Expert experimental design with

A total of eleven variables including inulin, initial pH, yeast extract, fermentation time, peptone, temperature, liquid medium volume, inoculum concentration and three dummies or unassigned variables were studied in twelve experiments. In addition, three dummies were used as the variables that could be assigned numerical values. The experimental design with the name, symbol code, and actual level of the variables is shown in Table 1. The analysis of variance (ANOVA) for the data and the model coefficients were computed with Design-Expert 8.0.6. 

Nancy Sharkey, Irene Clark, Suzanne Patterson, Patricia P. Dobson and Howard Schwartz and the secretarial efforts of Ms. Madle Tyler. A special thanks to Dr. Kurt W. Kohn, Chief of the Laboratory of Molecular Pharmacology, who has supported all of this work with advice In experimental design and expert data analysis and Interpretation, as well as valuable critique of this and all past manuscripts. 

The situation is not hopeless however, it does require that a more complicated approach be used. In particular, it is necessary that suitable data be collected that will allow a direct calculation of the error due to any particular source. This in turn requires that data should be collected according to a statistically designed scheme, so that the source of variation to be studied can be separated from the others. One example of how this approach can be applied to NIR data to extract the values of several sources of variation at the same time has appeared in the literature [7]. Statistical texts are available that describe a number of other experimental designs that may be applied. For unusual circumstances it may be necessary to consult a statistician such a consultation is a good idea anyway, as statistical methodologies are fraught with pitfalls to trap the unwary and inexperienced, and an expert guide is invaluable. 

In order to choose an appropriate array, degrees of freedom must be computed first. Given seven factors with three-levels, one factor with two-levels, and one interaction of a two-level and a three-level factor, the number of degrees of freedom for the experiment is computed to be 7(3-1) + 1(2-1) + (3-l)(2-l) +1 = 18. Table 10.10 shows the experimental design for an LIS array. In all, 18 treatments must be used for the experiment with the factor levels as shown. For this study, however, three levels of risk are used in Table 10.11 representing judgements made by three experts. 

Educational TV in India Experimental TV Service was inaugurated on September 15th, 1959. This programme was planned for educational and cultural values of community and designed for community viewing. In the meantime Ford Foundation in India was approached to assist in the development of educational TV. Ford Foundation team of TV experts visited India in February 1960. In 1961 Television was started in India as an educational TV project. The project was launched in close collaboration with the Delhi Directorate of Education and with the financial assistance of the Ford Foundation in 1961. 

The experiment design involves selecting multiple levels of each factor to define the conditions for each experimental run, and some subject matter expert/end-user knowledge is required to select the range of the factor levels to cover. As the optimum level of the response is not known in advance, at least five factor levels should be used, and the range of these levels should be as wide as possible with the low and high levels determined by the end user s input of feasibility and safety considerations. 

---