A quantitative analysis of the 2017 Honduran election and the argument used to defend its outcome

TL;DR

This paper quantitatively examines the 2017 Honduran election, analyzing the validity of the incumbent's explanation for the election outcome and finding strong evidence against it.

Contribution

It formalizes the 'Hernandez conjecture' mathematically and uses geodemographic data to assess its validity with a rigorous quantitative approach.

Findings

Overall probability of the conjecture being true is less than 0.0001.

Analysis shows negligible chance of a fair win by the incumbent.

Geodemographic data does not support the incumbent's explanation.

Abstract

The Honduran incumbent president and his administration recently declared victory in an election riddled with irregularities and indicators of fraud. Perhaps most curious, however, was a numerical anomaly: the primary challenger carried a very significant lead of five percentage points more than half way through the election but was ultimately defeated by the incumbent. The incumbent (Hernandez) offered a plausible explanation for the surprising turnaround in the ballots: his popularity is greater in remote areas of the country but votes from remote areas were not counted until later in the election. Here, we mathematically formalize this argument, which we will call the Hernandez conjecture, and employ the resulting formulae together with geodemographic data from Honduras to quantitatively assess the conjectures veracity. When the departamentos were analyzed individually, three sparsely-populated departamentos (of 18 total) showed small but non-negligible probability of the conjectures veracity; however, when the country was analyzed as a whole, the overall probability of the conjectures veracity was calculated to be less than 0.0001 under a wide range of different assumptions. Results of our three-pronged analysis, taken together, indicate a negligible probability of a fair win by the incumbent.