At least since the big attention achieved by the scenario of the Club of Rome "The Limits to Growth" (Meadows et al., 1972), scenarios are a well accepted means for knowledge acquisition in research.
As part of the learning software EMILeA-stat we develop a scenario, which allows the user to apply capabilities learned in an EMILeA-stat course in a realistic environment. Also, the scenario can be used to acquire new knowledge in a field with little prior knowledge. This is achieved by 'transferring' the user into a chemical engineering environment letting him or her plan the production, evaluate the outcomes, and try to optimize the process by means of statistical data analysis and experimental design.
Styrol is one of the most important intermediate products in chemical industry. It is used in many technical synthetic materials, e.g. in Styropor(R). Styrol is produced on the basis of ethylbenzene and hydrogen in an adiabatic reactor. The production process of Styrol depends on many influences, some are controllable, others are only observable. Theoretically, the Styrol yield can be determined by means of a set of differential equations from the values of the influential factors. This can be used to simulate the process. Since the influential variables, however, are themselves influenced by random variation, such 'noise' has to be simulated also.
By means of the Styrol process the users of EMILeA-stat have a facility to get acquainted with statistical data analysis of realistic data, as well as with design of experiments in a realistic environment. In the first part of the scenario the user works with a fixed data set of 100 observations representing past runs of the production. To these data the user is asked to apply various methods of statistical data analysis, e.g. in order to find relationships between the influential factors and the Styrol yield. This data set was constructed to include many specialities, like e.g. mismeasurements, which should be found by the user. This way, the usefulness of the various statistical methods is demonstrated.
In the second part of the scenario the user is asked to go through the most important steps of quality optimization, i.e. screening, modeling, and optimization (Weihs and Jessenberger, 1999). In each step experimental designs can be chosen. The Styrol process is simulated with the corresponding values of the influential factors, and the results can be evaluated afterwards.
The evaluation of the data sets of the first, as well as the second step, is realized by means of the software R. On the one hand, prepared evaluations can be activated by mouse click. On the other hand, individual evaluations and graphics can be activated by the input of R-code.
By the intuitive structure of the scenario the user is enabled to manage the learning steps by him- or herself, to find structure in the data, to get an intuitive understanding of the process, and to design experiments to optimize the process in order to maximize yield. The user is supported by a context-sensitive helping system, which includes descriptions of possible analysis steps as well as of statistical methods. Most of the help pages are taken from the EMILeA-stat kernel including interactive examples. Moreover, the user is guided by means of questions leading to the next analysis steps or animating the user to 'play' with the data independently.
According to Moore (1997), learning by means of individual action is the most important new model of pedagogy. In particular, the combination of statistical content, learning in realistic data situations, and the use of multimedia with various interaction facilities supports effective and modern learning best, following Moore. Moreover, the facility to fit tempo and detail to individual needs makes an individual learning process possible. All these concepts are realized in the presented scenario of the learning software EMILeA-stat.
Literature
D. H. Meadows, D. L. Meadows, J. Randers, W. W. Behrens III. (1972): The Limits to Growth: A Report to The Club of Rome. Club of Rome
D. S. Moore (1997): New Pedagogy and New Content: The Case of Statistics. Int. Stat. Rev., 65, 2, 123-165
C. Weihs, J. Jessenberger (1999): Statistische Methoden zur Qualitätssicherung und -optimierung. Wiley-VCH, Weinheim
Keywords:
scenario, statistical education, e-learning, multimedia, design of experiments, chemical engineering
