0

I have a question about using linear mixed model effects in R using lmer.

I have a repeated measure experiment with 117 participants. They all perform a task with 5 categories (Prime_Names). The dependent variable is reaction times (Score). I want to compare those 5 categories with each other. There is a lot of missing data so I think a RM anova is not an option.

I have two questions:

  • Am I using the correct analysis if I do a linear mixed model effect analysis in R with lmer?
  • I am not sure if my model is completely correct, especially for the random effects. When do you use only "+ (1|Resp_ID)" and when do you use "+ (Prime_Name|Resp_ID)"

Two options:

Option 1:

model <- lmer(Score ~ Prime_Name + (1|Resp_ID), data=df)

Option 2:

model <- lmer(Score ~ Prime_Name + (Prime_Name|Resp_ID), data=df)

Any help will be appreciated.

Thank you

benson23
  • 16,369
  • 9
  • 19
  • 38
Dieuwertje van Dijk
  • 1
  • You can definitely use `lmer()` to analyze your data. The difference between the two models is that the first model specifies a unique intercept for each participant in the data. The second model has these intercepts, but it also adds a random slope, which allows the effect of `Prime_Name` to vary for each participant. The model will then report the effect of Prime_Name for the average participant. I think it may be helpful to read a conceptual introduction to mixed models http://www.bodowinter.com/uploads/1/2/9/3/129362560/bw_lme_tutorial1.pdf – sjp Nov 18 '20 at 19:35
  • As for the missing data, there are ways of dealing with that, but not all of them are commonly used in every field, so I would ask a PI what is common in your field. You may also be interested in this paper, which talks about some best-practices for analyzing reaction time data: https://dialnet.unirioja.es/servlet/articulo?codigo=3405162 – sjp Nov 18 '20 at 19:37

0 Answers0