Another statistician has joined the blogosphere, and I highly recommend his blog. The name is Good Stats Bad Stats, and the author is Larry Cahoon who is retired after a career in government (official) statistics. Larry is a long-time reader of Junk Charts as well, and I'm late to the party, as he has been blogging for almost a year.
Here are some pieces that caught my attention:
In this post (link), he takes down someone who's trying to use "simple models" (a random walk !?) to overturn the global warming consensus. Watch how he shows the model exhibits non-sensical results. It's easy to fit a given set of data but models often fail when they are tested with new data. It's a failure of generalization.
In this post (link), he takes Romney and his supporters to task for his record in fighting unemployment in Massachusetts. This is an example of finding the context to interpret the data. A favorite trick of junk data hucksters is to divorce the data from its context.
In this post (link), he highlights a cool chart about the trend of divorces in Mexico. I wish the designer has taken the bold step of smoothing out the curves. All of the information is essentially contained in an inverted triangular curve. See my post over at Junk Charts (here).
There are other gems and thank me for ruining your day's plans. Happy reading.