Joseph Ray Brillantes

ENGL315-90

Research Paper

15 Apr 2008

 

OR is Essential to Educational Policy Making and Administration

The Operations Research: The Science of Better website reported that because of using Operations Research (OR) the Ford Company saved $250 million in designing and testing vehicle prototypes, the UPS saved $276 million in ten years by redesigning its overnight delivery network, the US Army increased its recruitment by 17%, NBC increased its revenues in advertising by over $200 million, and the City of New Haven, CT reduced HIV infection rates with its needle exchange program. Clearly from these figures, OR provides a remarkable opportunity to economically use resources.

Despite OR’s power of making a tremendous impact on society, OR is a relatively young field. J.E. Beasley summarized the history of OR in his OR-Notes, which is hosted in Brunel University’s website. He wrote that OR started in the United Kingdom just before World War II, when the military sought the help of mathematicians to quantitatively evaluate the utilization and optimization of military resources by researching its operations, thus the term “Operations Research”. Furthermore, he added that after the war, OR became a field of study in American and British universities. OR is so young that it doesn’t even have an established course name, yet. In some universities, instead of calling it OR, it is called Management Science (MS) or Decision Science (DS).

Notwithstanding the variety of names the field is called, most academics would agree with the Operations Research website’s definition of OR, MS or DS as the use of mathematical programming “to help make better decisions”. Analysts study the operations of a system, approximately model the system mathematically, and then use mathematics to find the optimal mix of independent variables that would result to the best outcome. Analysts then recommend alternatives to the decision makers.

Even though decisions are made in every human undertaking, OR is more commonly studied and used in business, engineering, defense and the natural sciences. In comparison to the number of publications tackling applications of OR in the list above, there are very few publications wherein OR was used in the social sciences. However, OR can be used in the social sciences, specifically in educational policy making and administration. The use of OR is essential in educational policy making and administration to ensure that resources are optimally allocated in pursuit of educational goals. Several cases of how OR was used in policy making and educational administration demonstrates this.

OR can be used in educational policy making because it was used in policy making. Numerous OR practitioners have analyzed several significant urban and community problems, and their analyses have influenced policies of community decision makers in solving those problems. Some of their findings also influenced regional and national policies.

One example of OR studies influencing community, regional and national policies was done by Edward Kaplan. Kaplan was invited by the City of New Haven, CT to evaluate its needle exchange program in 1991. He wrote in “Adventures in Policy Modeling! Operations Research in the Community and Beyond” that in the 1980’s and in contrast to the nationwide average of 30 percent, 60 percent of the city’s Acquired Immune Deficiency Syndrome cases were drug injectors (5). When Human Immunodeficiency Virus (HIV)-infected drug injectors share their used syringes with other uninfected drug injectors, the infected injectors transmit the virus to the uninfected injectors through the used syringes. To prevent the transmission of the virus through used syringes, injectors must not use and share used syringes. A needle exchange program provides drug injectors clean syringes in exchange for used ones. He continued that despite the opposition of community groups, police officials and drug treatment providers, the city passed legislation for a needle exchange program (5). Kaplan was invited to design and conduct an evaluation of the program, with two important restrictions. His team could not ask questions from the participants of the study, and the participants would not be tested for HIV. Otherwise, participants might be anxious to take part in the needle exchange program. Kaplan worked with the imposed restrictions by using a mathematical model he previously developed. The mathematical model paralleled the spread of HIV through needles with the spread of malaria through mosquitoes. He explained this parallelism as such,

Viewing needle exchange through the lens provided by this model, it became clear that exchanging used needles for clean ones is equivalent to replacing the mosquitoes in a malaria outbreak with newborns free of disease. [. . . Needle] exchange forces needles to share fewer people! Finally, if fewer needles become infected fewer people will as well (5).

Instead of monitoring and surveying participants of the program, Kaplan and his team developed a syringe testing and tracking system “that would record the use of the program by clients, [and . . . ] detailed the travel and infection status of individual needles” (5). The team then used the data from this testing and tracking system in Kaplan’s mathematical model, and “estimated conservatively that rate of new HIV infection among program participants declined by 33%” (5). The results of the evaluation convinced city officials to continue the needle exchange program. The results also convinced state officials to expand the program in Connecticut. Other countries also initiated needle exchange programs based on Kaplan’s work. Kaplan also reported that his work also made the development of an economic analysis model of needle exchange possible to help other legislators in identifying conditions that would justify such a program in their jurisdictions (6). Kaplan’s work demonstrated in a convincing manner that OR is a powerful tool in policy making and analysis. Needle exchange programs seem immoral and should not be initiated and sustained because it might encourage drug use. Kaplan’s evaluation implies that because HIV infection among participants declined, then needles are not being shared. Because a used needle is exchanged for only one clean syringe, and because syringes are not being shared, then these imply that the number of drug injectors have remained constant. Despite the fears of an increase in drug use because of accessible syringes, the evaluation has proven the opposite, and that the use of drugs and the accessibility of syringes are uncorrelated. Hence, the sustaining of the needle program was not influenced by politically-motivated opinions but by empirical evidence, which provided a feasible solution with promising results. Empirical evidences are better motivations in formulating feasible solutions which can be adapted to reflect the values of a society, than politics. If OR were used in educational policy making, then empirical evidence would drive educational policies rather than politics resulting in better formulated policies.

Another OR study that influenced policy, though only in a city, was done by Charles Larson, as well as the founders of Enforth Corporation. Larson wrote in “Public Sector Operations Research: A Personal Journey” that he, as well as the Enforth Corporation, was selected by the New York City Department of Sanitation (NYCDOS) to determine the optimal number of barges to meet the doubling of garbage tonnage that needed to be transported by barges to Staten Island’s Fresh Kills Landfill (FKL) (140). The reason behind the doubling of garbage tonnage that needed transportation was because environmental regulation required NYCDOS to close all its in-city landfills except FKL. He reported that NYCDOS needed to increase the size of its barge fleet to meet the demand, and it was ready to buy 120 barges costing $1 million each. He continued that by using OR methodologies and models, his company determined that NYCDOS needed only 80 to 90 barges to meet the demand (140). Despite their recommendation, NYCDOS bought 100 barges instead. Larson concluded that because of the result of their study, NYCDOS saved $20 million less $100,000, which was the cost of hiring his corporation to conduct the study (140). The results of Larson’s OR study influenced the decision of buying only 100 barges instead of 120, and definitely influenced the policy makers who approved this capital investment. NYCDOS also made a policy to maintain 100 barges in its fleet until a sudden increase in demand would require additional barges. Doing this, NYCDOS used its resources better than it was planning to do. If OR studies such as this were done to determine educational expenses, then the education sector would have used its resources much more efficiently while maintaining effectiveness. OR studies in the educational sector are better in influencing policy makers because these studies provide empirical evidence. As previously stated, solutions that are empirical are better than politically-motivated ones because they provide feasible solutions and foreseeable outcomes.

Despite OR’s impressive applicability in policy making and community development and its practitioners’ claim of its interdisciplinary nature, OR has not been extensively used in educational policy making. The lack of publications applying OR to educational policies and universities granting undergraduate and postgraduate degrees linking the study of educational policy analysis through an OR perspective makes this very apparent. However, OR can be used in educational policy making. Educational systems are not very different from urban and community systems. Both systems are highly-influenced by political or legal, economic, socio-cultural, demographic, technological and even global factors.  In addition, because of educational systems’ similarity with urban and community systems, methodologies and models developed for urban and community systems can be translated to educational systems without extensive difficulty. Notwithstanding the lack of OR analysis of educational policies, there are a few publications reporting the successful application of OR to educational administration.

The most common application of OR to educational administration is school busing. Almost all school districts in the US contract external organizations to transport children from bus stops near their home to schools in the district and vice versa. Obviously, the more buses the school district needs, the more expensive the contract becomes, even if contracts are periodically offered to the lowest bidder. Larson and Amedeo R. Odoni wrote in their book Urban Operations Research, which is hosted in the Massachusetts Institute of Technology (MIT) website, a classic example of improving school bus routing using OR. Larson and Odoni wrote,

For their project, a group of three of our [graduate] students [were hired by and] worked with members of the [Boston] School Committee to redraw school bus routes and reschedule their arrival and departure times (the same bus would make several trips in the morning, as elementary, junior high, and high schools each had different starting times). By employing a good deal of common sense and a number of heuristic routing techniques from network analysis, the students prepared a commended redeployment of buses for [. . .] the entire school system.

Before the implementation of the new routing system, the Boston School System added bus routes on top of existing bus routes to meet the increasing demand of transporting school children to an increasing number of schools because of the dramatic rise of population in the area. This addition of new routes to previous ones often resulted in different buses stopping at the same bus stops to transport children to the same schools at different times. This was clearly an uneconomical method of busing children, and was the motivation behind the project of the MIT graduate students. Larson and Odoni reported that because of the new routing scheme the town saved “about 26 percent of its annual [. . .] bill for the school-bus services”. Results such as these are not exclusive to the US because they were also duplicated in Japan.

Nanzan Educational Complex in Nagoya started preparing for a foreseeable increase in competition for top students in 1995. As the population reduction of Japan continues, there would be fewer students enrolling in schools and the competition for the talented students would increase in the coming years. To ensure that it attracts enough talented students to continue its operations, Nanzan Educational Complex started to improve the quality of education it offers by ensuring that its compensation package was competitive to retain and attract talented faculty, and by improving its facilities. These improvements require Nanzan Educational Complex to increase its coffers. Atsuo Suzuki, Katsushige Sawaki and Tosheharu Hasegawa wrote in “An OR/MS Approach to Managing Nanzan Gakuen (Educational Complex): From Strategic to the Daily Operational Level” how the educational complex improved its school bus system and saved “more than US$5 million – money the university can spend to improve students’ education and the university’s competitiveness” (49). Previous to the implementation of the recommendations of the study, Suzuki, Katsushige and Hasegawa wrote that they found several inefficiencies of the bus system. They discovered that the bus system operated 29 buses, the number necessary during peak times, the whole day and even at non-peak times, which only required five or six buses. Because the bus system always operated as if during peak times, bus drivers had a lot of idle times during non-peak times. The bus drivers were all full-time employees of the contracted bus operator, so even if they had several hours of idle time, they required payment from the operator, which the operator transferred to Nanzan Educational Complex. Also, the high schools and the college had separate bus fleets, instead of sharing buses (47). After running several simulations of feasible scenarios, the authors recommended new timetables, sharing of buses by the high schools and the college, and the delay of start time for the college by 20 minutes to reduce the number of buses operated during peak times (48). By reducing the number of buses during peak times, the variation of necessary buses during peak and non-peak times became smaller, resulting to a lean bus fleet. The authors also recommended that the administration negotiate with the bus operators to hire less full-time and more part-time drivers for its school buses to reduce the bus drivers’ idle time (48). The company agreed to the new timetables, and the hiring of more part-time drivers in fear of losing the contract to a more competitive bidder. The improvement of Nanzan Educational Complex’s bus system implied that the applicability of OR in the administration of educational systems transcend any country’s political, legal, economic, socio-cultural, demographic and technological constraints.

Finally, Raymond G. Taylor in his “Operations Research Solves Real Problems for School Administrators” also reported another instance of using OR to increase the efficiency of a school bus system. He wrote that network algorithms were used in “finding the residual capacity of school bus networks”, though he neither elaborated on the details of the problem nor how it was solved (346).

OR applications to educational systems are not limited to just school bus systems. There were some publications that described using OR methodologies and models in addressing varied educational administration issues. Some of these issues were on school schedules, assignment of proctors in examinations, development and assembly of an examination, and creation of study groups.

The first two, school schedules and assignment of proctors in examinations, were done in Nanzan Educational Complex also by Suzuki, Sawaki and Hasegawa. For fifty years before 2006, university classes in the complex started at 9AM. Most of the university students commuted for more than an hour from their homes to the school, requiring them to wake up no later than 7AM, and to travel during rush hour. A sizeable number of students were frequently late for their 9AM class. Thus, most of the professors refuse to teach 9AM classes.  According to Suzuki, Sawaki and Hasegawa that during the first period of classes at 9AM the “classroom usage rate is between 48.2 percent and 59.1 percent, which is low compared to the highest classroom usage rate, which is 90.9 percent” (50). To improve the utilization of its facilities, the university wanted to decrease the variation of classroom usage rates. After running simulations and analyzing the data, Suzuki, Sawaki and Hasegawa recommended “delaying the time of the first period by [. . .] 20 minutes [. . .] so that students can avoid peak congestion and the university can increase classroom utilization” (50). Furthermore the authors believe that because of the delay, the university “may [. . .] gain a better geographical advantage over other universities in the area” because people living farther than the rest of the student body can also study in the university making the pool of candidates for admission to the university larger (50).

With an increase in the pool of candidates for admission, the Nanzan Educational Complex must also increase its efficiency of assigning proctors to its admission examinations. Suzuki, Sawaki and Hasegawa reported that previous to the OR method of assigning proctors, the “administrative staff member in charge of the assignments used to spend three days [assigning 200 proctors to administer the examinations to 12,000 students in five days] doing the work by hand” (51). The authors continued that they solved the problem by devising “a heuristic algorithm to make a special cost matrix for the minimum-cost network-flow problems that automatically obtained integer solutions using the linear programming method [. . . and] using [. . .] Excel-based optimization software [. . .] in a couple of hours” (52-53). If schools were to be managed such as the Nanzan Educational Complex, schools would minimize their costs, increase their facilities utilization efficiency, and decrease their labor hours.

The versatility of OR is not limited to just school-wide management, but also to classroom management. Dmitry Krass’ and Anton Ovchinnikov’s “The University of Toronto’s Rotman School of Management Uses Management Science to Create MBA Study Groups” proves that OR can be used to assign students to create well-balanced groups. Well-balanced groups were defined in Krass’ and Ovchinnikov’s article as “heterogeneous [. . .] groups whose members possess diverse personal characteristics and backgrounds” (126). Furthermore, Krass’ and Ovchinnikov’s article states that “[researchers] in organizational behavior [. . .] have found that [. . .] well-balanced groups [. . .] are more effective than homogenous groups on complex projects and problem-solving tasks, [. . . and] that the more diverse the groups are, the deeper the perspective each student can gain from his or her peers” (126-127). Based on this research result, the Rotman School of Management centrally assigns study groups to ensure that groups are homogenous, but that the members of each group are diverse. Each group must have the same proportion of men and women, of people with technical and humanistic backgrounds, and of people who are native speakers of English and not as with other groups. For example, if there were 126 males and 53 females, and it was determined that each group must have four members, then each group must have two or three men, and one or two women. The same logic is applied to the other criteria. Obviously, each member in a group has a stronger skill set in a field than the other members of the group, and because each person in the group has a stronger skill set than others, there is a tendency to assign the work of the group solely to the person with the stronger related skill set. For example, people with technical backgrounds are better in quantitative courses than people with humanistic backgrounds, and so any technical work that is supposedly cooperatively done and learned by each member of the group is exclusively assigned to the member with a stronger quantitative skill. This situation is defined in the authors’ article as work-splitting. They wrote about the effect of work-splitting to learning,

[Work-splitting diminishes] the supposed benefits of group work [. . .] and risk subverting the learning process: in many cases, the students who do the assignments have the strongest prior background in that particular area and thus gain the least benefit by doing the additional work (127).

The authors continued,

To address the issue of splitting work within groups, the MBA administrators decided to implement a multiple-groups policy. Under this policy, they assigned each student to several study groups, with different groups employed by different courses. [. . .] For this strategy to be effective, we must ensure that [. . .] each student should have [significantly] different set of partners in each of his or her study groups (128).

The authors formulated this problem as a “classical management science assignment problem with side constraints, [. . .] developed a simple heuristic-based approach, [. . . and] implemented [their] ideas and algorithms in a [. . .] Microsoft Excel macro [. . . that was] versatile and simple to use” (129-132). Previous to the authors’ software implementation, they reported that two employees would spend a week creating well-balanced groups manually, while after the implementation, it only took 20 minutes (128, 133). Though their approach is only feasible when assigning hundreds of students to many different groups, Krass’ and Ovchinnikov’s work clearly implies that OR can be used to ensure that learning objectives are efficiently met by creating well-balanced groups with a software in 20 minutes.

Aside from ensuring that learning objectives are met with the creation of well-balanced groups, OR can also be used in developing and assembling standardized examinations. Ronald Armstrong, Dmitry Belov and Alexander Weissman reported in “Developing and Assembling the Law School Admission Test” that they used OR to assemble the quarterly Law School Admission Test. They wrote in the abstract of their article,

Operations research tools can help in developing rigorous standardized tests. Our mixed-integer program [. . .] provides a model for assembling forms for the Law School Admission Test (LSAT). Since 2002, our LSAT assembler – software we developed using a Monte Carlo approach – has produced test forms meeting all specifications. This software has saved thousands of hours of personnel time (140).

In their conclusion, the authors added,

Many problems in testing and educational measurement can be addressed by using operations research techniques. [. . .] Analysts can also use operations research in designing tests, calibrating items and sampling examinees from populations of interest (148).

Their work implies that numerous standardized tests meeting set specifications can be produced automatically from pre-tested pool of questions in lesser labor hours than manual production. Thus, shorter standardized tests with the same predictability than their longer counterparts can be administered more often. Clearly, OR can help save time teachers spend in creating tests and diverting time saved to preparing and administering instruction in the classroom. Thus, teachers can devote more time to student learning.

In addition to the articles of Larson and Odoni, Suzuki, Sawaki and Hasegawa, Krass and Ovchinnikov, and Belov and Weissman, Taylor also enumerated several OR methodologies that was successfully applied to educational administration, though he did not expound on the details of each. He wrote,

In the past two years, in cooperation with the [OR/Ed] laboratory, my colleagues and I have promulgated about 40 applications of OR to educational administration.

They include several linear programming applications: manpower allocation; distribution of merit pay; resource allocation to curricular alternatives; capital purchasing decisions; budgeting; assignment problems; and, the timing of capital projects.

[The] lab has devised several examples of the use of goal programming to resolve resource allocation problems with multiple objectives: assignment of students to schools; finding the optimal location of a new school; and working on public trust for school boards.

Two non-linear programming problems have been explored and their solutions promulgated: finding the optimal layout of space; and curricular scheduling using an industrial bin-packing model.

A number of applications have been developed which do not rely on mathematical programming, but on the standard algorithms for dealing with networks: communications system interconnections in schools; finding the residual capacity of school bus networks; and probabilistic PERT as a project management tool.

Two types of problems (Markov and Queuing), which fit into the broad category of stochastic processes, have been investigated. These problems include: forecasting teacher shortages; evaluating flow of students through university programs; personnel policy appraisal; concept mastery and the earliest expected time to the criterion; determining the minimum number of stations needed for college registration; and, determining the minimum number of on-line terminals needed to satisfy certain library service policies.

And finally, the relevance of Bayesian analysis has been demonstrated as it relates to the school closing problem in geographic areas where there is a great deal of snow (346).

Clearly from the promising results reported in these few articles of the application of OR to educational administration, OR should be used in managing schools.

However, it is disappointing to realize that there is still no movement to use OR in schools. OR is still not widely used in schools because those who are managing the schools are not oriented and trained to use OR. In fact, most of the graduate programs preparing teachers to be educational administrators still have not included OR in their quantitative analysis courses. Taylor reflected upon this weakness in educational administration programs,

[It] seems strange that we invest so much energy in teaching educational administrators a mode of research that has little relevance to their work [. . .]. The kinds of questions I asked as an administrator, and the kinds of problems I was expected to solve, simply did not yield to statistical hypothesis testing [. . .]. My problems had much more to do with finding optimal arrangements for resources, evaluating policy alternatives given uncertain data, planning and forecasting (346).

Though graduate school coordinators might contend that their programs prepare teachers to be leaders whose primary responsibility is to improve learning in the schools, it is unacceptable to ignore the management functions of these leaders when they assume their roles as educational administrators in schools and school districts by not including any OR course in graduate programs.

It would certainly help the education sector in achieving its goals by having more resources in its disposal. However, because resources are limited, it is essential that resources are used effectively. It was demonstrated by the several cases that OR could be used to make substantial savings in some areas of education by studying its operations and making them efficient. The savings can then be allocated to fund programs that would increase learning. Educational politicians could also make their policies more relevant by using OR to model educational systems. The several cases mentioned in this paper demonstrated that OR could model behaviors of urban service systems and optimize their outputs or minimize their costs. The urban service systems are not very different from educational service systems. Thus, models developed for the former can be used with few adjustments for the latter. With this type of analysis, educational politicians can lobby for changes in the education sector. These changes in policies would reflect scientific conclusions rather than politically-driven assumptions. These policies would also powerfully justify reallocations of resources to other educational initiatives that would bring about more substantial results than the status quo. Thus, it is folly for educational administrators and politicians to ignore OR, and for practitioners to exclude the education sector in their field.

Works Cited

Armstrong, Ronald, Dmitry Belov and Alexander Weissman. “Developing and Assembling the Law School Admission Test.” Interfaces 35.2 (Mar-Apr 2005): 140-151. Academic OneFile. 2 Mar 2008 <http://find.galegroup.com/itx/start.do?prodId=AONE>.

Beasley, J.E. OR-Notes. 18 Mar 2008 < http://people.brunel.ac.uk/~mastjjb/jeb/or/intro.html>.

Kaplan, Edward H. “Adventures in Policy Modeling! Operations Research in the Community and Beyond.” Omega: The International Journal of Management Science 36.1 (Feb 2008): 1-9. Academic OneFile. 19 Feb 2008 <http://find.galegroup.com/itx/start.do?prodId=AONE>.

Krass, Dmitry and Anton Ovchinnikov. “The University of Toronto’s Rotman School of Management Uses Management Science to Create MBA Study Groups.” Interfaces 36.2 (Mar-Apr 2006): 126-137. Academic OneFile. 25 Feb 2008 <http://find.galegroup.com/itx/start.do?prodId=AONE>.

Larson, Richard C. and Amedeo R. Odoni. Urban Operations Research. Cambridge, MA: Massachusetts Institute of Technology, 1997. 1 Mar 2008 <http://web.mit.edu/urban_or_book/www/book/index.html>.

Larson, Richard Charles. “Public Sector Operations Research: A Personal Journey.” Operations Research 50.1 (Jan-Feb 2002): 135-145. Academic OneFile. 2 Mar 2008 <http://find.galegroup.com/itx/start.do?prodId=AONE>.

Operations Research: The Science of Better. 2007. 18 Mar 2008 <http://www.scienceofbetter.org/index.htm>.

Suzuki, Atsuo, Katsushige Sawaki and Toshiharu Hasegawa. “An OR/MS Approach to Managing Nanzan Gakuen (Nanzan Educational Complex): From the Strategic to the Daily Operational Level.” Interfaces 36.1 (Jan-Feb 2006): 43-45. Academic OneFile. 26 Feb 2006 <http://find.galegroup.com/itx/start.do?prodId=AONE>.

Taylor, Raymond G. “Operations Research Solves Real Problems for School Administrators.” Education 114.3 (Spring 1994): 346(3). Academic OneFile. 18 Feb 2008 <http://find.galegroup.com/itx/start.do?prodId=AONE>.