Understanding Mixed-Effects Models through Simulation

WORKSHOP OVERVIEW:
In this talk, we will delve into the increasingly popular mixed-effects models, frequently employed in experimental fields such as psycholinguistics where data are nested within participants and items. These models can often appear opaque, making it challenging to grasp what specific parameters signify and how they are reflected in the data.
Through the power of simulation, we aim to demystify these models by constructing and manipulating their parameters from scratch. This approach not only aids in a deeper comprehension of the models but also serves as an invaluable tool for conducting power analysis and refining experimental predictions.
Key areas we will explore include:

• Simulating data and see whether the model estimates mirror our expectations.
• Understanding random effects and how to simulate them.
• Exploring how simulation techniques can be used for power analysis.

This session is designed to be useful for a wide audience but of course familiarity with the lme4 package and R will help. If you would like to familiarize yourself with data simulation, you can have a look at my blog posts about this topic (https://julianquandt.com/categories/power-analysis/).

ONLINE PROGRAMME
The workshop will be held online, starting on 12th July at 12:00 pm (UK Summer time). Please join us via this Zoom link:
https://bham-ac-uk.zoom.us/j/84906376457?pwd=alZGSFY4SmJzVjRHb2FJMm9zS0NuQT09
Meeting ID: 849 0637 6457
Passcode: 163231

SPEAKER:
Julian Quandt, currently a postdoctoral researcher at WU Vienna University of Economics and Business, completed his PhD studies in Psychology at Radboud University, Nijmegen, focusing on metacognition in value-based decision making. During his doctoral studies, faced with the absence of a standard tool for power analysis for Mixed-Effects Models, he used data simulation instead. Through this process, he discovered the power of data simulation as a tool for learning and enhancing his own understanding of statistical models. Julian authored four blog posts on data simulation, covering both simple and complex statistical models. His intention was to share his intuition that simulation can serve as a valuable tool for comprehending these models, in addition to enabling us to conduct power analysis for (nearly) any statistical model.

CODE OF CONDUCT
Please take a minute to read our code of conduct - we are committed to creating a safe and harassment-free environment:
https://github.com/rladies/.github/blob/master/CODE_OF_CONDUCT.md
R-Ladies events are always free and open to anyone who wants to participate as long as they follow the code of conduct. We are pro-actively inclusive of queer, trans, and all minority identities, with additional sensitivity to intersectional identities. Our priority is to provide a safe community space for anyone identifying as a minority gender who is interested in and/or working with R.