added verbose to dr estimator, readme updated

R/nonprob_dr.R

@@ -65,6 +65,9 @@ nonprob_dr <- function(selection,
   if (control_inference$vars_selection & control_inference$vars_combine) {
 
     ## estimate the mi
+    if (verbose) {
+      cat("MI variable selection in progress...\n")
+    }
     results_mi <- nonprob_mi(outcome = outcome,
                              data = data,
                              svydesign = svydesign,
@@ -84,6 +87,9 @@ nonprob_dr <- function(selection,
                              se = FALSE,
                              pop_size_fixed=pop_size_fixed)
 
+    if (verbose) {
+      cat("IPW variable selection in progress...\n")
+    }
     results_ipw <- nonprob_ipw(selection = selection,
                                target = reformulate(outcomes[[1]]),
                                data = data,
@@ -114,10 +120,19 @@ nonprob_dr <- function(selection,
     mi_coefs_sel <- lapply(results_mi$outcome, coef)
     dr_coefs_sel <- lapply(mi_coefs_sel, function(x) {
       mi_cols <- names(x[abs(x)>0])
-      combined <- union(ipw_coefs_sel, mi_cols)
+      combined <- sort(base::union(ipw_coefs_sel, mi_cols))
       combined[!grepl("Intercept", combined)]
     })
 
+    if (verbose) {
+      cat("IPW vars selected:", ipw_coefs_sel, "\n")
+      cat("MI vars selected:\n")
+      print(lapply(mi_coefs_sel, function(x) names(x[abs(x)>0])))
+      cat("DR combined vars:\n")
+      print(dr_coefs_sel)
+    }
+
+
     selection_vars <- all.vars(formula.tools::rhs(outcome))
     outcome_vars <- all.vars(formula.tools::rhs(selection))
     target_vars <- all.vars(formula.tools::lhs(outcome))
@@ -139,7 +154,7 @@ nonprob_dr <- function(selection,
     control_inference_$vars_selection <- FALSE
 
     for (o in outcomes$f) {
-
+      ## this is not saved in the output list
       results_ipw_combined[[o]] <- nonprob_ipw(data = X_nons,
                                                target = reformulate(o),
                                                selection =  reformulate(dr_coefs_sel[[o]]),

README.Rmd

@@ -45,7 +45,7 @@ population or probability sample is available:
 -   inverse probability weighting estimators with possible calibration
     constraints [@chen2020],
 -   mass imputation estimators based on nearest neighbours [@yang2021],
-    predictive mean matching and regression imputation [@kim2021],
+    predictive mean matching [@chlebicki2025], non-parametric [@chen2022nonparametric] and regression imputation [@kim2021],
 -   doubly robust estimators [@chen2020] with bias minimization [@yang2020].
 
 The package allows for:
@@ -56,7 +56,7 @@ The package allows for:
 -   estimation of variance using analytical and bootstrap approach (see
     @wu2023),
 -   integration with the `survey` and `srvyr` packages when probability sample is
-    available [@Lumley2004, @Lumley2023],
+    available [@Lumley2004; @Lumley2023; @srvyr2024],
 -   different links for selection (`logit`, `probit` and `cloglog`) and
     outcome (`gaussian`, `binomial` and `poisson`) variables.
 

README.md

@@ -27,12 +27,14 @@ for non-probability samples when auxiliary information from the
 population or probability sample is available:
 
 - inverse probability weighting estimators with possible calibration
-  constraints ([Chen, Li, and Wu 2020](#ref-chen2020)),
+  constraints ([Y. Chen, Li, and Wu 2020](#ref-chen2020)),
 - mass imputation estimators based on nearest neighbours ([Yang, Kim,
-  and Hwang 2021](#ref-yang2021)), predictive mean matching and
+  and Hwang 2021](#ref-yang2021)), predictive mean matching ([Chlebicki,
+  Chrostowski, and Beręsewicz 2025](#ref-chlebicki2025)), non-parametric
+  ([S. Chen, Yang, and Kim 2022](#ref-chen2022nonparametric)) and
   regression imputation ([Kim et al. 2021](#ref-kim2021)),
-- doubly robust estimators ([Chen, Li, and Wu 2020](#ref-chen2020)) with
-  bias minimization ([Yang, Kim, and Song 2020](#ref-yang2020)).
+- doubly robust estimators ([Y. Chen, Li, and Wu 2020](#ref-chen2020))
+  with bias minimization ([Yang, Kim, and Song 2020](#ref-yang2020)).
 
 The package allows for:
 
@@ -42,7 +44,9 @@ The package allows for:
 - estimation of variance using analytical and bootstrap approach (see Wu
   ([2023](#ref-wu2023))),
 - integration with the `survey` and `srvyr` packages when probability
-  sample is available Lumley ([2023](#ref-Lumley2023)),
+  sample is available ([Lumley 2004](#ref-Lumley2004),
+  [2023](#ref-Lumley2023); [Freedman Ellis and Schneider
+  2024](#ref-srvyr2024)),
 - different links for selection (`logit`, `probit` and `cloglog`) and
   outcome (`gaussian`, `binomial` and `poisson`) variables.
 
@@ -449,7 +453,7 @@ sample_prob
 #>     flag_srs == 1))
 ```
 
-or with `srvyr`
+or with the `srvyr` package
 
 ``` r
 sample_prob <- srvyr::as_survey_design(.data = subset(population, flag_srs == 1),
@@ -490,7 +494,7 @@ result_dr
 #>  - variance estimator: analytic
 #>  - population size fixed: false
 #>  - naive (uncorrected) estimator: 3.1817
-#>  - selected estimator: 2.95 (se=0.0414, ci=(2.8688, 3.0312))
+#>  - selected estimator: 2.95 (se=0.0415, ci=(2.8687, 3.0313))
 ```
 
 Mass imputation estimator
@@ -553,6 +557,14 @@ Work on this package is supported by the National Science Centre, OPUS
 <div id="refs" class="references csl-bib-body hanging-indent"
 entry-spacing="0">
 
+<div id="ref-chen2022nonparametric" class="csl-entry">
+
+Chen, Sixia, Shu Yang, and Jae Kwang Kim. 2022. “Nonparametric Mass
+Imputation for Data Integration.” *Journal of Survey Statistics and
+Methodology* 10 (1): 1–24.
+
+</div>
+
 <div id="ref-chen2020" class="csl-entry">
 
 Chen, Yilin, Pengfei Li, and Changbao Wu. 2020. “Doubly Robust Inference
@@ -562,6 +574,22 @@ Statistical Association* 115 (532): 2011–21.
 
 </div>
 
+<div id="ref-chlebicki2025" class="csl-entry">
+
+Chlebicki, Piotr, Łukasz Chrostowski, and Maciej Beręsewicz. 2025. “Data
+Integration of Non-Probability and Probability Samples with Predictive
+Mean Matching.” <https://arxiv.org/abs/2403.13750>.
+
+</div>
+
+<div id="ref-srvyr2024" class="csl-entry">
+
+Freedman Ellis, Greg, and Ben Schneider. 2024. *Srvyr: ’Dplyr’-Like
+Syntax for Summary Statistics of Survey Data*.
+<https://CRAN.R-project.org/package=srvyr>.
+
+</div>
+
 <div id="ref-kim2021" class="csl-entry">
 
 Kim, Jae Kwang, Seho Park, Yilin Chen, and Changbao Wu. 2021. “Combining

references.bib

@@ -87,3 +87,34 @@ @article{wu2023
 	url = {https://www150.statcan.gc.ca/n1/pub/12-001-x/2022002/article/00002-eng.htm},
 	langid = {en}
 }
+
+@Manual{srvyr2024,
+    title = {srvyr: 'dplyr'-Like Syntax for Summary Statistics of Survey Data},
+    author = {{Freedman Ellis}, Greg and Schneider, Ben },
+    year = {2024},
+    note = {R package version 1.3.0},
+    url = {https://CRAN.R-project.org/package=srvyr},
+  }
+
+
+@article{chen2022nonparametric,
+  title={Nonparametric mass imputation for data integration},
+  author={Chen, Sixia and Yang, Shu and Kim, Jae Kwang},
+  journal={Journal of survey statistics and methodology},
+  volume={10},
+  number={1},
+  pages={1--24},
+  year={2022},
+  publisher={Oxford University Press}
+}
+
+
+@misc{chlebicki2025,
+      title={Data integration of non-probability and probability samples with predictive mean matching},
+      author={Piotr Chlebicki and Łukasz Chrostowski and Maciej Beręsewicz},
+      year={2025},
+      eprint={2403.13750},
+      archivePrefix={arXiv},
+      primaryClass={stat.ME},
+      url={https://arxiv.org/abs/2403.13750},
+}