It is especially important to recognize that the different results do not only stem from assembling the consumption categories differently.There are also differences for the categories that overlap in both studies. Since I use a largerdata set, account for fixed individual effects, and use the more appropriate ordered choice model, there is good reason to put stronger belief in my results.