Variations of One Factor ANOVA Designs - Agricultural Statistical Data Analysis Using Stata - page 80

Each of the first five lines are calculated local variables or, in the
vernacular of Stata, local macros. The last line displays the results
of the last calculated macro (RE). If you are familiar with other
programming languages, Stata's use of macro is somewhat differ-
ent. he local term is required to distinguish these macros (variables)
from global macros. The values assigned to these macros are calcu-
lated from the list of scalars (numbers) from the previously executed
estimation command ( anova ) and listed with the ereturn list
command. You will notice in the last two lines that these local macros
have `and ' quotes around them. This is different from other program-
ming languages as it indicates to the Stata program to use the value
of the macro rather than the macro name. With the ` and ' around `r',
it recognizes the value (3); without the ` and ' it would recognize it as
r. The ` is an accent mark located at the upper left side of most key-
boards. The other (') is a closed quote found near the return key. We
will look more closely at this in Chapter 7 on programming.
It is important to remember that the values listed in the ereturn
list are only available until the next estimation command is exe-
cuted. If you were to calculate another ANOVA, the values would
change to the new estimation.
In this case, the R.E. is 1.0022889, which is quite small. It means
by using an RCBD instead of the CRD we are only seeing 0.2%
increase in efficiency. This will not always be the case; in fact, the
relative efficiency can be quite large in some cases. Finally, if the error
degrees of freedom or Residual degrees of freedom is below 20, a cor-
rection factor should be calculated and multiplied against the relative
efficiency. In this case, with an error degrees of freedom of 38, it is
not necessary; however, for those cases where it would be required the
correction factor is

( ) ( ) ( ) +

r − 1 t − 1 tr − 13

k =

( ) ( ) + ( ) +

r − 1 t − 1 3 tr 11 −

Latin Square Designs

Latin square (LS) designs add another source of variation and, hence,
consist of both row and column variations. This design requires that
the number of treatments equal the number of rows and columns in