Are Preferences for Food Quality Attributes Really Normally Distributed? An Analysis using Flexible Mixing Distributions