in R / statistics / model selection / GLM / AIC / deviance on Mon 26 October 2015

Here's what you need to report in a paper about the model comparison:

  • residual deviance
  • residual df
  • delta AIC
  • AIC weight

You should also report the null deviance and degrees of freedom, maybe in a table caption.

Thanks to Emilio Bruna for prompting this post and suggesting its title ...

in R / statistical graphics / visualization / ggplot2 / lattice on Tue 28 April 2015

It's a long tradition in statistical graphics going from Tufte back to Tukey and Cleveland to advise against using bar charts. Many folks, including me, have pejoratively called the common (in ecology, at least) bar chart + SE a "dynamite plot". Although Ben Bolker has questioned the wisdom of this ...

In a live walk-through on April 10 at the Davis R-Users Group, I gave a brief presentation motivating this topic. This post expands and cleans up the code from that talk. If you just want the code: download the .R file.

Updated 2015-04-28

The point of this post isn't ...