Split Plot ANOVA Calculator

Split plot ANOVA online. Two error strata, main and sub plot F tests, CV percent and post-hoc.

When to use it

Use when one factor (the main plot) is much harder to randomise than the other (the sub plot). Classical examples: irrigation regime on whole plots with varieties on sub plots, tillage system on whole plots with seeding rate on sub plots, processing batch on whole plots with formulation on sub plots.

When NOT to use it

Do not use when both factors can be randomised independently at the same scale; that is a factorial RBD (more powerful for both factors). Do not use when the main-plot factor is the one of primary interest at high precision (split-plot deliberately gives the main plot less precision than the sub plot).

What you get

Two-tier ANOVA: Block, A (main plot), Error a, B (sub plot), A x B, Error b. Three F tests using the appropriate error term each time. Two CD values: CD_A uses Error a, CD_B and CD_AxB use Error b. CV percent is reported per stratum. Post-hoc letter display per significant effect.

Worked example

3 main-plot by 2 sub-plot by 3 blocks demo dataset.

25.4, 27.1
31.2, 29.8
22.1, 23.4
24.9, 26.3
30.5, 28.7
21.8, 24.1
26.8, 28.2
32.1, 30.5
23.4, 25.1

Source: StatVeda built-in example (not from a published textbook).

How to interpret the output

Sub-plot factor B and the A x B interaction are tested with higher precision than the main-plot factor A. If A x B is significant, interpret cell means with CD_AxB; the main-effect tables can be misleading. The hierarchical randomization is what creates the two error terms; if you do not randomise hierarchically the analysis is wrong (Cochran and Cox 1957 ch. 7).

Common pitfalls

Mixing the two error terms for post-hoc (CD_A uses Error a only, never Error b).
Treating A as the high-precision factor; it is deliberately less precise than B.
Forgetting that randomization happens twice: A within block, then B within each main plot.
Reporting a single CV percent (split-plot has two error strata, hence two CVs).

Try it in StatVeda

The Split-Plot Design engine runs entirely in the browser. No signup, no install, no data sent to a server. Paste your data, hit Run, copy the output.

Open Split-Plot Design

References

Cochran, W. G. and Cox, G. M. (1957). Experimental Designs, 2nd edition. John Wiley and Sons, New York. Chapter 7.
Gomez, K. A. and Gomez, A. A. (1984). Statistical Procedures for Agricultural Research. John Wiley and Sons, New York. Chapter 4.
Federer, W. T. (1955). Experimental Design: Theory and Application. Macmillan, New York.