Overview

Details

A small package for joint calibration of totals and quantiles for probability and non-probability surveys as well as causal inference based on observational data. The package combines the following approaches:

• Deville, J. C., and Särndal, C. E. (1992). Calibration estimators in survey sampling. Journal of the American statistical Association, 87(418), 376-382.
• Harms, T. and Duchesne, P. (2006). On calibration estimation for quantiles. Survey Methodology, 32(1), 37.
• Wu, C. (2005) Algorithms and R codes for the pseudo empirical likelihood method in survey sampling, Survey Methodology, 31(2), 239.
• Zhang, S., Han, P., and Wu, C. (2023) Calibration Techniques Encompassing Survey Sampling, Missing Data Analysis and Causal Inference, International Statistical Review 91, 165–192.

which allows to calibrate weights to known (or estimated) totals and quantiles jointly. As an backend for calibration sampling (sampling::calib), laeken (laeken::calibWeights), survey (survey::grake) or ebal (ebal::eb) package can be used. One can also apply empirical likelihood using codes from Wu (2005) with support of stats::constrOptim as used in Zhang, Han and Wu (2022).

backend	method	function called
sampling	c("raking", "linear", "logit", "truncated")	sampling::calib
laeken	c("raking", "linear", "logit")	laeken::calibWeights
survey	c("raking", "linear", "logit", "sinh")	survey::grake
ebal	eb	ebal::eb
base	el	R code and stats::constrOptim

Currently supports:

• calibration of quantiles,
• calibration of quantiles and totals,
• calibration using standard calibration (i.e. Deville and Särndal, 1992), empirical likelihood and entropy balancing method,
• covariate distribution entropy balancing for ATT and QTT (distributional entropy balancing; DEB) via the ebal package,
• covariate distribution balancing propensity score for ATE and QTE (distributional propensity score; DPS) via the CBPS package.

Further plans:

• generalized calibration via sampling::gencalib,
• calibration for Gini and other metrics,
• …

For details see:

• Beręsewicz M, and Szymkowiak, M. (2023). A note on joint calibration estimators for totals and quantiles, working paper (arxiv 2308.13281).
• Beręsewicz M (2023). Survey calibration for causal inference: a simple method to balance covariate distributions, working paper (arxiv 2310.11969).

Funding

Work on this package is supported by the the National Science Centre, OPUS 22 grant no. 2020/39/B/HS4/00941.

Installation

You can install CRAN version of the package using

install.packages("jointCalib")

You can install the development version of jointCalib from GitHub with:

# install.packages("remotes")
# remotes::install_github("ncn-foreigners/jointCalib")
install.packages("jointCalib", repos = c('https://ncn-foreigners.r-universe.dev', 'https://cloud.r-project.org'))