A marketing expert for a publishing house wants to measure reader preference for three different covers of the same paperback novel. She has chosen 10 cities and 3 newsstands in each city which are going to sell the novel. She wants to use one of two experimental setups described below.

(a) In each city each cover is assigned randomly to one of the 3 newsstands. The number of books sold during a three-week period following the assign- ment is used to compare the effect of the covers on sale of the novel.

(b) In each city each of the 3 newsstands will sell the book using each cover for one week (that is, the trial extends over 3 weeks) in such a way that during a given week the 3 newsstands in a city will display the book with a different cover. The same 3-week period will be used in all cities. Sales figures for each week will be used for the analysis. For each of the two scenarios described above: (i) Give the name of the experimental design used. (ii) Identify the experimental units. (iii) Give the model for each of the designs and the ANOVA table, including sources of variation and d.f. (iv) Indicate how you would test whether the covers had the same effect on sales. (v) Which of the two designs would you prefer in this situation and why?