•Press Release Latin America and the Caribbean World
October 17, 2011
Statistical Analysis Shows that OAS Action Was Inconsistent with Election Data
For Immediate Release: October 17, 2011
Contact: Dan Beeton, 202-239-1460
En Español
Washington, D.C. – An Organization of American States (OAS) Mission overturned the results of the first round of Haiti’s presidential elections last year, despite that it had no statistical evidence to do so, a new Center for Economic and Policy Research (CEPR) paper finds.
“The OAS’ actions in taking the unprecedented step of overturning an election, without a recount or evidence for its action, casts serious doubt on the institution’s credibility as an independent, neutral arbitrator or election observer,” CEPR Co-Director Mark Weisbrot said. “It’s difficult to see this as anything other than a political intervention.”
Weisbrot added: “Any government considering having the OAS involved in their election in any way should reconsider until the organization has conducted an investigation of their abuses in Haiti and taken steps to make sure that this can’t happen again.”
Weisbrot noted that the purpose of the CEPR paper was not to cast doubt on the legitimacy of Haiti’s government, nor to provide evidence as to who should have been elected president. Rather, the purpose was to investigate whether the OAS had any statistical or empirical basis for its unprecedented action in reversing the election results.
The paper, “The Organization of American States in Haiti: Election Monitoring or Political Intervention?” by David Rosnick, shows that the OAS’ prescribed methodology of discarding “suspect” vote tally sheets would not be expected to move the vote count closer to the intent of the voters. After throwing out a number of vote tally sheets, the results of the first round of the election were reversed, with Michel Martelly (now president) taking second place, and Jude Célestin pushed into third, and thereby eliminated from the second round runoff election. The United States government subsequently put enormous pressure on the government of Haiti to accept the elimination of the government’s candidate (Célestin). Recently revealed Wikileaks cables show that Washington had turned against the government of Haiti for political reasons.
“The point is not that the OAS and CEP [Haiti’s Provisional Electoral Council] ought to have incorporated imputations into the official count, but that they did in fact impute zeros for these sheets,” the paper states.
By means of a thorough statistical analysis of all of the tally sheets from the first round of elections, the author was able to model hundreds of possible scenarios based on imputations for the missing and excluded data. The results showed that Célestin, not Martelly, was by far the most likely second place finisher in the first round.
The OAS Mission, the paper notes, only examined a portion of the total tally sheets, and the tally sheets it chose to discard were from disproportionately pro-Célestin areas. The OAS did not use any statistical inference to in order to estimate what the result might have been had they examined the other 92 percent of tally sheets that they did not examine. Nor did the OAS attempt to account for the impact of more than 150,000 missing or excluded votes, nearly 12 percent of the total.
The paper concludes that the OAS Mission “considered four approaches to remedy the high rate of irregularities in the tally sheets. Of these, three— voiding the entire election, conducting a revote in selected problem areas, or conducting a nationwide recount—would have at least addressed that question. It is most unfortunate that the OAS chose to simply throw out selected ballots for technical reasons.”
“It is absurd for the OAS to then reverse the results of the first round of the election, or to support any results,” the paper states in its conclusion.
The election was also marred by the exclusion of the country’s most popular political party, and a record low turnout in both rounds (less than 24 percent participation in the first round).
###
