Dear Ho, 

Check this link out:

http://www.psy.unipd.it/~masin/Masin-1980d.pdf

@Béatrice Marianne Ewalds-Kvist 

tanks for your sharing.

I have read this paper before. In my understanding, this paper focused on the differnence of two ways to caculate the PSE using constant stimuli. The way I perfer to use in my study is similar to use the psychometric function. But I couldn't find the author use what specific function to fit the data, maybe just because I couldn't understand well.

actually I have read many materilas about caculating the PSE, but I can't find the way for the yes/no task. Most of them is about testing threshold.

I don't know if you have any idea about it , or you have a better understanding about the paper?

manythanks.

Hi Ho:

If I understand your question, I believe one method is to plot your data as a function of absolute value of the deviation from the middle along the x-axis (that is, collapse across left and right deviations) and proportion of responses that the landmark was deviated from the middle on the y-axis. Then you can fit a Sigmoid function to your data and calculate the point along the x-axis corresponding to the 50% point along the y-axis.

Now, the point along the x-axis corresponding to the 50% point along the y-axis corresponds to how far deviated the landmark needed to be from middle before observers were able to report the landmark as deviated above 50% of the time.

I hope this was helpful.

Best,

Greg