(Exercises 12 to 15 in Advanced Copyediting Practice for Chinese Technical Writers by Ted Knoy)
 [ Exercise 12 ] It is a fact that engineers select an appropriate variable and the transformed observations are treated as though they are normally distributed with a constant variance. Those methods neither require previous knowledge of how the variables are distributed nor are the censored data stipulated to be available. The procedure for analyzing singly censored data in a replicated experiment is as follows: S tep 1: Distinguish the experimental results as the uncensored (complete) data and the censored (incomplete) data. Step 2: The relationship between the two values must be found by performing regression analysis. Step 3: Estimate the two variables. Step 4: The estimated censored data must be ranked. Step 5: Find the regression models for response avaerage and standard deviation for each trial. Step 6: The factors that significantly affect the response average and standard deviation must be identified. Step 7: The optimal factor/level combination must be determined. The derived model provides an extension of an earlier concept  and helping industrial managers in determining a feasible number of replenishments. Experimental design is used in this method to arrange the design parameters and noise factors in the orthogonal arrays and computing the signal-to-noise (SN) ratio based on the quality loss for each experimental combination. The relative importance of each response can be transformed into a fuzzy number through means of the establishment of a formal scale system that can be used to convert linguistic terms into their corresponding fuzzy numbers and to express the relative importance of each response by the linguistic term. The Taguchi approach provides a combination of experimental design techniques with quality loss considerations and that the average quadratic loss is minimized. The conventional approach happens to be cumbersome, complicated and wastes too much time. The two-step procedure not only identifies those factors that significantly affect the signal-to-noise (SN) ratio, but also the levels that maximize SN are found. Logethetis (1988) proved that strong non-linearities exist and the B technique was also recommended for use by him. This work not only proposes an effective procedure based on the rank transformation of responses and regression analysis, but also the singly censored data are discussed. The following steps describe the procedure: Step 1: Calculate the normalized decision matrix. Step 2: The weighted normalized decision matrix is calculated. Step 3: The ideal and negative-ideal solution is determined. Step 4: Calculate the separation measures. Step 5: The relative closeness to the ideal solution is calculated. Step 6: The preference order is ranked.