friendly / VCDR

Visualizing categorical data with R
0 stars 0 forks source link

Pc #30

Closed philchalmers closed 9 years ago

philchalmers commented 9 years ago

Mostly typos here, but check the TODO's.

friendly commented 9 years ago

Thanks again. I made some comments re the TODOs on the first commit. About to merge pull request.

philchalmers commented 9 years ago

No problem. Did you see the second commit though? Had a few more TODOs, especially regarding the generated ggtern figures.

philchalmers commented 9 years ago
...but may be highly skewed in small to moderate samples.%
\footnote{When computing confidence intervals for models estimated via maximum likelihood, profiled-likelihood confidence intervals can be a suitable alternative to using the ASE estimates because they are more effective at capturing asymmetric intervals \citep{Pawitan:2001}.}

With the reference

@Book{Pawitan:2001,
  Title                    = {{In All Likelihood: Statistical Modelling and Inference Using Likelihood}},
  Author                   = {Yudi Pawitan},
  Year                     = {2001},
  Address                  = {New York}
}
friendly commented 9 years ago

I read the \TODOs on ggtern figs; Fig 4.21 is particularly ugly. I emailed Nick Hamilton on this issue

friendly commented 9 years ago

I'm thinking to remove the \TODO and suggested footnote, because while it is accurate, it seems overly pedantic, in that this paragraph was written to set up the motivation for using log odds ratios, described in the next paragraph.

...but may be highly skewed in small to moderate samples.% \footnote{When computing confidence intervals for models estimated via maximum likelihood, profiled-likelihood confidence intervals can be a suitable alternative to using the ASE estimates because they are more effective at capturing asymmetric intervals \citep{Pawitan:2001}.}

philchalmers commented 9 years ago

Understandable, I'm okay with whatever here. On Mar 26, 2015 2:46 PM, "Michael Friendly" notifications@github.com wrote:

I'm thinking to remove the \TODO and suggested footnote, because while it is accurate, it seems overly pedantic, in that this paragraph was written to set up the motivation for using log odds ratios, described in the next paragraph.

...but may be highly skewed in small to moderate samples.% \footnote{When computing confidence intervals for models estimated via maximum likelihood, profiled-likelihood confidence intervals can be a suitable alternative to using the ASE estimates because they are more effective at capturing asymmetric intervals \citep{Pawitan:2001}.}

— Reply to this email directly or view it on GitHub https://github.com/friendly/VCDR/pull/30#issuecomment-86664918.