Re: Contrasts

Dear Jason,

Let me join the discussion.

To calculate the contrast "Variety_Marvellous vs. Variety_Victory in Nitrogen_0.2_cwt" is the interaction of a covariable '0 1 -1 derived from the coding of Variety (assuming order GoldenRain Marvelous Victory) and a covariate 0 1 0 0 derived from the coding of Nitrogen (assuming order 0cwt 0.2cwt 0.4cwt 0.6cwt)

Brian provide code to create the first contrast.
In his interaction example, he calculated the contrast for all 4 levels of Nitrogen.

ASReml-r has a model function at() which creates the single contrast.

So temp.asr <- asreml(yield ~ v1 + Variety + at(Nitrogen,2).v1 + v1:Nitrogen + Variety:Nitrogen,
                   random = ~ Blocks/Wplots, data=temp)

should separate out that particular contrast.

However, if you have multiple contrasts, you need to be careful in interppreting them;

One approach would be to calculate the predicted Variety:Nitrogen means
with their variance matrix, and then do a t test of the difference between the
mean for Marvelous:0.2cwt and Victory:0.2cwt

In this balanced example, this should give the same result but in general,
the test of a regression coefficient, an an ANOVA statistic may give different results.

I trust this helps explain the principles behind the calculation.

------------------------
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)

-------------------- m2f --------------------

Sent using Mail2Forum (http://www.mail2forum.com).

Read this topic online here:
http://www.vsni.co.uk/forum/viewtopic.php?p=647#647

-------------------- m2f --------------------
Received on Sat Jul 10 2009 - 04:33:50 EST