Replicated difference and preference tests: how to account for inter-trial variation