Participants performed a tasks with a 2x2 design. Each trial was either correct or incorrectly answers and reaction times were measured.
Each condition contained multiple trials.
In order to maintain the data provided by the multiple trials I'm running a linear mixed model rather than repeated measures. Additionally I would like to include correct/incorrect as a factor rather than covariate since I'm expecting a strong 3 way interaction.
The model looks like this:
Subjects: Subjects
Repeated: TrialNumber
Dependent Var: ReactionTime
Factors: Condition A, Condition B, Correctness
Fixed: Full factorial of all conditions (since It's the 3 way interaction I'm after).
Random: Empty
Am I running this correctly? It feels like I'm missing something, I'm just not sure what.
Have you considered models that are designed to account for the joint distribution of response time and response accuracy (e.g., Ratcliff's diffusion model, accumulator models, etc.)?
Hi Rik
Just a point of clarification was there an intervention (some change or treatment) through the repeated measures? Are you looking at between- or within-participant comparison?
Terry
Participants did not undergo any change or or treatment. The effect of interest is the within-subject effect resulting from the multiple conditions.
Since it's the behavioral data of an fMRI session the design was structured to optimize the MRI analysis, leaving a less than ideal setup for the behavioral data.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS