Studies on the Computerization of Design of Experiments(I)

실험계획법의 전산화에 관한 연구(I)

This paper studies the handling of significant digits and rounding off methods in domestic industries. ANOVA tables made by six well-known big companies are selected and analyzed. There exist various mistakes in handling of significant digits and rounding off methods such as: * too many significant digits in the Sum of Squares values in comparison to the original data * too many significant digits in the variance ratio in comparison to the F table values. * no consistancy in the number of significant digits * no consideration for the number of significant digits in computations * ignoring the KS A 0021 in rounding off methods etc. Such mistakes are caused from the characteristics of the personal computers rather than the misunderstandings about the significant digits conception. A subroutine is developed for PC in BASIC language to help the handling of significant digits and rounding off.