Power Calculations in R: How much data is enough?

Power Calculations in R: How much data is enough?
A panel meetup with Ethan Brown, Jianmei Wang, and Richard Webster

Sample size calculations and power analysis are fundamental for study design, and allow you to trust the findings from your data. Yet performing sample size estimates in R can be difficult as:

i) there is a sea of open source packages
ii) there is a steep learning curve

This panel discussion will cover topics and common questions on power analysis, as well as introduce the Task View page on the topic.

Agenda items:

Introductions: (2 mins)
Task View Intro: (3 mins)
Reproducibility research & power calculations (10 mins)
Practical approaches to performing power calculations (10mins)
Challenges (10 mins)
Power analysis Task View launch (3 mins)
Reflections (2 mins)
Audience Q&A (15 mins)

Example panel questions:

Has sample size estimation ever led to changes in one of your studies' models or data collection?
Do people rush power analysis and why?
How do you estimate the effect size for a future study?

There will be a slido link provided for audience questions as well.

Panelists:

Ethan Brown is Associate Director of the Research Methodology Consulting Center at the University of Minnesota. His interests lie at the intersection of improving methodological practice, statistical simulation, and studying people's understanding of statistics. As a consultant, he works with social science researchers to improve their methodology; as a software developer, he develops tools to make advanced methods more accessible; as a researcher, he studies how people understand uncertainty. He holds a Ph.D. in Educational Psychology from the University of Minnesota.

Jianmei Wang works in the statistical method group at Roche and is based in the UK. She has a PhD in mathematics and has been working in the Pharma industry as a statistical scientist for 19 years. Jianmei has worked in both preclinical research, providing statistical consultation, including experimental design, to research scientists, and in clinical development, designing clinical trials.

Richard Webster is the Data Science Team Lead at the Clinical Research Unit, within CHEO (a Canadian children’s hospital). As a biostatistician in the field of child health he is familiar with data from health administrative, survey and observational studies. His research interests are twofold, i) reducing the carbon footprint of healthcare delivery and ii) making science more impactful and cost-effective with improved statistical power analysis. Richard was awarded an R Consortium Infrastructure Grant to build a Task View page for sample size estimation.