*Running Randomized Evaluations: A Practical Guide*

That is the new Princeton University Press book by Rachel Gelnnerster and Kudzai Takavarasha.  I am not the one who should be evaluating this work, and I won’t have the chance to look at it before final exams and papers descend upon me, but it appears to be a highly useful and readable guide to what the title suggests.

You can buy it here, the book’s home page is here.


Randomized evaluations are used to collect statistics in Third World countries, since they don't have the means to collect data from everybody. Actually even in First World countries it's been found that the census is inaccurate as people lie to the census takers, so indirect ways are preferred, but I digress.

Most surveys are flawed due to lying, due to non-random selections by the interviewers, and due to the lack of incentives to give data to a stranger (most traditional societies don't trust strangers with information, often on animist/magical grounds).

I have found in my own private research (unpublished) that most data--including population data--is flawed. Do China and India really have as many people as they claim? My indirect research says no. Their five year plans dictate that a large headcount be found and this "top-down" mandate drives the population stats. Indirect measures show me the data are inflated. But I don't have time to publish --even self-publish--this research. A while ago I did take a time out and self-published on Usenet data indicating China's GDP is inflated by 30% (about the same BTW as I estimate their population count is), using an energy analysis, and, sure enough a few years later, a university professor validated my findings.

Why are these findings not more common? Because there's no intellectual property save fame associated with publishing a study such as the above. And the only people given incentive by fame are either egomaniacs (not me, despite my provocative prose style) and/or university professors.

You read it here first: China, India, Brazil, Russia, all of Africa pretty much over-count the number of people by 10% to 30% (I have an Excel chart that tracks the exact approximate number). You're welcome.

Comments for this post are closed

Brazil: Yes. No one knows how many people are in the favelas. But they know it's a lot.

Comments for this post are closed

Ray, I apologize for my arrogance but what are some of the reasons for this top down mandate to inflate population? Off the top of my head...a high population gives governing bodies an argument to receive more funding, such as collecting more taxes or departments receiving more from government budgets. What could be some other drivers?

Comments for this post are closed

Comments for this post are closed