Nested Designs and Latin Squares - Latin Squares, Greco-Latin Squares, Generalized Latin Squares
A general class of balanced nested designs are the Latin squares (see for example Hays, 1988; Lindman, 1974; Milliken & Johnson, 1984; Winer, 1962; the term Latin square was first used by Euler, 1782). An example of a Latin square design is also discussed in the Introductory Overview. In nested designs, the combination of treatment levels is arranged so as to yield unconfounded main effect estimates with a minimum number of observation.
In Latin Squares designs, only main effects can be estimated. Therefore, to analyze these designs, simply select Main Effects ANOVA from the General ANOVA/MANOVA Startup Panel - Quick tab or GLM Startup Panel Quick tab. Then select the factors of the Latin squares design as Categorical predictor variables (via the Variables button the ANOVA/MANOVA Quick Specs - Quick tab or GLM Quick Specs Dialog - Quick tab. On the GLM and ANOVA Results - Summary tab, click on the All effects button to review the ANOVA/MANOVA results.