Talk:Preston curve

Latest comment: 14 years ago by Radeksz in topic Outliers

Outliers

edit

Has anyone studied the outliers in the data. The curve appears to me as an example of what not to do in regression, namely take the data as a package, throw them in a regression routine and name the resulting curve after oneself. Of course I might be mistaken, but the graph looks like a textbook example on how one does not treat data. Cheers 85.220.111.211 (talk) 20:27, 18 January 2010 (UTC)Reply

(some) of the outliers are already discussed in the text - mostly have to do with the AIDS epidemic in Sub Saharan Africa. Otherwise, as far as cross section relationships go, it's been a pretty strong one throughout the 20th century. And I think you can see the logarithmic nature of the relationship just from "eyeballing" the data, a couple outliers notwithstanding.radek (talk) 21:15, 18 January 2010 (UTC)Reply
The other big outlier - the one at the end of the income scale is the US, which has consistently lower life expectancy than rest of the developed countries. In earlier years this wasn't as pronounced of an outlier as it is here.radek (talk) 21:17, 18 January 2010 (UTC)Reply