Comments on: Weekend Minis for Your Lazy Weekend – 7/19/08

By: Jon Peltier

Jon Peltier — Sun, 20 Jul 2008 00:24:00 +0000

Aha. I had looked at the numbers, and hadn’t noticed that the cutoffs fit into 2-percentage-point bins. I understand why the rnages have to adjust to a moving distribution, I just didn’t know what guidelines had been used.

By: CalorieLab

CalorieLab — Sat, 19 Jul 2008 14:21:00 +0000

Jon, each state has a number representing the percentage of its adult population that is obese (i.e., BMI 30 or over). The colors in the map represent 2-percent ranges (except for the two extremes). The divisions were empirically chosen to maintain a more-or-less equal numbers of states per color, so that the general distribution can be grasped without having to resort to reading the legend.

Think of it as similar to adjusting the levels in Photoshop so that the histogram for a photograph extends from left to right, giving the best tonal range.

If we simply used the same scale every year, all states would soon be dark orange. The average increase in obesity nationwide is 1 percent per year, so we’ve been shifting the color legend 1 percent right every year.

By: Jon Peltier

Jon Peltier — Sat, 19 Jul 2008 13:47:00 +0000

Fattest States 2008 – A lot more red-tinged states than green-tinged. They don’t state what criteria they use to divide into different color groups, just that they had to change it this year.

The informal correlation between obesity and evangelicals is interesting.

By: Jon Peltier

Jon Peltier — Sat, 19 Jul 2008 13:39:00 +0000

Pew Study Stacked Bars

1. If it’s only Yes/No/Undecided, the stacking is okay. By question 3, where there are numerous bars in the stack, it becomes difficult to interpret.

2. Sorting by ‘Yes’ is fine, except Total Population should definitely stay at the top, and other/unaffiiated should probably stay at the bottom.