You may find these of help

http://www2.hawaii.edu/~taylor/z631/multcomp.pdf

Overview of methods

There are many methods for multiple-comparisons. Most are for pairwise comparisons of

group means, to determine which are significantly different from which others. Other

methods, though, are for more specialized purposes (e.g. comparing each of several treatments

to a control) or allow testing of more-general hypotheses contrasting sets of group

means (as is often done in preplanned “contrasts”).

The various methods differ in how well they properly control the overall significance level

and in their relative power; some, such as the popular “Duncan’s multiple range test” do not

control the overall significance level. The ones described in this handout all adequately

control overall significance and are either easy to use or powerful. They are:

• Bonferroni extremely general and simple, but often not powerful

• Tukey’s the best for all-possible pairwise comparisons when sample sizes are

unequal or confidence intervals are needed; very good even with equal

samples sizes without confidence intervals

• stepdown the most powerful for all possible pairwise comparisons when sample sizes

are equal

• Dunnett’s for comparing one sample (“control”) to each of the others, but not comparing

the others to each other.

Multiple-Comparison Procedures 2 of 9

• MCB compares each mean to the “best” (largest or smallest, as you specify) of

the other means.

• Scheffé’s for unplanned contrasts among sets of means

When more than one method is applicable, and it is not clear which is more efficient (i.e. will

give the narrower CI, for given α), it is legitimate to try all the applicable methods and use the

one which proves most efficient for the given data.

You may find these of help

http://www2.hawaii.edu/~taylor/z631/multcomp.pdf

Overview of methods

There are many methods for multiple-comparisons. Most are for pairwise comparisons of

group means, to determine which are significantly different from which others. Other

methods, though, are for more specialized purposes (e.g. comparing each of several treatments

to a control) or allow testing of more-general hypotheses contrasting sets of group

means (as is often done in preplanned “contrasts”).

The various methods differ in how well they properly control the overall significance level

and in their relative power; some, such as the popular “Duncan’s multiple range test” do not

control the overall significance level. The ones described in this handout all adequately

control overall significance and are either easy to use or powerful. They are:

• Bonferroni extremely general and simple, but often not powerful

• Tukey’s the best for all-possible pairwise comparisons when sample sizes are

unequal or confidence intervals are needed; very good even with equal

samples sizes without confidence intervals

• stepdown the most powerful for all possible pairwise comparisons when sample sizes

are equal

• Dunnett’s for comparing one sample (“control”) to each of the others, but not comparing

the others to each other.

Multiple-Comparison Procedures 2 of 9

• MCB compares each mean to the “best” (largest or smallest, as you specify) of

the other means.

• Scheffé’s for unplanned contrasts among sets of means

When more than one method is applicable, and it is not clear which is more efficient (i.e. will

give the narrower CI, for given α), it is legitimate to try all the applicable methods and use the

one which proves most efficient for the given data.

I appreciate your help, thanks.

My vote is for Tukey's Honest Significant Difference test

I have done BALANCE ANOVA for four different plant treatmentS , three different sugars and six different time intervals