Saturday, January 17, 2009

Statistics that Aren’t Even Lies? It Could Be Even Worse

Nobody knows for sure who coined the phrase popularized by Mark Twain “lies, damn lies, and statistics”, but some figures are so bad they aren’t even statistics. In the comments on this post, I looked behind a comparative UNDP table of the ratio of urban and rural populations by nation at the underlying definitions used by the Japanese and U.S. governments, and I think that I’ve shown that the table is utterly useless. It is not to be hoped for that this table is not being referenced as fact by myriad of derivative sources and used as the basis for UNDP and other UN policy recommendations, nor that such things do not happen with regularity. I stand ready to be corrected. (And sometimes am. One reader politely pointed out a fundamental misstep in my original post.)

I spent too much time on making my point not to draw attention to it and be finished with posting for the day. Besides, I think that it serves as an independent reminder of the dangers of relying on poorly sourced facts and figures.


Janne Morén said...

That source is indeed rather problematic for comparisons as you point out. Should have checked that better. If it matters, my original posts about this used a different source for a subset of countries, presumably collected using common criteria. I'm not really engaged enough (or with enough free time) to unearth that data again at this point, and will just let it rest.

Jun Okumura said...

Janne: Everything goes back to the census for the kind of detailed data sets you need for analysis on a national basis, let alone international comparisons.