1. Basics
   1. Contrasts
      1. def. of contrast
      2. use of contrast
         1. hybrid yield
         2. comparison
      3. estimate
         1. gmodels
            1. fit.contrast()
            2. estimable()
         2. aov()
            1. model.tables()
      4. confidence interval
      5. hypothesis test
      6. SSw
         1. orthogonal contrast
         2. calculate SSw
      7. default: contr.treatment
   2. Multiple Comparisons
      1. multiple comparisons
         1. error rate
            1. simultaneous error rate
            2. comparisonwise error rate
            3. experimentwise error rate
            4. FDR
            5. SFW
         2. methods
            1. Benferroni
            2. linear.contrast()
            3. Holm
            4. SFW
            5. FDR
            6. FDR
            7. Scheffe
            8. for all contrasts
            9. simultaneous CIs for all contrasts
            10. linear.contrast()
      2. pairwise comparison
         1. methods
            1. Tukey HSD
            2. simultaneous CIs for all possible pairwise comparisons
            3. familywise error rate
            4. studentized range statistics
            5. require equal sample size
            6. aov(); TukeyHSD()
            7. SNK
            8. FDR
            9. studentized range statistics
            10. REGWR
            11. SFW
            12. studentized range statistics
            13. LSD
            14. experimentwise error rate
            15. Duncan's
            16. Dunnet's
            17. compare with control
            18. compare.to.control()
            19. glht()
            20. MCB
            21. compare with best
            22. compare.to.best()
            23. comparison after looking at data
            24. library(cfcdae); pairwise()
2. General Steps
   1. Check the design type
      1. Factorial Design
         1. step1: set contr.sum
         2. n=1 replicate
            1. half-normal plot to get the significant factors
            2. pseudo p-value
            3. pooling high-order interactions
      2. Complete Block Design
         1. CRB design are balanced
            1. effects are orthogonal
         2. block for variance reduction
         3. Latin Square Design
            1. two block variables
            2. replicating Latin Square
         4. design for residual effect
         5. Graeco Latin Square
            1. three blocking variables
   2. Check the data type
      1. unbalanced data
         1. SS type
            1. contr.sum
         2. contrast
            1. linear.contrast
            2. contr.sum + fit.contrast
         3. multiple comparison
            1. glht() to do TukeyHSD
         4. marginal means
            1. emmeans
         5. interaction plot
            1. emmeans
   3. Fit a model
   4. Check assumptions
      1. normality
         1. qqplot()
      2. constant variance
         1. diagnostics
            1. side-by-side boxplot
            2. residual vs. fitted
            3. diagnostic plots
         2. transformation
            1. power family
            2. boxcox
            3. other stabilizing methods
         3. balanced design preferred
      3. independence
         1. DW test
   5. Investigate effects
      1. interaction effects
         1. interaction plot
         2. pairwise comparison
            1. TukeyHSD
            2. underline diagram
      2. main effects
         1. profile plot
         2. pairwise comparison
            1. TukeyHSD
            2. underline diagram