search

ucd-ipo / agroft

Agricultural Field Trial Statistics Package
http://ucd-ipo.github.io/agroft/
BSD 3-Clause "New" or "Revised" License
4 stars 5 forks source link

`yield.pred.sq` row missing from Tukey test? #90

Closed rkingdc closed 8 years ago

rkingdc commented 9 years ago

Using the following parameters,

Data: split-plot design Experimental Design: Split-Plot RCBD DV: yield IVs: main plot treatment: A sub plot treatment: plot blocking factor variable: block Transformations: None

Under Tukey's Test for Nonadditivity, a message says to "refer to only the 'yield.pred.sq' row in this table...", but there is no such row in the table. Only a row for 'A', 'plot', and 'Residuals'. Is this an error?

moorepants commented 9 years ago

The row should be there if it runs the Tukey test. It is an error if not.

msimmond commented 9 years ago

I deleted the Tukey 1df Test for the split-plot RCBD example (see pull request), as I think we should only use it for the basic RCBD with a single treatment rep per block.

rkingdc commented 9 years ago

In the ANOVA table output, we also get the error 

Error in MEEM(object, conLin, control$niterEM) : 
Singularity in backsolve at level 0, block 1

I'm pretty sure this has to do with the data not being nested but the model is trying to analyze it as if the data are nested. I'll need to implement an error handler that is more informative for this issue.

rkingdc commented 8 years ago

@msimmond I just want to confirm that we don't want to run the tukey test for split plot designs?

msimmond commented 8 years ago

That's correct.

rkingdc commented 8 years ago

@msimmond Tukey 1-df test is for 2-way ANOVA only, correct? So should we only be doing this for RCBD2 and CRD2?

msimmond commented 8 years ago

Tukey 1-df test is for the RCBD tests only (i.e. RCBD1 and RCBD2). Not for any CRD or split-plot.

Maegen Simmonds Postdoctoral Scholar Earth Systems Science Stanford University Green Blg, Rm 321 Stanford, CA 94305 Phone: (707) 694-6079 mbsimmonds@stanford.edu

On Tue, May 10, 2016 at 5:06 PM, Ian K. Kyle notifications@github.com wrote:

@msimmond https://github.com/msimmond Tukey 1-df test is for 2-way ANOVA only, correct? So should we only be doing this for RCBD2 and CRD2?

— You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub https://github.com/ucd-ipo/aip-analysis/issues/90#issuecomment-218327707

rkingdc commented 8 years ago

fixed