Results of the Experiment, June 2002

In this study the stores were created to be paired in neighborhoods according to their strategy. Therefore, each of the eight neighborhoods had two stores pursuing the same strategy. However, one of the stores stuck strictly to their "scripts" and received no tutoring, while the other store changed operating procedures by following the tutors advice as closely as possible.

The stores were operated by a volunteer group of Governors School students. These students had over two weeks experience with playing the game, and thus were easily able to interpret both the scripts assigned to them and the advice the tutor provided. The students were given a certain latitude in their operation of the stores. For example, if the tutor advised them to use market research to choose more expensive advertising, they were encouraged to decide for themselves what that meant, and what advertising to buy. In addition, they were free to choose their own employee within the bounds of their script, were advised to make their own decisions in terms of price setting (again within the bounds of their scripts and the tutors advice), and to decide for themselves about borrowing extra capital from the bank. It turned out that none of them chose to borrow.

After seven consecutive days (eleven virtual weeks) of this competition a snapshot of the simulation was taken. (see http://dbay.cs.ndsu.nodak.edu/~mooadmin/DollarBay/scoreboards/scores-30jun02.html)

The following observations are of interest:

1. The top three players are Tutored, although the third is not significantly ahead of its counterpart.
2. The next two tutored stores are in the middle of the pack (8th and 10th place), and virtually tied with their counterparts (within 3% net worth in the worst case)
3. The next tutored store (LF_bt_T) is an anomaly in the data, and was removed from the analysis, as explained below.
4. The next tutored store is only in 12th place, but still three positions ahead and 28% more profitable than its counterpart
5. The lowest tutored store, in 14th place, is virtually tied with it's counterpart in 13th place, with only 4% difference in net worth.
6. The lowest two players are non-Tutored.

Thus we see that case-based tutoring had positive effect, with no harm caused in any case. The tutor advice either proved to be highly beneficial, or showed no real difference in the course of this short study. That some tutored players still finished at the bottom of the scoreboard can be attributed to the weakness of the strategy employed by the player (as determined by the schema, above), and not the effect of tutoring itself. Indeed, the tutoring had significant effect in some cases. It is interesting to note that the first place and last place stores pursued the same strategy in the same neighborhood, and that the Tutored store proved to be 459% better ($89,812, or nearly 4 standard deviations).

Explanation of the Anomaly: