This actually illustrates the point nicely. Surveys are NOT just a collection of anecdotes. Since each person who fills out the survey has to answer the same questions, you get (roughly) the same information from each person. In a collection of anecdotes, who knows what each person is choosing to include in his/her story and what the person is leaving out. By putting a carefully selected structure onto the information collection, you are making a "collection of anecdotes" into useful data that can be used for scientific reasoning.