I did a test where I have 7 audio systems which where compared with each other and the tester have to say if the second one is better or worse than the first (scale -3 to 3). After comparing all 21 possibilities , they where repeated so I get two results for every pair. Now my Problem is how I have to evaluate the data. My thoughts would be with a cross table like this:
system1 system2 system3 system1 - 2 3 system2 -1 - -1 system3 -2 0 -
Above the middle diagonal the results are inverted. Is it right to calculate the mean over a line and the inverted mean for the row and then calculate the mean over these two for every system ?