I have all in all 600 measurements: 20 subjects x 15 different stimuli x 2 ears (presented in each ear separately). I use 4 different methods to detect whether there is a significant response in each measurement. The output of each method is either 0 (not significant) or 1 (significant). I need to test which method gives me the most number of significant detections. I also use different amount of data (10-20-30-40-50-60-70-80-90-100 % of the measurement time) for each method to find out how fast the methods can detect the significant response, i.e. which method is more effective.
I am not interested how the methods perform for each individual stimulus, but how they perform in general.
Please help me to find the correct model to compare the methods!
I tried glm (output ~ Method + NrEpochs + Method:NrEpochs, family=binomial(link = 'logit'), data = D), but the model failed at homogeneity test.
Best regards Anna
- Badges
- Science topic