Exit polls and partial vote counts

As promised, the Tribunal Supremo Electoral broadcast some official results at 9 PM Sunday.

Their summary, with only 24% of the vote counted: 34% Juan Orlando Hernández, 28% Xiomara Castro, 20% Mauricio Villeda, 15% Salvador Nasrallah.

The margin dividing the two top candidates is quite small: 249,660 to 202,501-- so less than 50,000 votes separate LIBRE and the Partido Nacional. The eligible electorate is 5.3 million.

Does this tell us who will win? no, it does not. We do not know which results are included; there is no way to project from likely voting patterns in areas already counted to other similar areas.

Long before the TSE broadcast these partial counts, the Honduran press owned by supporters of Hernández was calling the election for him, based on exit polling by Ingenieria Gerencial. This is the same firm that did polling for the Partido Nacional, those private polls that were alluded to during the campaign but never published.

Meanwhile, LIBRE, relying on other exit polls, saw its candidate emerging as the winner. Without a newspaper ready to declare Xiomara Castro the winner, this would only matter if you were someone (like us) who expects exit polls in Honduras to be inherently unreliable-- and thus expect contradictory results.

Before the TSE circulated their preliminary counts, Xiomara Castro announced that she has been elected; on twitter, the statement read

Con los resultados que he recibido de boca de urna de todo el país, puedo decirles: Soy la Presidenta de Honduras. [With the results that I have received from the edge of the ballotbox from throughout the country, I can say to you: I am the President of Honduras.]

This at least should serve to prevent all the Honduran press from prematurely calling the election for Hernández. Of course, it also has opened Castro up to critique from pundits nationally and internationally.

Meanwhile, Bloggings by Boz tweeted

I analyzed the exit poll data with an adjusted turnout model and got 31.5% to 31% in favor of Hernandez, well within any margin of error.

Except for the absolute number (we were kicking around 34-35% in discussions internally) that sounds about right to us: two diametrically opposed candidates separated by a threadbare margin. Not 6%-- this election should turn on 1-2% of the final vote count.