USING THE RANDOM PARAMETERS LOGIT MODEL TO COMBINE REVEALED AND STATED PREFERENCE DATA

Recent literature has combined Revealed (RP) and Stated Preference (SP) data in the Multinomial Logit Model (MNL) to estimate the value of environmental goods. However, emerging research has identified that a limitation of the MNL is the assumption of Independently and Identically Distributed (IID) errors, resulting in inaccurate model predictions and inconsistent utility parameters. Our analysis applies an alternative method to combine RP and SP data that takes into account the heterogeneity in both the observable and unobservable components of utility. This allows us to test whether such heterogeneity has an important effect on predicting behavioral choices