Automation of sampling error production using R


Statistics New Zealand



Project description

Sampling errors are an important quality indicator in sample survey results. Stats NZ uses them to decide when results are significant enough to comment on in media releases of data, and they are also available to help data users understand the confidence to put in survey estimates.

Calculating and tabulating these can be a large task, especially when additional tables or variables are required, or a new survey is run. We have made a start on automating this process, and are looking for a suitable intern to complete this work.

The required product will be an R package that provides functions to calculate sampling error as well as customise the outputs and create and format reports. It needs to work with different data sources (different surveys) and be well modularised so it can be easily modified to meet future requirements.

Skills required

  • Experience with programming in R, with the ability to research and create packages and write code that is configurable and efficient
  • The ability to quickly learn about Stats NZ sample survey design and the calculation methods for sampling errors for different kinds of survey estimates   (study of maths and/or statistics beyond stage 1 is desirable).