FROM THE EDITOR
BARBARA B. FLYNN, Decision Line Editor, Babcock Graduate School of Management, Wake Forest University
Interested in becoming more involved with the Decision Sciences Institute? We are looking for a new feature editor for the Doctoral Issues feature in Decision Line. This regular feature focuses on issues that are relevant to doctoral students and to doctoral programs. As with all of the features, columns may be written by the feature editor or contributed by guest columnists. If you are interested in learning more, please let me know.
I would like to highlight two special articles in this issue of Decision Line. The first features another of the instructional innovation award finalists from the 1996-97 competition. Huei Lee writes about the development and use of a multimedia tutoring system for the introductory production/operations management course. In the second article, Bill Bassin speculates about the process of developing the seedings for schools in the NCAA basketball tournament. I'm sure that you will find both articles interesting and entertaining.
As the weather turns cold, think warm weather and San Diego!
A Test-based Distributed Multimedia Tutoring System for Production and Operations Management
by Huei Lee, Department of Management and Marketing, College of Business, Lamar University
Advances in production technology and intense global competition during the last decade have made the study of production and operations management (POM) one of the most challenging areas in business schools today. Several scholars have indicated that the course of POM should teach students manufacturing technology and its role in developing business strategy. However, POM represents a departure from traditional business courses due to its heavy technical content.
Many students in the college of business resist the POM course. First, most students who choose business careers think they are not interested in manufacturing. Second, students normally have difficulty in understanding such terms as flexible manufacturing systems (FMS), cellular manufacturing (CM), and computer-aided design/computer-aided manufacturing (CAD/CAM). This situation becomes worse when instructors explain these concepts without using overhead projectors, transparencies, video demos, or other visual aids in the classroom. Some students also have difficulty with quantitative models and analysis. These students complain that they seem to understand a problem in the classroom, but they do not know how to work a similar one in the examination.
Multimedia and Internet access are viable teaching methods to minimize students' resistance to the POM course and to help students understand the context of POM. Multimedia refers to the use of diverse electronic tools, such as videotape, videodiscs, and CD-ROMs in presenting information in the form of images, videos, text, graphics, and sound. Distributed multimedia refers to the distribution of multimedia processing among multiple geographically or functionally separate locations linked by a communication network (i.e., local area network, Internet, or Intranet).
The author has developed an instructional tool, the Distributed Multimedia Tutoring System (DMTS), to help students understand the context of POM and other technology-related courses. The DMTS, funded by a university research grant, is a multimedia system that combines text, graphics, audio, and video with the interaction of a computer in a network environment. The DMTS has been designed, developed, and implemented by the author and student assistants since 1993.
The DMTS includes three tools:
Implementation of the Innovative Approach
There are three major ways to use DMTS in student learning: (1) multimedia applications displayed by an instructor on a big-screen monitor, (2) multimedia applications used by students in a stand-alone system, and (3) multimedia applications used by end users or instructors in a network environment. Multimedia used as a presentation tool allows the instructor and students to convey information through the use of more than one of the senses. The more senses used, the greater the likelihood that concept clarity will be developed by students. The output of the multimedia screen can be sent to a liquid crystal display (LCD), a projector, or a big screen TV/monitor while the instructor explains the concept. With the use of a network, students are able to access information in an interactive multimedia network.
Why is DMTS important as a teaching and learning tool in POM? The answer is that there are a multiplicity of reasons, the main one being that it will get the attention of students. Students today, especially youngsters, spend hours accessing the Internet and playing video games that have moving pictures and sound. It is not reasonable to think that an instructor with a chalkboard and lecture notes will be able to capture the attention of students who thrive on the excitement and challenge of video and computer games. Another reason for the importance of Internet/multimedia usage in the classroom is that it enhances the process of educating students in the use of new and emerging technologies.
Effectiveness of the Innovative Approach
To investigate the uses and successes/failures of the Distributed Multimedia Tutoring System, the following null hypothesis is derived: There is no difference in students' achievements between pre-test and post-test of multimedia use in POM classes. A pre-test/post-test control experiment was conducted to test this hypothesis. The tests were administered by the author in POM classes. Students in the POM classes were divided into experimental and control groups, and were given written tests to determine their achievements. The control group was instructed in the traditional manner, while the experimental group was instructed using the multimedia system, DMTS. At the end of each semester, questionnaires were devised to determine, from the student's perspective, the success of the POM course as taught. Measurement tools in this project are: (1) written pre/post tests that objectively measure performance measure students' achievement, and (2) a questionnaire survey that represents subjective opinions from students.
The pre-test was administered during the beginning of the semester, before the DMTS was used in classrooms. The students were primarily juniors and seniors who had no previous working experience in manufacturing. A post-test was conducted after using the DMTS for each topic unit. Pre-tests and post-tests were administered in the following POM topics: production technology, quantitative models, and total quality management.
Based on the data presented in the study, there was a significant difference between pre-tests and post-tests. The test scores for either the pre-tests or the post-tests resulted from the summation of the three topics in this research: production technology, quantitative models, and total quality management. In order to ensure that the pre-tests and the post-tests were significant in each testing area, three separate t-tests were conducted. The test results also indicated that there were significant differences between pre-test and post- test in each area.
Using the DMTS in teaching POM has been a good experience in the last three years. Research findings confirm the author's belief that the major advantages of using multimedia instruction over the traditional method are: (1) it gets more attention from the students; (2) students gain better understanding of new production terminologies; and (3) using multimedia as a tool also introduces a new emerging technology.
Acknowledgments: The author would like to thank Kuo-Lane Chen and Ramona Porter for assistance in conducting this research. Portions of this paper were published in the 1996 Proceedings of the Decision Sciences Institute.
The Road to the Final Four, or, Are NCAA Committee Seedings Informationally Efficent?
by William M. Bassin and Amy D. Spencer, Shippensburg University
Since the 1984-85 season, the NCAA Division I men's basketball tournament has had four regions with 16 teams each. In the first round, the team seeded first in a given region plays the 16th seed, the team seeded second plays the 15th seed, and so on, and the team seeded eighth plays the ninth seed. There are thus eight first-round games in each region, and the match-ups are arranged so as to maximize the difference in seeding for each game.
In the second round there are four games in each region. If the first round has gone according to form, the team seeded first in the region plays the eighth seed, the team seeded second plays the seventh seed, and so on. However, for example, if in the first round the team seeded second has lost to the 15th seed, and the seventh seed has won its game, then the seventh seed plays the 15th seed in round two. In other words, the teams are not "reseeded."
In the third round there are two games in each region, and in the fourth round (or regional finals) there is just one game. The winners of the four regional finals advance to the so-called "final four." Round five is the national semi-final round, while round six is a single game for the national championship.
The goal of this research is to develop an equation that predicts the probability of a given team winning a tournament game based on available information. Figure 1 shows the results of using the following logistic equation developed from data for 672 games in rounds one, two, and three from 1985 through 1996:
P' = 1 - (1 / exp (-0.157 + 0.162D + 1.019F + 0.564S)),
P' = the probability that the higher seeded team wins the game,
D = the difference in the seeding (e.g., if a team seeded first plays a team seeded ninth, then D=8),
F = 1 if the team is a first seed, and 0 otherwise, and
S = 1 if the team is a second seed, and 0 otherwise.
Note that the weighted average of the absolute values of the residuals is 2.33%. Without the F and S variables the average absolute residual would be 3.62% -- a difference which is highly significant statistically. Thus, first and second seeds have an advantage over and above the difference in seed.
However, when variables representing other potential sources of information (national ranking, win/loss percentage, and quality of conference) are added to the equation, none of these variables turns out to be statistically significant. In other words, the NCAA seeding information is efficient. That is, no additional data can improve the probability of picking the winner of a particular game. Put another way, despite some famous early round upsets and near upsets over the years, the first three rounds of the tournament generally go according to form.
A Possible Anomaly
There are two reasons why only data from the games in the first three rounds are used in developing the above equation. First, while there are data for 672=384+192+96 games in rounds one, two, and three, there are data for only 48 round four games and data for only 36 games in rounds five and six. More important, a better use of the rounds four, five, and six data is to validate the performance of the equation. This turns out to be especially interesting for teams seeded first.
Panel 1 of Figure 2 shows that the equation fits the actual results quite well for teams seeded first in the first three rounds -- about as well as for all teams. In the fourth round, however, teams seeded first perform relatively poorly compared to the predictions of the equation. The equation predicts that 24.6=72.4% (34) first seeded teams should reach the final four. In fact, only 19 first seeded teams do. At the same time, as Panel 2 shows, there is no drop off in fourth round performance for teams seeded second. In other words, this is a possible anomaly in the efficiency of the seedings for teams seeded first.
At the same time, Panel 3 shows that teams seeded first and second still have significant probabilities of reaching the final four. These teams make up 64.6%=39.6%+25.0% of all final four berths, despite the anomaly. By the same token, teams seeded fifth and lower (despite the memorable performance by eighth seeded Villanova in 1995, for example) hardly ever reach the final four. (Note that the 1.0% probability of winning for teams seeded fifth and lower is 6 divided by the number of such teams, i.e., 576. The 12.5% fraction of berths is 6 divided by the number of final four berths, i.e., 48.
Panel 4 shows that the underperformance of teams seeded first continues into rounds five and six. These teams win only 65% of their games against lower seeds, vs. the 77.1% predicted by the equation. The overall difference in winning percentage in these three rounds vs. the equation is 14.9%. The "t" ratio for this difference is -2.498, which is statistically significant.
Why does this happen? Perhaps first seeded teams in rounds four, five and six do not play up to potential when confronted with teams which are (1) of equal ability or (2) on a hot streak, such as the 1995 Villanova team. They may find the level of stress unusually great and succumb with more than predicted frequency.