A bakery wants to evaluate new recipes for carrot cake. It decides to ask a random sample of regular customers to evaluate the recipes by tasting samples of the cakes. After a customer tastes a sample of the cake, the customer will provide scores for several characteristics of the cake, and these scores are then combined into a single overall score for the recipe. Thus, from each customer, a single numeric score is recorded for each recipe. The taste-testing literature indicates that in this type of study some consumers tend to give all samples low scores and others tend to give all samples high scores.
a. There are two possible experimental designs. Design A would use a random sample of 100 customers. From this group, 20 would be randomly assigned to each of the five recipes, so that each customer tastes only one recipe. Design B would use a random sample of 100 customers with each customer tasting all five recipes, the recipes being
b. The manager of the bakery asked for a progress report on the experiment. The person conducting the experiment replied that one recipe tasted so bad that she eliminated it from the analysis. Is this a problem for the analysis if Design B was used? Why or why not? Would it have been a problem if Design A was used? Why or why not?