January 24, 2023
This is the fourteenth in a series of blog posts addressing a report by Diego Escobari and Gary Hoover covering the 2019 presidential election in Bolivia. Their conclusions do not hold up to scrutiny, as we observe in our report Nickels Before Dimes. Here, we expand upon various claims and conclusions that Escobari and Hoover make in their paper. Links to posts: part one, part two, part three, part four, part five, part six, part seven, part eight, part nine, part ten, part eleven, part twelve and part thirteen.
To sum up:
This blog post series examined assumptions and biases in Escobari and Gary A. Hoover’s 2020 report, “Evo Morales and Electoral Fraud in Bolivia: A Natural Experiment and Discontinuity Evidence.” We looked at counting biases and observed that support for a given candidate need not be uniform over the count. A candidate’s vote share may trend upward and even swing quickly upward near the end of the count.
We saw that the Bolivia elections in 2019 as well as the referendum vote in 2016 showed similar tendencies, as arguably “left” voters tended to have their votes reported later than others. To be certain, rural precincts — where the voters more heavily favored the referendum — came in later. Precincts in capital cities — where Mesa was especially favored, even among opposition candidates — tended to more fully report early in the count. The vote patterns at the time of the TSE announcement, combined with knowledge about which polling stations remained unannounced, strongly suggested that Morales would win in the first round of the 2019 presidential election.
This clearly disproves the OAS claim that the late results and a first-round victory were “inexplicable.” It is not possible by any means to explain an election of any size down to the last vote. However, it appears, based on the data made available by the TSE at the time of the announcement, that a Morales victory in the first round was not only possible, but very likely. As of this writing, the OAS has not been held to sufficient account for its part in tearing down the credibility of Bolivia’s electoral system. Consequently, we reiterate our call for the OAS to engage openly and transparently concerning their activities surrounding the 2019 elections.
We saw that naive “difference model” estimates mislead. Justified on the suspicion that the late polling stations were contaminated by fraud, these models measure the effects of confounding factors such as rurality.
We saw that the addition of even small amounts of information about the election, precinct-level results from 2016, for example, go a long way in explaining election results in 2019 but that care must be taken to not overstructure our model by making assumptions that are too strict. Benign differences between 2016 and 2019 can lead to a violation of the difference-in-difference model’s assumption of parallel trends and result in biased estimation.
We saw that once Escobari and Hoover begin to allow for these benign differences, very little of the election is left unexplained. However, Escobari and Hoover then reinterpret their results by marking the nonparallel trends themselves as a sign of fraud. We emphasize that this reinterpretation is entirely inconsistent with their previous modeling assumptions. In any case, the reinterpretation reimposes sweeping assumptions dismissing very important benign causes for those differences in trends.
We strongly question Escobari and Hoover’s “triple”-difference results. The common baseline of 2016 results reduces the model to a difference-in-difference contrasting the major party (MAS-CC) margins to the minor party (MTS-21F) margins even though these different margins can in no way be expected to trend in parallel. We question how Escobari and Hoover could find such small “triple”-differences regardless of interpretation.
Even accepting the results Escobari and Hoover present, their reinterpretation that the measured difference in trends indicates 2.5 percentage points of fraud again requires unjustifiably strong modeling assumptions.
We reiterate our call for Escobari and Hoover to engage openly and transparently on a sensitive political issue into which they chose to insert themselves.
