Statistical analysis and visualization of the drug war in Mexico

Christian points me to this interesting (but sad) analysis by Diego Valle with an impressive series of graphs. There are a few things I’d change (notably the R default settings which result in ridiculously over-indexed y-axes, as well as axes for homicide rates which should (but do not) go town to zero (and sometimes, bizarrely, go negative), and a lack of coherent ordering of the 32 states (including D.F.),

I’m no expert on Mexico (despite having coauthored a paper on Mexican politics) so I’ll leave it to others to evaluate the substantive claims in Valle’s blog. Just looking at what he’s done, though, it seems impressive to me. To put it another way, it’s like something Nate Silver might do.

1 thought on “Statistical analysis and visualization of the drug war in Mexico

  1. Would you really call those y-axes over-indexed? Most of them have 5 or 6 labeled tick-marks, which doesn't seem a lot to me. They look to be made using ggplot so presumably use the defaults in that package rather than from R.

    The main gripe I'd have is with the grouped plots with free-y-axes like this one. That can be changed with a simple flag in the plot command.

    Agree with the negative y-axes problem — it looks to be caused by plotting the error of the best-fit line.

Comments are closed.