Db/adddatadir

[dmbates]

• add data directory
• add two longitudinal data sets
  • Oxboys.arrow from the nlme package for R
  • box.arrow from Davis(2002) "Statistical Methods for the Analysis of Repeated Measurements" and G.E.P. Box (1950)

[kliegl]

@dmbates : elstongrizzle.jl : That's a nice set of figures. • Would it be possible to set the x-tick labels to 8.0, 8.25, ... 9.5 on the first and perhaps to .50 increments on the second graph? • The y-axes could also be improved in this respect, I think. • In the caterpillar plots it would be nice to add vlines at the origins /zeroes.
• Similarly, I would propose to add hlines and vlines for shrinkage plots • Finally, also for shrinkage plots, does AoG allow selective labeling of isolated point with, say, the subject number?

[kliegl]

@dmbates : Oxboys.jl • I would propose to use only lower-case letters for notebook and data names -- oxboys.jl and oxboys.arrow
• Are there metadata for these data? • I don't think we should start with scaled scores such as age here, but carry out scaling in the script -- at least when we know the original metric. I also think that the descriptive plots at the beginning should show the original ages. Do we know them for Oxboys.? • Perhaps add a sentence of explanation when introducing functions not well-known for migrants from other computing environments, e.g. only(). • Would it be useful to add the zero-lines for caterpillar and shrinkage plots by default in MixedModelsMakie?
• I like this geomdof(m::LinearMixedModel) = sum(leverage(m)); way of defining new functions. Not sure where, but this could be a good example to introduce a Julia-specific feature.

[kliegl]

@dmbates : box.jl - nice data • m1 has a misplaced parenthesis (w/o consequence); it should be: form = @formula(resp ~ 1 + (time + time^2) * Group + (1+time+time^2|Subj)) • Add an LRT for test of interactions removed when going from m1 to m2? • This was mentioned in a different context, but I want to second the proposal to include the n of groups and n of obs in the default model output. For me this always serves as a quick automatic check whether the basics are correct.

[dmbates]

@dmbates : Oxboys.jl • I would propose to use only lower-case letters for notebook and data names -- oxboys.jl and oxboys.arrow • Are there metadata for these data? • I don't think we should start with scaled scores such as age here, but carry out scaling in the script -- at least when we know the original metric. I also think that the descriptive plots at the beginning should show the original ages. Do we know them for Oxboys.? • Perhaps add a sentence of explanation when introducing functions not well-known for migrants from other computing environments, e.g. only(). • Would it be useful to add the zero-lines for caterpillar and shrinkage plots by default in MixedModelsMakie? • I like this geomdof(m::LinearMixedModel) = sum(leverage(m)); way of defining new functions. Not sure where, but this could be a good example to introduce a Julia-specific feature.

Thanks for the suggestions. Could you take a look at the renamed notebook and data set that I just uploaded? It doesn't meet all your suggestions but I incorporated those that I could. In Pinheiro and Bates (2000) we stated that the data appear in Goldstein 1987 without an indication of the scaling of the age. Apparently the book is available in our main library so I will pick it up and check for the data there. It might be part of the comparison of multilevel modeling software test suite.

[dmbates]

It turns out that there is an additional reference for oxboys at http://www.bristol.ac.uk/cmm/learning/mmsoftware/data-rev.html#oxboys where they state that the boys' ages were from 11 to 13 years and the zero point is 12 years, so it looks like we can relabel them.

[palday]

I think this is outdated with the way we're now hosting data on OSF, so I'm going to close this. We can re-open it if it turns out that I was wrong!