Agriculture Reference
In-Depth Information
Analysis of covariance can be used in a number of different situations.
It can be used to estimate missing data, to control experimental error,
to adjust treatment means, and as an aid to experimental interpretation.
In controlling experimental error, covariance analysis introduces the
covariate, which is considered to have an effect on the dependent vari-
able. This effect, when removed, will generally lower the mean square
error or residual. Load the dataset Covariance.dta. This is a dataset of
a lima bean variety trial with 11 varieties arranged as an RCBD with
four replications (Steel and Torrie, 1980, p. 412). Enter the command
anova
ascorbic var rep
then enter the command
anova
ascorbic var rep
c.
cov
This results in the two anova tables:
Number of obs = 55 R-squared = 0.9040
Root MSE = 12.1935 Adj R-squared = 0.8704
Source | Partial SS df MS F Prob > F
----------+---------------------------------------------------
Model | 55987.1188 14 3999.07991 26.90 0.0000
|
var | 51018.1786 10 5101.81786 34.31 0.0000
rep | 4968.94012 4 1242.23503 8.35 0.0001
|
Residual | 5947.30397 40 148.682599
----------+---------------------------------------------------
Total | 61934.4227 54 1146.93375
Number of obs = 55 R-squared = 0.9644
Root MSE = 7.51657 Adj R-squared = 0.9507
Source | Partial SS df MS F Prob > F
---------+----------------------------------------------------
Model | 59730.9684 15 3982.06456 70.48 0.0000
|
var | 7457.62247 10 745.762247 13.20 0.0000
rep | 756.392711 4 189.098178 3.35 0.0190
cov | 3743.84965 1 3743.84965 66.26 0.0000
|
Residual | 2203.45433 39 56.4988289
---------+----------------------------------------------------
Total | 61934.4227 54 1146.93375
Search WWH ::
Custom Search