating racial differences in mean telomere length.

You can install the development version of svycdiff like so:

#--- CRAN Versioninstall.packages("svycdiff")#--- Development Version# install.packages("devtools")devtools::install_github("salernos/svycdiff")

This is a basic example usage via simulated data:

library(svycdiff)N<-1000dat<- simdat(N)S<- rbinom(N,1,dat$pS)samp<-dat[S==1,]y_mod<-Y~A*X1a_mod<-A~X1s_mod<-pS~A+X1fit<- svycdiff(samp,"DR",a_mod,s_mod,y_mod,"gaussian")fit

Once you havesvycdiff installed, you can type

vignette("svycdiff")

inR to bring up a tutorial onsvycdiff and how to use it. To accessthe vignettes in the developer version, please install the package with

devtools::install_github("salernos/svycdiff",build_vignettes=TRUE)

For technical details on the method, see please refer to Salerno etal. (2024+)“What’s the weight? Estimating controlled outcomedifferences in complex surveys for health disparities research.” Toreproduce the analysis results for the main paper, seeinst/nhanes.Rmd. For questions and comments, please contact StephenSalerno (ssalerno@fredhutch.org).

sessionInfo()#> R version 4.4.1 (2024-06-14 ucrt)#> Platform: x86_64-w64-mingw32/x64#> Running under: Windows 11 x64 (build 22631)#>#> Matrix products: default#>#>#> locale:#> [1] LC_COLLATE=English_United States.utf8#> [2] LC_CTYPE=English_United States.utf8#> [3] LC_MONETARY=English_United States.utf8#> [4] LC_NUMERIC=C#> [5] LC_TIME=English_United States.utf8#>#> time zone: America/New_York#> tzcode source: internal#>#> attached base packages:#> [1] stats     graphics  grDevices utils     datasets  methods   base#>#> loaded via a namespace (and not attached):#>  [1] compiler_4.4.1    fastmap_1.2.0     cli_3.6.3         tools_4.4.1#>  [5] htmltools_0.5.8.1 rstudioapi_0.17.1 yaml_2.3.10       rmarkdown_2.29#>  [9] knitr_1.50        xfun_0.52         digest_0.6.37     rlang_1.1.4#> [13] evaluate_1.0.3