498 CHAPTER 13 DESIGN AND ANALYSIS OF SINGLE-FACTOR EXPERIMENTS: THE ANALYSIS OF VARIANCEFigure 13-9 presents the results graphically. The underlined pairs of means are not different.
The LSD procedure indicates that chemical type 4 results in significantly different strengths
than the other three types do. Chemical types 2 and 3 do not differ, and types 1 and 3 do not
differ. There may be a small difference in strength between types 1 and 2.13-4.3 Residual Analysis and Model CheckingIn any designed experiment, it is always important to examine the residuals and to check for
violation of basic assumptions that could invalidate the results. As usual, the residuals for the
randomized complete block design are just the difference between the observed and estimated
(or fitted) values from the statistical model, say,(13-33)
and the fitted values areThe fitted value represents the estimate of the mean response when the ith treatment is run in
the jth block. The residuals from the chemical type experiment are shown in Table 13-15.
Figures 13-10, 13-11, 13-12, and 13-13 present the important residual plots for the ex-
periment. These residual plots are usually constructed by computer software packages. There
is some indication that fabric sample (block) 3 has greater variability in strength when treated
with the four chemicals than the other samples. Chemical type 4, which provides the greatest
strength, also has somewhat more variability in strength. Followup experiments may be nec-
essary to confirm these findings, if they are potentially important.13-4.4 Randomized Complete Block Design with Random Factors
(CD Only)yˆijyi.y.jy..eijyijyˆij0 1234 61 3 245Chemical typeFigure 13-9 Results of Fisher’s LSD method.Table 13-15 Residuals from the Randomized Complete Block DesignChemical Fabric Sample
Type 1 2 3 4 5
1 0.18 0.10 0.44 0.18 0.02
2 0.10 0.08 0.28 0.00 0.10
3 0.08 0.24 0.30 0.12 0.02
4 0.00 0.28 0.48 0.30 0.10c 13 .qxd 5/8/02 9:21 PM Page 498 RK UL 6 RK UL 6:Desktop Folder:TEMP WORK:PQ220 MONT 8/5/2002:Ch 13: