Matt,
You may want to

incorporate some type of "weighted" average. This will minimize the impact of single data points. (e.g. when one

person out of a hundred rates something as a 10, it shows up as a 10, whereas the weighted results would show that

only 1 person out of a hundred rating it as a 10.)
Unfortunately (and as others have implied) there may be too many

factors (e.g. age, race, etc) at play to make a simple average useful. Someone else in another thread tried to

collect more data to perform a multivariable analysis. This may be the way to go. Collect age, race, build, etc

along with these ratings. Pop the data into a statistical software like Minitab and voila! If someone decides to

go this route - please set up a spreadsheet first, so that all the data is in the same format from the get-go.



And by the way:
NPA 8 (I love getting the lady who cuts my hair all "upset" with NPA)
AE 6
I have also tried

some mixes with mixed results

peace to all
mc