At USAF Test Pilot School (a long time ago, Class 85A) we had a matrix of configurations and airspeeds we used to obtain roll rate. You are correct in that roll rate will vary with speed, altitude, configuration, etc.
For the T-38 at each altitude/KCAS point we would do 1/4, 1/2, 3/4 and full stick deflections rolls left and right. As I recall with flaps down, we would time from +/- 60 deg. in clean configuration we would do a full roll right and then left. We taped a reference on the panel to help get the same deflection each time.
For lower performance aircraft like the RV-3/4/6/7/8 I would use +/- 60 deg bank though you could use the full roll technique. Choose altitudes you are comfortable with 5k, 7k, 9k and a/s such as 100KIAS, 120KIAS, 130KIAS etc. up to manuevering speed.
For any flaps down use only +/- 45deg roll.
Stating the obvious, it is very important that you reach the aileron deflection you are testing (1/2 deflection, full deflection) very rapidly (and consistently) instead of gradually as this greatly impacts the results/time you record. keep the a/s within 1-2 knots for a valid data point and altitude within 50-100 ft.
After a full flight of roll testing the T-38 in the clean configuration at the higher speeds your internal gyro would be uncaged (max roll rate of 720deg/sec!).
When you plot the data you can look at changes in roll rate vs a/s for a given deflection, altitude versus deflection etc. I think on the RV-6/7/8 you will see the roll rate is non-linear with deflection with not as much change between 3/4 and full as between 1/2 and 3/4 deflection.
Hope this helps. post the results if you decide to do a full test, might find some interesting results.
TJ