Multiple linear regression.

This study calls into question the default computation of the disposition effect that uses the average purchase price as a reference point. We show, through a lab experiment, that the reference price of participants can change depending on the experimental behavioral design that is used. Our results show that for the control group, different reference prices do not show significant differences in the computation of the disposition effect, thus supporting the use of the average purchase price as a reference. However, this is not the case for participants in the treatment group. With the addition of experimental treatment concerning the disclosure of the final balance of the participants, the reference prices to compute the disposition effect showed statistically significant differences. The need to display their results caused these participants to use the first purchase price as a reference point when selling their assets.