I'm running an experiment on perceptual thresholds in audio. I'll try not to bog you down with too many details:
The experiment is about vibrato speed; specifically, when can you tell the difference between two stimuli that differ in vibrato speed only. Our software determines subjects' 50% threshold as a function of speed-difference; i.e., the difference in vibrato that is imperceivable.
It is run in 5 conditions, based on 5 "base vibrato speeds" (without getting into technical language, basically: slow, medium, fast (and steps in between)).
So, in the end our data look like this:
Independent variable: "base vibrato speed" (slow, med, fast).
Dependent variable: largest imperceivable difference in vibrato speed (the 50% threshold as determined by the software/experiment).
We hypothesize that as the base vibrato speed gets higher, the dependent variable (perceivable speed-difference) grows exponentially. Linear would be the null hypothesis.
So I'd like to fit both a linear regression, and some kind of exponential function to the data, and calculate a goodness of fit on both. Somehow I'm thinking chi-square won't cut the mustard here.
Do you have any thoughts?
