What is a Latin Square design?
You have chosen your experimental design. On this page you will choose your treatment design by being led through a series of questions.
Identifying your treatment design is affected by the number of treatment factors, how those treatment factors are combined, and what experimental units are associated with each of the treatment factors. The questions below address these three determinants.
If this "sounds like Greek" to you, some background reading is strongly recommended. E.g. see Glossary or do some reading from these resources.
You can see illustrations of the treatment designs in the gray table in the ANOVA tab.
Treatment Design Selection Key
1. Do you have more than one treatment factor?
Yes >> Go to red arrow 2 below.
No >> You have a Latin Square Single Factor design
(example) (What is Single Factor?)
2. Are all levels of one factor combined with all levels of the other factor(s)
to form the treatment combinations in the experiment?
Yes >> Go to red arrow 3 below.
No >> You have a Latin Square Nested design (What is Nested?)
3. Are the treatment factors applied to different types of experimental
units (e.g. sub-plot or whole plot)?
Yes: you have some kind of "split-plot" >> Go to red arrow 4 below.
No >> You have a Latin Square Factorial design (What is Factorial?)
4. Are the large experimental units physically divided into smaller
experimental units (2 types of experimental units only)?
Yes >> You have a Latin Square Split-Plot design (What is Split-Plot?)
No >> Go to red arrow 5 below.
5. Do you have 2 types of experimental units as in question 4, but the smaller
experimental units are created by taking measurements over time or space
(so no physical divisions, and the levels can not be randomized, eg. time 1
must always come first)?
Yes >> You have a Latin Square Repeated Measures design (What is Repeated Measures?)
No >> Go to red arrow 6 below.
6. Are the smaller experimental units in question 4 physically divided again
to create another type of experimental unit (3 types of experimental units)?
Yes >> You have a Latin Square Split-Split-Plot design (What is Split-Split-Plot?)
No >> Go to red arrow 7 below.
7. The only choice remaining: you must have treatment factors that are both
applied to large experimental units, which are physically perpendicular to
each other to create all the treatment combinations.
Yes >> You have a Latin Square Strip-Plot design (What is Strip-Plot?)