Notes for the Evaluation of Pre-Election Polls: The 2010 Elections for Governor in Mexico

Many of the pre-election polls published in 2010 that centered on the elections for governor in Mexico were incorrect. This has strengthened many of the misgivings that already existed about public opinion polls that bring into question the effectiveness of the methods used to measure preferences or the honesty of the pollsters. For 2010, there is no systematic evidence that suggests that the discrepancies between the polls and the election results are due to election effects. On the other hand, there is evidence of polling house effect. In this note, I explore several hypotheses based on the available data from pre-election polls published for the twelve entities with elections for governors during 2010. I also mention other determinants that, because of the absence of any public data, cannot be substantiated at this time and thus must be left for future research.