STA442/1008 Assignment 11

Do this in preparation for the Quiz on Friday Dec. 9th.

A national retail chain wanted to study the effects of two advertising campaigns (Factor A) on the volume of sales of athletic shoes over time (Factor B). Ten similar test markets ("Subjects") were chosen at random to participate in the study. The two advertising campaigns were similar in all respects except that a different sports personality was used in each. Sales data were collected for three two-week periods:

1. The two weeks prior to the campaign
2. The two weeks during which the campaign occurred
3. The two weeks after the campaign was concluded

  Test                         Sales Volume at
 Market   Campaign    Period 1     Period 2    Period 3
    1        1          958          1047        933
    2        1         1005          1122        986
    3        1          351           436        339
    4        1          549           632        512
    5        1          730           784        707
    6        2          780           897        718
    7        2          229           275        202
    8        2          883           964        817
    9        2          624           695        599
   10        2          375           436        351

1. Using the multivariate approach to repeated measures, test for the main effect of campaign, the main effect of period, and the campaign by period interaction. If the main effect for period is significant, it would be nice to follow up with Bonferroni-protected pairwise comparisons. However, this is clumsy with proc glm, so we will not do it now. Instead, we'll wait until we've seen proc mixed. Based just on the means and the tests you've been asked to do, what appears to be going on here?
2. If you are just interested in which campaign is better and not in testing whether the advertising is better than nothing overall, there may be a better way to analyze the data. Using the (baseline) Period One measurement as a covariate, do a multivariate regression with sales at period 2 and sales at period 3 as multiple dependent variables.
   1. Controlling for Period one sales, did the campaign have an effect on Period 2 and/or Period 3 sales? If the multivariate test is significant, follow up with Bonferroni-corrected univariate tests. In plain language, what do you conclude (if anything)?
   2. Controlling for campaign, are Period one sales related to Period 2 and/or Period 3 sales? If the multivariate test is significant, follow up with Bonferroni-corrected univariate tests. In plain language, what do you conclude (if anything)?

Bring your log files and your list files to the quiz.