Kevin Delaney Managing Editor, WSJ.com
Neil McIntosh Editor, WSJ.com Europe
Dear Mr. Murray,
Dear Mr. Delaney,
On December 28, 2011, The Wall Street Journal published an article titled “Russia's Dubious Vote” (http://online.wsj.com/article/SB10001424052970203391104577124540544822220.html).
As I was directly involved in the preparation of the article, consulting WSJ Moscow office reporters on the statistical analysis approach and calculation methods, I consider it necessary to comment on the statistical results contained in the article and the way they are presented.
It is very important that WSJ attracted public attention to the issue of irregularities in recent Russian parliamentary elections. The problem is of great importance both for Russia and for the state of democracy worldwide. Still, the statistical findings in the article are presented as an original research of WSJ’s reporters, while they are a reproduction of prior results and approaches by many Russian researchers and bloggers, including me (Sergey Shpilkin, blogging as podmoskovnik.livejournal.com).
The approach to fraud detection based on analysis of precincts and votes distributions as a function of turnout have been used in statistical analysis of Russian elections since 2007 (see, e.g. a 2008 article in The Times http://www.timesonline.co.uk/tol/news/world/europe/article3768223.ece). The approach was detailed in 2009 in “Troitsky variant” bi-weekly (http://trv-science.ru/2009/10/27/statisticheskoe-issledovanie-rezultatov-rossijskix-vyborov-2007-2009-gg/) and was widely discussed in Russia. A more popular account was published recently in Russian edition of Esquire: http://esquire.ru/elections.
For the December 4 parliamentary elections, first quantitative estimates of fraud appeared in blogs in just several hours (see, e.g. http://podmoskovnik.livejournal.com/129843.html), and soon after that many statistical analyses were published in blogs (http://oude-rus.livejournal.com, http://kobak.livejournal.com, http://dmitrykogan.livejournal.com, to name just a few) and media (e.g. http://lenta.ru/articles/2011/12/06/elections/, http://en.gazeta.ru/news/2011/12/13/a_3926402.shtml, http://trv-science.ru/2011/12/20/matematika-vyborov-2011/). These analyses revealed all the statistical irregularities that are mentioned and illustrated in the WSJ publication, including peaks in distributions at round percentages of votes for United Russia and differences in voting patterns for UR and other parties. On December 23, 2011, I presented a summary of the findings to the Russian Presidential Council for Civil Society and Human Rights.
After December 4 election, the head of WSJ Moscow bureau asked me to give him an explanation of my approach, stating that WSJ is going to check the calculations and prepare a publication. I considered it a good chance to bring the results to worldwide audience with support of one of the world’s leading newspapers. So I explained the statistical analysis, formulae and calculations in detail, and provided further clarifications in the following email exchange with him. As seen from the WSJ publication, the results presented are based exactly on these methods, and the graphs are virtually the same as those published before by various Russian authors. The fact that WSJ’s graphs and calculations are based on final election data while some of the prior results used preliminary data covering 95% or 98% precincts cannot make any difference, as possible discrepancies are far below the accuracy of fraud estimates and far less than could be noticed in the graphs. So the results published by WSJ, while calculated and graphed independently, are obtained using known methods developed by other researchers, which is not stated anywhere in the article.
I want to emphasize that this is not a copyright issue, but an issue of attribution. Having these results published in a prominent English-language newspaper without proper attribution of approaches used creates substantial obstacles to any further scholarly research in this area. Any of the original researchers who would attempt to publish something on the topic in a peer-reviewed journal would have to prove that their results indeed belong to them and not to WSJ. This would be an unnecessary burden for researchers and might have a negative impact on WSJ image in the academic circles.
The WSJ blog post “Russian Bloggers’ Findings Support ‘Fingerprints of Fraud’ in Election Results” at http://blogs.wsj.com/emergingeurope/2011/12/28/russian-bloggers-confirm-fingerprints-of-fraud-in-election-results/ makes no significant improvement to the situation, as 1) it does not state clearly that WSJ used approaches developed by others, and 2) it is only linked from the article page and not included in the article itself.
So I would suggest that you update the Behind the Analysis: The Methodology section in the online publication of Russia's Dubious Vote, including clear attributions of the sources and approaches used in the research. This will clarify the situation, remove obstacles to subsequent academic research and improve the image of WSJ itself.
Sergey Shpilkin aka podmoskovnik.livejournal.com
Researcher of Russian elections since 2007
P.S. As I consider this an issue of public importance, I am also publishing this letter in my blog podmoskovnik.livejournal.com