Maciej Beręsewicz

Another #rstats package uploaded to CRAN! If you are linking data without identifiers and thinking about non-standard blocking methods to compare pairs, check out our {blocking} package. See the website and (hopefully) informative vignettes! ncn-foreigners.ue.poznan.pl/blocking/

I've just got my copy of your book, @bradytwest.bsky.social. Thanks for including the {nonprobsvy}!

Ok, I’ve checked the docs. Do you use these commands to impute values i.e. for mass imputation estimator? If for the IPW we do not support more then two sources (nonprobability and probability samples).

Any comments are more than welcome! :)

Unfortunately, don’t know stata so much so I need to check out the documentation and I’ll go back to you

If you are interested in the theory behind the {nonprobsvy} package, consider reading the working paper that we just uploaded on #arXiv arxiv.org/abs/2504.04255 #rstats

That’s next on our list. We're working on it, and we hope to finish the final version at the end of May or beginning of June.

Looking forward to your comments! @bradytwest.bsky.social @vincentab.bsky.social @rmkubinec.bsky.social @statstas.datascience.blue @saskiabrthlms.bsky.social @bschneidr.bsky.social @ivelasq3.bsky.social

As well as polishing the code, we also worked on the documentation. The package is based on the {survey} package (@tslumley.bsky.social), but you can also use the {srvyr} package (@bschneidr.bsky.social).

If you are interested in non-probability sampling, this is for you! #EconSky #StatsSky

main page

changes

output

documentation

New version of the {nonprobsvy} #rstats package is on its way to CRAN! We have made significant changes to the source code, its functionality and added new methods.

The output changed and was inspired by an excellent {WeightIt} pkg @noahgreifer.bsky.social ncn-foreigners.github.io/nonprobsvy

Thank you! I will check it out.

I’m familiar with {targets} but this is mainly for data analysis pipeline than my problem…but I will check it out and read the manuals/tutorials.

You are not far away from the truth as this the reason why I need such tool :)

Ok, it seems that the second package does what I was looking for! In general, I am interested in connections between functions I, or my colleagues, wrote i.e. say fun1 is used in fun2 and fun3 and fun2 in fun4. The goal is to understand how the functions effect each other and to manage changes.

#rstats developers, can anyone tell me if there is a tool (preferably an R package) that can count the number of functions used and their connections within my own package? @vincentab.bsky.social @zeileis.org

#rstats developers, do you know of a package that allows me to see how many times a given R/Rcpp function created by my package is used within my package (R/ and src/)? Is there a tool that allows me to visualise the relationship between funs? yes, I have my own funs that do it but I need a tool :)

Another bad day in the U.S. for any technical documents that talk about the "bias" in estimators...🙄

I agree with Stephen and Thomas. The sampling will be the same but you need to account for the correlation at the inference stage otherwise your standard errors may be to optimistic.

#rstats If you are interested in estimating population size, you may wish to use our package {singleRcapture} which has just been uploaded to CRAN (0.2.2). The package implements several methods (based on truncated distributions) and provides a user-friendly API. cran.r-project.org/package=sing...

singleRcapture: An R Package for Single-Source Capture-Recapture Models Population size estimation is a major challenge in official statistics, social sciences, and natural sciences. The problem can be tackled by applying capture-recapture methods, which vary depending on...

The package is suited for those of you who study hard-to-reach human or wildlife populations. For more see our paper: arxiv.org/abs/2411.11032

#rstats If you are interested in estimating population size, you may wish to use our package {singleRcapture} which has just been uploaded to CRAN (0.2.2). The package implements several methods (based on truncated distributions) and provides a user-friendly API. cran.r-project.org/package=sing...

A screenshot of a newsletter showing the title and abstract from the following article: https://isi-iass.org/home/wp-content/uploads/Survey_Statistician_2025_January_N91_06.pdf#page9

Today’s issue of The Survey Statistician includes an overview of two #rstats 📦’s (srvyr & svrep) for survey data analysis, facilitating Tidyverse workflows and flexible resampling methods such as the generalized bootstrap, building on ‘survey’ as a foundation #statsky

isi-iass.org/home/wp-cont...

Trump’s administration is stopping money for surveys it doesn't like?

We are open comments and suggestions!

@bradytwest.bsky.social We've finally finished a draft paper on the nonprobsvy package: github.com/ncn-foreigne.... Just bear in mind that the abstract and the section on classes
are incomplete. We're also doing a big update to the pkg, so expect breaking changes. We are open feedback :)

@mzloteanu.bsky.social If you want to know more about the package, have a look at the draft of the paper: github.com/ncn-foreigne... (just bear in mind that the abstract and the section on classes and s3meth are incomplete). We're also doing a big update to the pkg,so expect some breaking changes.

I am delighted to announce that our paper, "Quantile Balancing Inverse Probability Weighting for Non-probability Samples", has been accepted for publication in the Survey Methodology journal! You can find a preprint of the paper here: arxiv.org/abs/2403.09726

We use several nice algorithms: FAISS (META/ Facebook AI), Annoy/Voyager (Spotify) and many others to block records into small groups (for pairwise comparisons). The packages are in early development so comments are more than welcome!

#EconSky #AcademicSky are you dealing with linking data without identifiers (aka probabilistic RL, entity resolution)? Check our blocking packages in #rstats (ncn-foreigners.github.io/blocking) and #python (blockingpy.readthedocs.io) that use ANN to sig. reduce number of comparisons and time!

Dear all, I'm looking for good examples of how non-probability and probability surveys (or population data) have been used together (and data is available). Does anyone have any ideas? This is for testing our {nonprobsvy} package (ncn-foreigners.github.io/nonprobsvy) #rstats #EconSky #AcademicSky