# Which test to compare two datasets?

I have used a stimulus (stimulus1) in a perception test, to see if the participants perceive it as A or B in a forced choice task. (the stimulus is composed of several features f1,f2...)

Then I have changed one of the features (f1) of stimulus1 gradually in 5 different levels. All participants perceive the original stimulus1 as A, but when the f1 rises gradually more participants perceive the stimulus as B. At the 5th level of f1 all participants perceive it as B. So if I have a graph the X axis being five levels of f1 and the Y axis being frequency of response B, I will have an S shape (Sigmoid?) curve. I call it curve1.

Next I have used the same stimulus, but have changed another feature (f2) of it and made stimulus2. Then just like the previous part, I have raised f1 (not f2) in 5 levels for stimulus2, and have got a similar results as in part one. If you make graph for stimulus 2 (with the X axis being five levels of f1 and the Y axis being frequency of response B) the curve will be similar to the curve1, but slightly higher than it.

I have made several other stimuli and manipulated their f1 in five levels, and obtained responses and their curves.

Now I want to compare all the curves with the first curve to see if they are significantly higher or lower than curve1 or not.

I had 40 participants and they all replied to all stimuli and their variations.

I want to know which test should I run (which analysis in SPSS) to know which curve is significantly different from curve1.

-
Welcome to the site. You would have to define what you mean by one curve being higher than another. What if they cross? – Peter Flom Dec 31 '12 at 11:49
Good point by Peter. It may be that the test you need is much simpler than what you imagine: a test of means under different conditions. – rolando2 Jan 1 at 0:52