In this episode of the Academic Crisis Line, Stacey Humphries and I gave a practical introduction to linear mixed-effects models. We talked about the background and key concepts about LMEMs, focused around 5 key questions that people often have when starting to encounter LMEMs for the first time.
- Why is a LMEM better than an ANOVA?
- What are fixed- and random-effects?
- What is the difference between a random intercept and a random slope?
- Should I include random slopes in my model?
- How can I tell if my model predictors significantly affect my dependent variable?
For answers to all these questions and more, check out the video! We also briefly walked through some practical aspects of running these analyses in R, but unfortunately the stream died at this point. We tried to re-record the parts that were lost, so this is annoyingly split over 3 videos. We hope it is useful nonetheless.
We mentioned a lot of different papers and links in the video which are listed here for your convenience, along with a few others you might find interesting:
You can find the visualisations of random intercepts and random slopes that we walked through here: http://mfviz.com/hierarchical-models/
Bodo Winter put together some fantastic tutorial for LMEMs:
Clark (1973) The language-as-fixed-effect-fallacy: http://www.sciencedirect.com/science/article/pii/S0022537173800143
Baayen et al. (2008) Mixed-effects modeling with crossed random effects for subjects and items http://jakewestfall.org/misc/BDB2008.pdf
Barr et al. (2013) Random effects structure for confirmatory hypothesis testing: Keep it maximal https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3881361/
Bates et al. (2015) Parsimonious mixed models https://arxiv.org/pdf/1506.04967.pdf
Matuschek et al. (2017) Balancing Type I error and power in linear mixed models http://www.sciencedirect.com/science/article/pii/S0749596X17300013
Baayen et al. (2017) The cave of shadows: Addressing the human factor with generalized additive mixed models http://www.sciencedirect.com/science/article/pii/S0749596X16302467
Luke (2017) Evaluating significance in linear mixed-effects models in R https://link.springer.com/article/10.3758/s13428-016-0809-y
Summary of Luke (2017) by Richard Morey: https://featuredcontent.psychonomic.org/putting-ps-into-lmer-mixed-model-regression-and-statistical-significance/
Brysbaert (2018) Power analysis and effect size in mixed effects models: A tutorial https://psyarxiv.com/fahxc
If you have any questions that we didn’t cover, please feel free to tweet us (@_SHumphries, @FranHartung, @Ph_Dial), or email us (hstace[AT]pennmedicine.upenn.edu, fhartung[AT]pennmedicine.upenn.edu).