So who makes the top ten of your friendlenta? Let’s have it.

As another remark, if I may, I find the ЖЖ part of your blog reading list a bit peculiar. All your picks are on my френдлента too, but if there was, say, a ten or twenty friends limit, none of them would’ve made the cut. Well, except Kashin of course. By the way, while we’re at it, the article you attributed to Kashin is actually by Andrei “Я Путина видел” Kolesnikov.

I’m curious to know what you think could be an improvement, as I’ve been thinking about this myself.

“Not all sources rank all countries of the index.”

“A country must be covered by a minimum of 3
different sources to be ranked in the CPI.”

The first two organizations listed are the Africa Development Bank and the Asian Development Bank. Obviously, the first must have concentrated on Africa and the second on Asia. A question immediately arises: when the respondents to the Africa Development Bank’s survey rated African countries, were they comparing them to the rest of the world or only to other African countries? I’m sure they were told to rate on an absolute, universal scale, but people often think in relative scales instead. “Well, by African standards, this is a 6″ and so on. If you look at the table of correlation coefficients on page 5, you’ll see that the Africa Development Bank’s data correlated poorly with data from other sources.

Of course the above is just a guess of mine and the weirdness of the results could be due to other, unknown-to-me causes. And the results ARE weird. I’m looking at their 2008 list right now and Botswana is there at 5.8, while the Czech Republic is at 5.2. Russia is at 2.1, tied with Bangladesh, Kenya and Syria. Liberia is at 2.4. AGT, I’m curious, do you think that Russia is more corrupt or less corrupt than Liberia?

I see that one of TI’s sources is the Economist Intelligence Unit, which “uses its panel of experts’ assessment on the incidence of corruption.” So the EIU doesn’t interview businessmen, it uses its own experts instead. OK. The Economist magazine’s political stances are well-known. I would describe them as libertarian and neoconservative. Fans of Putin they’re not.

Again, this doesn’t prove anything and I am just trying to guess why the results look they way they do. Generally speaking, the most typical cause of survey results failing the face validity test is bad sampling. Are the samples representative of the population being studied? Here the relevant populations would be businessmen and journalists. I can’t answer that last question just by looking at a 10-page methodology booklet. All I can provide are guesses.

I’m not asking you guys to go master quantum physics here. That PDF laying out their methodology (see link above) is literally just a collection of questions with quantifiable answers, put to businesspeople and professional experts/scholars.

I think this discussion would benefit immensely from one of you discontents actually reading the thing.