DataTaunew | comments | leaders | submitlogin
How Americans Die (bloomberg.com)
3 points by auton1 3632 days ago | 2 comments


1 point by laurenstill 3628 days ago | link

Oh boy.... the analysis here is horrible. They don't define mortality or rather, their definition doesn't match the actual definition. The question is, how many people die per year in a given age group so the first couple of slides are flatout bullshit. Axis are all wrong, slide 5 the y scale is in whole numbers, so this isn't a mortality rate. It simply can't be. Next slide, how do 25yos die at the same rate as 80? Is this percent? Total deaths? Is this 60 / 100k? We're taking a large number and making it look like a small number. Misleading. So we're not representing the population size. If I have 60 people over 85 who die in a population of 100k and 120 people under 25 in a population of 200k voila, both are 60. Slide 5 they point out the blip where we see a rise from 1984 - 1996. I don't know the significance of this rise , ie what it is in the raw data, so we don't even know the fluctuation here, or the area under the rise. Then we switch from "rate" to raw numbers and conjecture put on a plot. Better, but not of any real value. Slide 10 isn't normalized, and therefore incredibly misleading. Later "cancer and heart disease have become less deadly" is false, it's a smaller fraction of deaths. Looking at raw numbers, drug-induced went from 1.1k to 11k, suicide doubled and cancer and heart disease both increased by 25%, but none of that is normalized in this deck. /rant

-----

1 point by satoshi 3630 days ago | link

How is this done? Is D3.js used, for example?

-----




RSS | Announcements