Variations of One Factor ANOVA Designs - Agricultural Statistical Data Analysis Using Stata - page 86

Enter the following command and look at the resulting output:
anova root run temp, sequential
Number of obs = 12 R-squared = 0.9544
Root MSE = .074633 Adj R-squared = 0.8996
Source | Seq. SS df MS F Prob > F
-------+----------------------------------------------------
Model | .582504477 6 .09708408 17.43 0.0033
|
run | .115075577 3 .038358526 6.89 0.0317
temp | .467428901 3 .155809634 27.97 0.0015
|
Residual | .027850417 5 .005570083
---------+----------------------------------------------------
Total | .610354894 11 .055486809
It is important to enter the command exactly as listed above includ-
ing the order of the variables. A sequential sum of squares is often
referred to as a type I sum of squares and the partial sum of squares
as the type III sum of squares or sometimes as the adjusted sum of
squares. Because not all observations occur simultaneously (we have
only three growth chambers and four treatments), the order of the
calculations is important. The sum of squares accounted for by tem-
perature is calculated after taking into account the sum of squares
for the runs. Normally the anova command defaults to the partial
sums of squares where the order of the independent variables does
not matter. Since the anova command defaults to the partial sums
of squares, it does not have to be explicitly listed as an option. The
sequential sum of squares calculated the first sum of squares, which
influences the subsequent calculation. Try it for yourself by revers-
ing the order of the run and temp variables. In addition, whether the
sequential option is specified or not will change the results. From
this ANOVA the germination temperature is significant, but the run
is significant as well. Because of this, the least squares means should
be reported rather than the arithmetic means when reporting these
results. To calculate the least squares or marginal means enter the
following command:
margins temp