Meta-analysis problem

I wonder if anyone could give me some advice about whether my idea for doing a meta analysis is OK.

I would like to combine the data from two psychology experiments and do a meta analysis. The two experiments have response time as the DV.

I was hoping to do a linear mixed effects regression, coding experiment with a binary dummy coded variable and look for interactions of predictors and experiment.

Experiment 1 used responses to visually presented words and has 7 orthogonal predictors.

Experiment 2 used responses to the same words but auditorily presented. It has the same 7 orthogonal predictors plus one extra orthogonal predictor - a measure relating to the length in milliseconds of the stimulus words (SoundFileLength).

The extra predictor in Experment 2 is problematic, especially as it is a robust predictor of response times.

I was wondering whether I could do the following...

With the Experiment 2 data - do a regression for each participant with response time as the DV and SoundFileLength as the IV and save the residuals as a measure of response time adjusted for effects of SoundFileLength.

Normalise RTs of each participant in Experiment 1.

Use these two measures as the DV in the meta analysis.

Would this be OK?


- Mike