One-way layout with fixed effects (comparison of treatments with no blocking): (i) Use model (2.1) with either of the constraints Σk 1𝜏i = 0 (zero sum) or 𝜏1 = 0 (baseline). Interpretation of the...

One-way layout with fixed effects (comparison of treatments with no blocking):

(i) Use model (2.1) with either of the constraints Σk

1𝜏i = 0 (zero sum) or 𝜏1 =

0 (baseline). Interpretation of the 𝜏i for each constraint can be found in

Section 2.1.

(ii) Use the ANOVA table in Table 2.2 and the F test in (2.7) for testing the

null hypothesis: 𝜏1 = 𝜏2 = = 𝜏k.

(iii) If the null hypothesis is rejected, multiple comparisons (Section 2.2) of the

𝜏i’s should be considered. The Tukey method in (2.20) is recommended.

The Bonferroni method in (2.14) can be used in very general situations. It is

recommended in situations where the critical values for the Tukey method

are not available.