One reason you should always look at your data rather than mindlessly running regressions is because some point may so far be out of place that it causes your results to be corrupted. A case in point is from the recently released Human Development Report. For people who follow these types of reports know that it is basically skewed towards welfare state governments, in other words, Scandinavian countries do quite well because of their large government spending on a variety of public welfare measures. It's still enjoyable reading through it and looking for those odd pieces of data that jump out at you. On page 258, I found a number which is so obviously out of whack that it should have raised red flags for anybody working with the data.

Did you know that only 92% of children in the U.S. go to primary school? Even worse, this number is down from 97% a decade earlier. Even accounting for 1 or 2 percent that home school, that still leaves over five percent not accounted for. If this was true, it would be a national scnadal. As a footnote says, there are discrepencies between the number of students and the number of school aged children which could account for this difference, but this strikes me as quite large and probably outside of any reasonable error measurement.


Powered by Movable Type 5.02

About this Entry

This page contains a single entry by Bob published on September 11, 2005 2:49 PM.

Working Poor was the previous entry in this blog.

Qual & Other Stuff is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.