The Americas | Reasons to disbelieve

What the data say about the integrity of Honduras’s election

Our analysis of a disputed vote count

THE electoral commission of Honduras (TSE) will not declare a winner in the presidential election, held on November 26th, until after a recount of some kind. The first count suggests that Juan Orlando Hernández won re-election. He beat Salvador Nasralla, a sports broadcaster, by 42.98% to 41.38%.

Mr Nasralla charges that the result is fraudulent. A weird and chaotic vote-counting process has strengthened that suspicion. After releasing preliminary results from 57% of ballot boxes, which showed Mr Nasralla with a lead of five percentage points, the TSE stopped reporting on November 27th without explanation. After publication of results resumed on the afternoon of November 28th, Mr Nasralla’s lead disappeared. That looks fishy.

The Economist has analysed the results to figure out whether someone falsified the count. Our findings are not conclusive, but they suggest there are reasons to worry.

If the results released by the TSE at each stage of the count were a representative sample of the country, the odds of the shift it reported from Mr Nasralla in early results to Mr Hernández in later ones would be close to zero. Mr Hernández has explained his luck by saying that the later ballots come from rural areas, where his National Party is stronger.

To test this theory, The Economist compared results reported from municipalities by the afternoon of November 28th with final results from the same areas. Honduras is divided into 298 municipalities; 288 had published incomplete results before reporting was interrupted. We looked at municipalities because they are small, and tend to be mainly urban or mainly rural.

Even controlling for that, the vote count shifted systematically from Mr Nasralla to Mr Hernández between early and later results. In chart one, each dot represents a municipality. The chart shows for example that Mr Hernández got 36% of the votes in La Libertad, in central Honduras, before the TSE stopped publishing results. After it resumed, Mr Hernández got 49% in the same place. Mr Nasralla’s share dropped from 51% to 36%. Our analysis shows that he lost 3.5 points on average relative to Mr Hernández within each municipality.

Proving fraud through such analysis is difficult. Statistical anomalies can have reasonable explanations. One possible objection, even though municipalities are in general fairly homogeneous, could be that those in which Mr Hernández outperformed have a large number of voters living in urban areas that reported early and many living in late-reporting rural areas.

We asked Rosemary Joyce, an anthropologist at the University of California, Berkeley who specialises in Honduras, to see if that was the case. She found that explanation for the vote shift implausible: municipalities in the departments of La Paz and Lempira, where Mr Hernández improved significantly between early and late counting, do not have large towns.

Ms Joyce’s claim is supported by our analysis of census data from 2013. We looked at the split between rural and urban households in the 284 municipalities for which data are available, as well as the proportion of houses with dirt floors, which correlates closely with the share of rural households. We found no relationship between how rural a municipality was and how sharply its vote shifted towards Mr Hernández (see chart 2).

Another possible objection to our analysis is that the early reports were based on vote tallies that were sent electronically to the TSE; 29% of vote tallies were not, according to monitors from the EU. There might be a reason why electronically transmitted votes would favour Mr Nasralla. But the difference would have to be huge to explain the shift in the later count to Mr Hernández. If votes sent electronically favour Mr Nasralla by five percentage points, he would have had to lose by over 18 points among votes reported on paper to explain the late shift towards the president. The odds are that that didn’t happen.

This article appeared in the The Americas section of the print edition under the headline "Reasons to disbelieve"

The choice that could save South Africa, or wreck it

From the December 7th 2017 edition

Discover stories from this section and more in the list of contents

Explore the edition

Discover more

Nicolás Maduro’s sham election: the sequel

Venezuela’s unpopular president is reusing a familiar script

Can Haiti’s police hold on?

Help is taking time to arrive, as politics delays an international security mission


The cocaine trade is booming in Europe’s Caribbean territories

The overseas possessions provide an easy route into the EU