I'm performing a study where subjects rate their perceived, subjective intensity of a stimulus presented to them at different magnitudes. To submit a rating, they click on on a logarithmic scale presented to them on a monitor, and that selected pixel value is used as the rating. They must rate several different stimuli, for roughly an hour. During pilot runs, subjects have told me that sometimes they change how they go about rating in the middle of a session. They arbitrarily find it easier to change the reference rating in their memory to something different from before, perhaps a more recent intense stimulus is used as the new reference.
Are there any traditional ways to compensate for this? My initial thought is to have the subject rate something in the training session, and then show them on every subsequent stimulus the location at which they rated that training stimulus.