Survey Standard Error Estimation for Cumulated Estimates and their Differences in Complex Panel Designs

Calculate point estimates and their standard errors in complex household surveys using bootstrap replicates. Bootstrapping considers survey design with a rotating panel. A comprehensive description of the methodology can be found under < https://statistikat.github.io/surveysd/articles/methodology.html>.

surveysd

This is the development place for the R-package surveysd. The package can be used to estimate the standard deviation of estimates in complex surveys using bootstrap weights.

Installation

# Install release version from CRAN
install.packages("surveysd")

# Install development version from GitHub
devtools::install_github("statistikat/surveysd")

Concept

Bootstrapping has long been around and used widely to estimate confidence intervals and standard errors of point estimates. This package aims to combine all necessary steps for applying a calibrated bootstrapping procedure with custom estimating functions.

Workflow

A typical workflow with this package consists of three steps. To see these concepts in practice, please refer to the getting started vignette.

• Calibrated weights can be generated with the function ipf() using an iterative proportional updating algorithm.
• Bootstrap samples are drawn with rescaled bootstrapping in the function draw.bootstrap().
• These samples can then be calibrated with an iterative proportional updating algorithm using recalib().
• Finally, estimation functions can be applied over all bootstrap replicates with calc.stError().

Further reading

More information can be found on the github-pages site for surveysd.

• The methodology is covered in the methodology vignette.
• A more comprehensive documentation of calc.stError() is available in the error estimation vignette.