MerQuR+Statistics Glossary

Statistics Glossary → Compact letter display (homogeneous subsets)

Compact letter display (homogeneous subsets)concept

A method that turns pairwise comparison results into a short display in which groups that do not differ significantly share a common letter.

When many groups are compared, the table of pairwise results becomes long and hard to read. In a compact letter display each group gets one or more letters: two groups sharing a letter are not significantly different according to the chosen post hoc test, while groups with no letter in common are significantly different. SPSS's homogeneous subsets table presents the same idea in another form.

A group can carry two letters, such as 'ab': it does not differ from the 'a' groups or the 'b' groups, although those two sets differ from each other. Non-significance does not prove that there is no difference; a shared letter only says that a difference could not be shown with these data. The letters depend on which post hoc method and adjustment were chosen.

An everyday example

Yields of five wheat varieties are compared and the table gives A 'a', B 'a', C 'ab', D 'b' and E 'c'. So A and B are similar, C cannot be distinguished from either A or D, A and D differ, and E differs from all the others.

In MerQur

In One-Way ANOVA, the 'Extra output: Letter display (homogeneous subsets)' checkbox builds it from the significant pairs of the selected post hoc test; it is also given as the result of the group-pair z-tests in the MR × Categorical Crosstab and in the group proportion comparisons of the Crosstab.

Analyses where this concept appears

One-Way ANOVACrosstabMR × Categorical Crosstab

Related concepts

Post hoc tests (pairwise comparisons)Tukey's HSD test