On October 20, 2019, Bolivia held a presidential election. The official count would show incumbent Evo Morales with a first-round victory with 47 percent of the valid vote — more than 10 percentage points ahead of the runner-up, former president Carlos Mesa. However, Morales was not able to finish his term in office — let alone begin his next term. On November 10, under pressure from the military, Morales was forced to resign and flee the country.
Even before the October election, many in the opposition had pledged not to accept the results of the vote if Morales won. The Organization of American States (OAS), in country to observe the vote, poured gasoline on the fire when — the day after the election and with the official count not yet complete — it issued a press release expressing concern about “an inexplicable change in trend [in the preliminary count] that drastically modifies the fate of the election and generates a loss of confidence in the electoral process.”1 The statement delegitimized the election in the eyes of many, both in Bolivia and internationally. It also provided opposition elements with the justification they needed to reject the election results and force Morales from office three weeks later.2
As CEPR and a number of other independent researchers have shown, the OAS has produced no credible evidence to back up its statement of October 21, 2019. While a long interruption of the preliminary vote count ― from around 7:45 p.m. on October 20, 2019 to 6:30 p.m. the following day ― raised understandable concerns, there was no “inexplicable” or “drastic” change in trend following the interruption, as the OAS alleged.
Irregularities were going to be inevitable in this election — as with any other — even absent any fraud. However, the OAS’s initial unsubstantiated allegation strongly colored how the OAS interpreted the existence of irregularities, confusion, and destruction of voting materials in the election as evidence of systematic and intentional fraud. The OAS sought out irregularities specifically on tally sheets that heavily favored Morales on the basis of its unjustified claim — and then used the fact of the irregularities to justify its prior actions. Likewise, the OAS held up any sign of confusion or chaos (or even destruction) by any party in the election as further evidence of opportunity for manipulation by the government, seen through the lens of its unjustified claim. The OAS went so far as to bemoan the difficulty in auditing tally sheets that were destroyed when opposition protesters burned several departmental election offices. The audit also included a deeply flawed statistical analysis purporting to substantiate the OAS’s initial claim of inexplicable change in the trend of the preliminary vote count. We have reviewed and responded to the audit in detail elsewhere.3
The importance of the OAS’s insistence that the late results were inexplicable cannot be overstated. The initial allegation and subsequent audit were cited by the de facto government of Bolivia in their persecution of members of the Tribunal Supremo Electoral (TSE) and prominent members of the political party that Morales led — the Movimiento al Socialismo–Instrumento Político por la Soberanía de los Pueblos (MAS-IPSP, or MAS). Human Rights Watch reported that prosecutors charged at least four members of the Santa Cruz TSE
solely on the report by the Organization of American States … according to the charging document and the official transcripts of three hearings.4
As The New York Times reported, the OAS allegations of fraud “fueled a chain of events that changed the South American nation’s history.” 5
Since Bolivia’s elections last year, several researchers have performed statistical analyses attempting to show fraud based on a failure to completely explain the “inexplicable change in trend” that so concerned the OAS the day after the vote. Once subject to minimal scrutiny, none of these studies credibly put the official election result in doubt.
In previous papers, we have critiqued the research of Diego Escobari and Gary Hoover, John Newman, and Irfan Nooruddin, the statistician hired by the OAS to perform its own statistical analysis of the vote.6 In a document named “Newman Response to Idrobo et al and Rosnick,” Newman responded to our critique of his work.7 The OAS has since cited his work as further justification of its actions in Bolivia.8
In this paper, we provide a detailed rebuttal to Newman’s response. Primarily, we showcase the general inadequacy of his approach and extend his framework for further analysis. With that framework, we revisit and build upon our previous critiques and expand these to include the work of additional researchers, including that of Rómulo Chumacero, and of Edgar Villegas.9 In Section I, we highlight a common fault in the primary statistical evidence against Morales. To varying extents, these researchers failed — partially or completely — to account for the tendency of tally sheets representing opposition-heavy precincts to be counted before those in areas supporting Morales. In Section II, we address additional concerns raised by Newman in his response.
In summary, we reaffirm our previous conclusions that Morales’s first-round victory was predictable based on the data available at the time of the October 21, 2019 OAS statement that, in the words of The New York Times, “fueled a chain of events that changed the South American nation’s history.”10
Read the entire report here.
The author wishes to thank Alex Main, Dan Beeton, Karen Conner, Kevin Cashman, Brett Heinz, Jake Johnston, and Guillaume Long.
1. OAS (2019a).
2. This is particularly distressing as the preliminary count has, in any case, no official standing. The purpose of this count is to provide the public a day-of sense of how the populace voted. By contrast, the official count is a lengthy process of scrutinizing tally sheets. As a fraud-prevention measure, no corrections are allowed on the tally sheets. Instead, when errors are made, these are presented as “observations” signed and fingerprinted in an allotted section of the sheet. Most frequently, observations report that votes for one party were incorrectly recorded on the tally sheet for another party. These observations are not taken into account in the preliminary count, resulting in inaccurate reporting of results.
3. Johnston and Rosnick (2020).
4. HRW (2020).
5. Kurmanaev and Trigo (2020).
6. Rosnick (2019) and Rosnick (2020).
7. Newman (2020a). His work in question is Newman (2020b).
8. OAS (2020a).
9. Chumacero (2019) and Alvarado, et al. (2019).
10. Kurmanaev and Trigo (2020).
This introduction originally appeared on CEPR.