Agriculture Reference
In-Depth Information
discussed in more detail in Chapter 7. Suffice it to say, this is one of
the great strengths of Stata—its extensibility.
local Er = e(ss_2)/e(df_2)/*Row mean square*/
local Ec = e(ss_3)/e(df_3)/*Column mean square*/
local t = e(df_1)+1/*Number of treatments*/
local Ee = e(rss)/e(df_r)/*Error mean square*/
local k = ([(`t'-1)*(`t'-2)+1]*[(`t'-1)^2+3])/([(`t'-1)*
(`t'-2)+3]*[(`t'-1)^2+1])/*Correction factor*/
local RE = (`Er' + `Ec' + (`t'-1)*`Ee')/((`t' + 1)*`Ee')*`k'
local RER = (`Er' + (`t'-1)*`Ee')/(`t' * `Ee')*`k'
local REC = (`Ec' + (`t'-1)*`Ee')/(`t' * `Ee')*`k'
display `k'
display `RE'
display `RER'
display `REC'
Remember, as before, the scalars are only available from the most
recently executed estimation command. The results of executing this
Do-File are a correction factor of 0.93 and relative efficiencies of
4.20, 1.14, and 4.81 comparing the LS to the CRD, RCBD (rows),
and RCBD (columns), respectively. This means that the LS design is
320% more efficient than the CRD or the CRD would require 3.2
times more replications to attain the efficiency of the LS design. In
addition, the LS design has increased precision with the row blocking
of 14% and column blocking of 381%.
With small LS designs two identical experiments can be conducted
and the results analyzed together. Load the dataset Latin square 2.dta
for the next analysis. This dataset represents an experiment with three
different initial fertilizer applications as part of an overall fertiliza-
tion program with direct seeded onions. The initial fertilizer treat-
ments were 0 fertilizer, 150 lbs/acre calcium nitrate (Ca(NO
3
)
2
), and
200 lbs/acre diammonium phosphate ((NH
4
)
2
H
2
PO
4
). Each treat-
ment was applied to a single row, so, in addition to the replication
source of variation, there were individual fertilizer hoppers.
Enter the following command to analyze these data. Note the ver-
tical bars in this command. This key (
|
) is usually found just above the
return key on most keyboards:
anova
yield exp rep|exp hop|exp trt
Search WWH ::
Custom Search