Comments on Diego Valle's Blog: Homicides in Mexico 2006-2009

Thomas, 1)I should have linked to my previous pos...

2011-02-02T11:07:29.685-06:00

Thomas,

1)I should have linked to my previous post on firearm homicides or at least to this chart. I should also probably switch to displaying the proportion in monthly increments instead of yearly to make the pattern more clear.

2)You're absolutely right, it's been corrected. Thanks for the feedback!

This is some very interesting data exploration, an...

2011-02-02T01:39:44.195-06:00

This is some very interesting data exploration, and good use of ggplot2. It's rather depressing how negative an impact the drug war seems to have had on Mexico.

I would like to offer a couple of constructive critiques, though.

First, you note that the rate of increase in percent of homicides by firearms seems to have slowed in the past year. To me, this looks more like noise, and might be measurement error rather than a real drop. Even if it's not a measurement error, your conclusion implies a trend, which the data does not appear to support. Unless there is a predictive model that predicts a slow-down in the increase and is supported by this data, we would be better to comment that there is year-on-year variation in the rate of increase.

Similarly, the "Sunday is the most dangerous day" comment appears to be based on noise rather than a real difference. We would need to perform an ANOVA using days of the week as the factor and homicides as the response to determine if there is a real, measurable difference between Sunday and the other days.

I'm glad you found my work useful.

2011-01-29T10:14:25.427-06:00

I'm glad you found my work useful.

This is impressive work, thank you for making the ...

2011-01-29T05:24:47.316-06:00

This is impressive work, thank you for making the code available! I started using ggplot2 a month ago, so it's very helpful looking at others peoples code, seeing how they do things. Thanks!