Model | Description | s | k | STY or MET | Reference | |||
---|---|---|---|---|---|---|---|---|

Single trial | Diagonal (s = 1) | 1 | 0 | STY | ||||

DIAG | Diagonal | s | 0 | ^{b} | STY | |||

US | Unstructured | s | 0 | MET | ||||

CS | Compound symmetry | s | 1 | MET | Patterson et al. (1977) | |||

CS+DIAG | CS+DIAG | s | 1 | MET | Cullis et al. (1998) | |||

FAMk^{c}^{,}^{d} | Factor analytic (main effect) | s | MET | Smith et al. (2001) |

STY, single trial year (note the DIAG model is equivalent to analyzing each trial year separately); MET, multi-environment trial.

↵a A similar table could be constructed for with , and .

↵b ⊕ represents a kronecker sum, so that results in a diagonal matrix with elements for the specific variance of trial

*t*.↵c is a matrix of

*k*factor loadings at each of the*s*trials.↵d For FAM

*k*let where , var and .