Errors in Election Pools

Errors in Election Pools

The Professor Alexandre Patriota recently hosted a great presentation from Raphael Nishimura called “Errors in Election Pools” hosted in the IME (Institute of Mathematics and Statistics).

During my period as a student, production engineering, I had contact with some of these issues, with regard to the design of experiments and the understanding of sources of errors that could add bias, whether measurement or sampling errors.

As a Brazilian and former resident of the city of São Paulo, this is a topic that interests me a lot.

Not only from a civic perspective, but also from a mathematical perspective and design of experiments and operational research, given that throughout my data career as I became more interested in the subject, I came to understand how bad the research was and how much a poll carried out with a bad methodology can directly affect the outcome of an election, as in these two symbolic cases that raised doubts about the usefulness and credibility of research institutes.

• In the 2010 election, in which absolutely all the research institutes gave two candidates (Marta and Netinho) with a very wide margin and in the end Senator Aluysio Nunes was elected.

• In 2012 in São Paulo, the current mayor Fernando Haddad who was practically removed from the second round by the research institutes and in the end he was elected.

The lecture is in Portuguese, but there is the option of subtitles in English and for those who just want to consult the slide deck, the link is in the references below.

Measurement errors and Sampling errors

Undercoverage

Overcoverage

References

• Slide deck
• YouTube Channel A Ciência da Estatística