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

May need to change split plot from aov to mixed effects #82

Closed moorepants closed 9 years ago

moorepants commented 9 years ago

See this discussion: https://groups.google.com/forum/#!topic/davis-rug/PnbK2ZDi9tM

msimmond commented 9 years ago

Yes, after consulting several sources we need to change the statistical approach for split-plot designs from fixed effect model using aov() to mixed effects model using lmer() or lme(). I'll check to see if the split-plot design code is still in the pull requests so I can edit it there.

msimmond commented 9 years ago

@ikkyle I'm not sure how to go about editing the two split-plot scenarios (CRD type and RCBD type). Can you add them to the pull requests? Or is there a new place to edit since a lot of those drafts were merged to the master code?

moorepants commented 9 years ago

If you go here: https://github.com/ucd-ipo/aip-analysis/blob/master/examples/split_plot_rcbd.R and then click the pencil you can edit the example. It will then give you an option to create a pull request or directly commit the changes to the master code. You can do that with any file in the repository.

moorepants commented 9 years ago

There is also this desktop program https://desktop.github.com/ that would let you edit the files on your Mac.

msimmond commented 9 years ago

Thanks, @moorepants.

rkingdc commented 9 years ago

This should be working now for split plot RCBD using nlme. I was a bit confused and didn't include split plot completely randomized design, but I'll work that in soon.

msimmond commented 9 years ago

Can you please merge the pull request that I made for the split plot RCBD, but with your changes to the nlme model? I'd like to have all the examples in their final form if possible.

rkingdc commented 9 years ago

Done. What should the split plot CRD formula look like?

msimmond commented 9 years ago

model <- lme(fixed = Y ~ A*B, random = ~1|Rep/A, data = my.data), where A is main-plot, B is sub-plot, and Rep is experimental unit for main-plot.

rkingdc commented 9 years ago

Done.