My friend Daniel Saunders recently released a neat little graphical toolbox for Matlab called QuestIntuition that allows you to play around with the QUEST procedure for adaptive sampling of psychometric thresholds. This will be a good resource for students learning how QUEST works.
Some questions :
- What happens if the initial guess is way off? How many trials does QUEST need to recover?
- What happens if the assumed slope is way off?
- What happens if the upper asymptote of the psychometric function is lower than the threshold QUEST is trying to find (I saw this in a paper I reviewed once)? Does it still produce reasonable samples?
Check it out!