0
$\begingroup$

I have a rather simple design, but I am unsure how best to (properly) handle the data analysis.

I have an experiment with two between-subjects factors and one within-subjects factor. I will use, as an example Gender(Male v Female) and Training(Noob vs Pro) for the between, and Shapes(Red and Blue) for the within.

My problem: the Shapes factor has unequal trials based on the Training condition, where Noob sees 5 Red and 6 Blue, while Pro sees 6 Red and 5 Blue trials. What I care about is "accuracy in correctly detecting the colour of a the shape".

Now, I have no idea how to analyze this data. If I just input the data in long form, with Trial as a factor, would a MEM in R analyze the data based on AVERAGES (due to the difference in trial number) or SUMS (which would be wrong)?

I also assume that there are differences in the scores people given to specific trials, so i need to have this as a random effect.

Any advice is appreciated.

$\endgroup$
  • $\begingroup$ Are differences such as 5R/6B vs 6R/5B random, or systematic? Do number of IDs always sum to 11, or does the number of IDs also vary? $\endgroup$ – BruceET Jul 30 at 19:29
  • $\begingroup$ The differences are systematic. There are always 5R/6B in one condition and 6R/5B in the other. So yes, at the end, the trials sum to 11 per each participant. $\endgroup$ – Hubris555 Jul 31 at 11:49
0
$\begingroup$

I (kinda) figured out what nlme and/or lme4 are doing when runing such a study. By entering the raw data (i.e. all the trials for each participant, without any pre-processing, e.g., sums, or means), the final results show that it is indeed calculating the scores for each factor and sub-factor by calculating averages. Namely, it is taking into account the unequal trials when analyzing the results. (yey).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.