Dr. Mark Kosmo on the Albanian Elections!

I read Dr. Mark Kosmo's brilliant analysis (below) of the recent Albanian elections with avid interest and greatly appreciated his astute and insightful review of the election tallies, and what they might portend in the near political future in Albania.

Hats off to Mark for helping to set the voting record straight. BRAVO!

***

OPEN LETTER TO ALBANIAN MEDIA, GOVERNMENTAL INSTITUTIONS, FOREIGN ORGANIZATIONS

The Tirana Mayoral Election: Were the Contested Votes Fair or Fraudulent?

May 23, 2011

For anyone wondering about the Tirana Mayoral election and the continuing intrigue, accusations, legal disputes, and character assassination, it is helpful to sometimes take a more scientific and logical approach and ask one simple question regarding the counting of the contested votes this past week.

Can one explain how one candidate received 67% of the contested votes after receiving 50% of the general vote?

We know that: (i) out of 250,000 votes cast, each candidate received 50% of the vote; and (ii) for the 256 contested ballots counted from 125 boxes this past week, one candidate won by a margin of 173-83 or 67.6% to 32.4% -- i.e., about twice as many of the contested votes went to one candidate over the other candidate.

What are the possible explanations? There are two: (i) the voters supporting one candidate were twice as likely to insert their ballots into the wrong box; or (ii) the contested votes were fraudulently manipulated and/or judged in a biased manner. Neither explanation can be proven, but each can be assessed.

For the voters supporting one candidate to be twice as likely to put their ballots into the wrong box at least one of the following must be true:

(i) due to inexperience, first-time voters are more likely to cast their ballots erroneously AND first-time voters are twice as likely to support the same candidate; or

(ii) due to inexperience, young voters are more likely to cast their ballots erroneously AND young voters are twice as likely to support the same candidate; or

(iii) due to color-blindness, color blind voters are more likely to cast their ballots erroneously AND color blind voters are twice as likely to support the same candidate; or

(iv) due to lower levels of intelligence or awareness, less intelligent or aware voters are more likely to cast their ballots erroneously AND less intelligent or aware voters are twice as likely to support the same candidate; or

(v) one set of officials in the polling stations were twice as likely to deliberately mislead voters whom they suspected of voting for the other party to insert their ballots into the wrong boxes with the hope that these votes would be declared invalid; or

(vi) one set of officials in the polling stations was twice as likely to deliberately mislead voters whom they suspected of voting for their party to insert their ballots into the wrong boxes with the hope that these votes would be declared valid; or

(viii) supporters of one candidate were twice as likely to receive inadequate assistance from polling officials to ensure that their ballots were inserted into the correct box; or

(viii) the ballot boxes that were counted were from voting stations where one candidate, on average, got 2/3 of the votes.

Is there any reason to believe the first seven explanations? If yes, then the contested votes could split 67-33%. If no, then the chances of a contested vote going to either candidate are 50% each. Of course, there is no way to answer the the first seven points definitively -- each person has to make his or her own judgment based on common sense regarding what he or she thinks are reasonable possibilities.

Could the ballot boxes with contested votes all have been from polling stations where one candidate received 2/3 of the votes?? This is highly doubtful given that only 2 of 11 districts had any candidate get over 55% of the vote -- http://www.shekulli.com.al/2011/05/14/rezultatet-e-numerimit-te-10-njesive-vendore-ne-tirane.html.

If none of the above is true then the chances that the recount was a fraud are about equal to the chances that a DNA match on a paternity test correctly identifies the father – i.e., well over 99%. This can easily be verified in 15 seconds at http://stattrek.com/Tables/Binomial.aspx. From a total of 250,000 votes where the general population voted 50% for each candidate, the chances of 67% supporting any one candidate in a pool of 256 votes is approximately 1 in 200 million -- see table below. The result is the same as flipping a coin 256 times and getting 173 heads -- there is a 1 in 200 million chance that 2/3 of the coin tosses coming up heads on a fair coin.

A skeptical person might argue that the chances of any one outcome is always low and therefore that the above table and analysis is not very illuminating. To further clarify realistic possibilities one can look at a range of outcomes in the table below for vote differences between the two candidates of 0, 10, 20,...90 votes out of 256 votes. The chances of a difference of 90 votes as realized under the recount this past week falls in the range of extremely low possibilities from 1 in 45,000 to 1 in 200 million. In other words, if the vote was manipulated, it was manipulated too much to be credible and enough to be statistically detected.

Similarly, an argument such as Candidate A could win 125 boxes by one vote each and each of those votes is likely to have been fair has an astronomically low chance of being true. To win 256 votes from 125 boxes as the count of the contested votes shows means that for approximately two votes per box one candidate averaged 1.33 votes and the other 0.67 votes. To use a clearer example, imagine if a box has three votes. To win one vote 90 times in a row, on average, is the same as winning a bet 90 times in a row with less than 40% chance of winning each time since the chances of getting 2 votes out of 3 is 37.5% in any one box. To win such a bet 90 times in a row would suggest it is time to move from Tirana to Monaco or Las Vegas since the chances are less than winning 90 coin tosses in a row.

There has been much debate and argument regarding the role of the Albanian Election Commission. It is for others to debate and question its integrity or professionalism and whether or not the contested votes were transparently and diligently counted according to its mandate and the law. However, two questions remain -- were the voters that supported one candidate twice as likely to insert their ballots into the wrong box or were the contested votes fraudulently manipulated and/or judged in a biased manner? Which of these possibilities is more likely? Each of us has to answer that question ourselves -- in Tirana, in Albania, in Kosova, in the United States, and everywhere where Albanians live -- and vote!!!

In closing, I would like to invite anyone who reads this document to send me comments -- English preferred, but Albanian is OK. I also invite you to forward this to others as you please -- especially to international statistics and election evaluation specialists who could give an unbiased and professional review to better assess the validity of the contested votes.

Mark Kosmo, PhD Economics
Former Economist for the World Bank
Boston, MA

May 23, 2011