Dear Rebecca,
A few things in your query confuse me.
The .pvs file will contain retransformed predicted values.
e.g.
GRP Logit_value Stand_Error Ecode Retransformed_value approx_SE
1.0000 1.4987 0.1450 E 0.8174 0.0227
2.0000 0.7591 0.1324 E 0.6812 0.0294
3.0000 -0.3179 0.2200 E 0.4212 0.0525
In this example,
1/(1+exp(-1.4987))
[1] 0.8173805
So, if you have implausible predicted values, there must be a problem with the way you are combining the fitted effects. These effects are printed in the .sln file (in simple models also in the .asr file).
The way you describe the model, I would assume you have defined 'treatment' and 'situation' as factors. But the way you have written the predict calculation makes it look like they are covariates. If so, this would explain getting different values from PREDICT when compared with your hand calc.
Without seeing the model and effects, I can't tell what the problem is but in a simple case, it should be straight forward to calculated the underlying predicted value manually. I'm happy to look closer if you want to send me the details.
------------------------
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=545#545
-------------------- m2f --------------------
Received on Wed May 19 2009 - 10:30:58 EST
This webpage is part of the ASReml-l discussion list archives 2004-2010. More information on ASReml can be found at the VSN website. This discussion list is now deprecated - please use the VSN forum for discussion on ASReml. (These online archives were generated using the hypermail package.)