Practical population inference with weights and survey design variables

This event is now fully booked. If you would like to be added to the waiting list please email the UK Data Service Events Team. 

Most users of survey microdata know they should use weights and, where possible, survey design information when producing estimates. However, users will find the lack of straightforward, practical guidelines or the technical nature of the available resources confusing. This free workshop is aiming to fill this gap.

This event is the second in a series of pilot workshops on this topic run by the UK Data Service. Its aim is to provide a practical introduction and clear guidance on the correct use of weights and survey design variables when producing population estimates from UKDS survey data. The session will combine plain English explanations of core concepts in survey design and statistical inference with their practical implications for real-world data analysis.

What this workshop will cover:

• Key concepts of survey design and their impact on statistical estimation.
• Software-related issues and common misunderstandings about weights.
• Balancing precision and time cost in real-world data analysis.
• Practical examples and exercises using R and Stata.

Objectives:

• Provide practical guidelines for survey estimation for applied researchers and data users.
• Introduce core statistical inference concepts in plain English.
• Link statistical concepts to real world survey data applications and examples.
• Demonstrate how to compute population totals, proportions, and confidence intervals using survey functions in R and Stata.
• Show how to compute domain (subgroup) estimates.
• Develop awareness of common mistakes when using weights and survey design variables in statistical software.

Target audience: This workshop is intended for researchers or data analysts with some prior experience of computing estimates with survey microdata. Basic familiarity with R and/or Stata is preferable.

Please note: This event will take place in person at the University of Manchester. Participants should bring their own laptop with R and/or Stata installed. Tea/coffee and a light lunch will be provided.