Statement About Vote Count Is Wrong, and Destabilizing
For Immediate Release: October 22, 2019
Contact: Dan Beeton, 202-239-1460
Washington, DC ― The OAS statement yesterday on Bolivia’s election should be retracted, said Mark Weisbrot, Co-Director of the Center for Economic and Policy Research.
On Monday, October 21, the OAS issued a statement expressing “its deep concern and surprise at the drastic and hard-to-explain change in the trend of the preliminary results after the closing of the polls.”
“The OAS statement implies that there is something wrong with the vote count in Bolivia because later-reporting voting centers showed a different margin than earlier ones,” Weisbrot said. “But it provides absolutely no evidence — no statistics, numbers, or facts of any kind — to support this idea.
“And in fact, a preliminary analysis of the voting data at all of the more than 34,000 voting tables — which is all publicly available and can be downloaded by anyone — shows no evidence of irregularity.”
See below for an example of this analysis at the city and precinct level. As can be seen from the data, the change in the vote margin in the later-reporting voting centers is a result of geography — i.e., pro-government areas, on average, reported later than those that have a higher proportion of voters who are against the government.
“This kind of change in voting results, due to later-reporting areas being politically or demographically different than earlier ones, is quite common in election returns — as anyone who has watched election returns come in on CNN in the United States knows,” said Weisbrot. “That is why it is wrong to draw conclusions from a change in the voting pattern without any statistical analysis or even looking closely at the data.”
Weisbrot said that it was “irresponsible” to make such accusations when there was already postelection violence in Bolivia. “As this narrative gets repeated in the media, it will take on a life of its own, and will be difficult to correct, even as more people look at the data, or produce statistical analysis.”
Weisbrot also said it was “disturbing” that the OAS release followed, and was followed by, statements by powerful, partisan US political actors who have long been outspoken against the Bolivian government.
Senator Marco Rubio, hours before the OAS release, declared: “Evo Morales failed to secure necessary margin to avoid second round,” and expressed concern that “he will tamper with results.” Trump administration officials made similar statements.
Appendix 1: Preliminary Analysis of Electoral Results: The Case of Cochabamba
At 7:30 p.m. on Sunday, October 20, 2019, the Supreme Electoral Tribunal (TSE) of Bolivia released a quick count of electoral results covering 83 percent of votes. At that point, Morales’s lead over his closest rival, Carlos Mesa, was less than the 10 percentage points needed to avoid a runoff election. After a nearly 24-hour delay, the TSE resumed its quick count. By 10:30 a.m. on Tuesday, October 22, 2019, the quick count covered about 95 percent of votes. Morales’s lead over Mesa had been extended and had surpassed the 10 percentage point threshold.
In a press release, the OAS noted an “inexplicable change in trend that drastically modified the fate of the election and generates a loss of confidence in the electoral process.” However, a preliminary analysis of the quick count in the city of Cochabamba indicates that the overall change in trend observed between the results posted the evening of October 20 and then in the days after is, in fact, easily explainable.
When the TSE released results of the initial quick count on October 20, there had been 336,563 votes counted in the city of Cochabamba. Of those votes, Evo Morales (MAS Party) had secured 38.9 percent while Carlos Mesa (CC Party) had secured 51.4 percent. By the morning of October 22, 106,925 new votes had been added to the tally in Cochabamba. In those 106,925 votes, the trend shifted dramatically — Morales received 52.2 percent and Mesa 35.4 percent.
The Table below shows the shifts in trend from the seven cities with the largest number of new votes counted. As can be seen, the most significant trend shift occurred in the city of Cochabamba.
In order to analyze that shift in trend, we looked at the data at a more refined geographical level — the precinct level. The results of that analysis can be seen in the Table below, which shows changes in the ten precincts with the largest number of new votes counted within the city of Cochabamba.
What the table above shows is that there was not actually a trend shift within precincts. As can be seen, Morales’s vote share among the 100,000 new votes counted in Cochabamba actually decreased in a number of precincts, and increased in others. This indicates that the overall trend shift observed in Cochabamba was due to later reporting, on average, of precincts that favored Morales; and not to some unexplained trend shift that could cast doubt on the results of the election.
Appendix 2: Projections of Electoral Results in Bolivia*
As of 3:00 p.m. local time in Bolivia, on October 22, there were 4,300 voting tally sheets remaining to be counted. Zero were in the Department of Santa Cruz; 2,730 were in the Department of La Paz. Therefore, the results of this election will depend heavily on the Department of La Paz.
If we include the votes of Bolivians living outside the country, with 88 percent of all votes counted, President Evo Morales leads Carlos Mesa by 45 to 39 percent. In order to win in the first round, Morales needs to lead Mesa by a margin of at least 28 percentage points, among the remaining votes.
In the municipality of Nuestra Señora de La Paz, Mesa leads by a margin of 49 to 40. But there are only 90 tally sheets remaining to be counted in this municipality.
In the rest of the department (El Alto and the smaller municipalities), Morales leads by a margin of 58 to 21. In these areas, there are 2,640 tally sheets remaining to be counted.
If this margin is replicated in the pending tally sheets (which is likely), Morales would win in the first round, surpassing 40 percent and with a difference of more than 10 points above the second-place finisher Carlos Mesa.
These results can be verified by downloading the Excel sheet with the results of each tally sheet. The scan of each tally sheet can be verified here.
*Note: Appendix 2 uses data from the official vote count; Appendix 1 is using data from the quick count. The results of both have been widely, and legitimately, cited and reported in the media.