Friday, March 20, 2009

Consent and Dissent

UPDATE (3/26/09): While double-checking my referendum data, I discovered that the original data I downloaded was not correctly labeled. Most of the precinct results were shifted over one column, except for the results of Q29 which repeated the results of an earlier question. I am not sure if the error was mine or the CEC's, but I have updated the information I reported earlier.

Azerbaijan's referendum has received praise and criticism, and you may review some of these voices here, here, here, and here. The final turnout was 71.08%, and all 29 questions received at least 87% support nationally according to the CEC.

This post does not review general conclusions about the quality of the voting process, but rather focuses on some of the data coming from Azerbaijan's CEC. The CEC deserves credit for transparency in publishing precinct-level turnout data and results quickly and efficiently. Moreover, just as in the presidential election in October 2008, the CEC installed web cameras in many precincts which allowed anyone to watch the proceedings.[1]

I combined the turnout data with the electronically recorded results protocols for all precincts (as I noted earlier, some precincts report incomplete data). Let's take a deeper look at results for question 21, the question receiving the strongest support according to official results:

101-ci maddənin V hissəsi aşağıdakı redaksiyada verilsin: «V. Müharibə şəraitində hərbi əməliyyatların aparılması Azərbaycan Respublikası Prezidenti seçkilərinin keçirilməsini mümkün etmədikdə Azərbaycan Respublikası Prezidentinin səlahiyyət müddəti hərbi əməliyyatların sonunadək uzadılır. Bu barədə qərar seçkilərin (referendumun) keçirilməsini təmin edən dövlət orqanının müraciətinə əsasən Azərbaycan Respublikasının Konstitusiya Məhkəməsi tərəfindən qəbul edilir.»

In 153 247 precincts, the question received unanimous support from voters (accounting for 66,039 122,037 votes, or around 2% 4% of all votes cast on this question). Some of these precincts were small, with 27 recording under 100 votes from voters. But several were large; 11 29 included over 1,000 voters. As I noted in a previous post, unanimous results - especially in large precincts - are questionable. Even if there is little dissent on a ballot issue, voters and election officials are not perfect. Voters make mistakes and sometimes render their responses (or entire ballots) invalid. The CEC reports that 3.64% of votes on question 21 were invalid, lower than any other question (question 16 had 4.64% invalid; all other questions were over 5%).

Indeed, many voters cast invalid votes on the question. Invalid votes could reflect mistakes, or intentional efforts to show dissent without recording a no vote. In 4,521 4,297 precincts (out of 5,367 precincts), at least one ballot was invalid ("etibarsız") on question 21. In 1,052 418 precincts, 10% or more were recorded as invalid, accounting for 124,796 50,176 votes (or a bit more than 3.5% 1% of the vote based on precinct returns). No precincts recorded a majority of "no" votes on question 21. In only one precinct did the sum of "no" and invalid votes exceed 50%.

[1] While Internet video monitoring increases openness, it also could influence less sophisticated or older voters who may believe that their presence at the polls will be recorded (inducing them to turn out and vote). But, this is a subject for another post.

No comments:

Post a Comment