They must have studied a group through hypothetical choices (instead of actual choices that had any weight on their budget), or they did an open study in which

They must have studied a group through hypothetical choices (instead of actual choices that had any weight on their budget), or they did an open study in which the tested subjects could have chosen something based on how they wanted to be seen choosing, or it may be just plain memetics, or any other reason. That is, however, most of the time the close as you can get to the actual human behavior within a reasonable effort (i.e. unless you're willing to go with covert surveillance or something).