Friday, June 6, 2014

#PredictiveCOL - Final prediction

Adhesions are ready: Ramírez is now supporting Zuluaga and López is supporting Santos. Peñalosa has given freedom to his voters. However, there is something interesting that can change the future of the election: Santos voters were less than voters of Unidad National voters to the congress. That is somehow explained because the abstention in the caribbean coast was very atypical. It is assumed that parties are working hard in order to “encourage” voters to get effective votes.


According to pollsters, this is the evolution of voting intention:


NewImage


Now, if election were held today, this is the SNAPSHOT (mere opinion) of the run. The credibility interval for blank vote is (8.7 %, 15.2 %), for Santos is (39.4 %, 49.3 %) and for Zuluaga is (38.6 %, 48.5 %).


NewImage


On the other hand, having into account the auxiliary information, this is the forecast of the election for the day to be hold (June 15, 2014). The credibility interval for blank vote is (3.9 %, 10.4 %), for Santos is (41.9 %, 51.8 %) and for Zuluaga is (40.9 %, 50.8 %).


NewImage

That said, this is the posterior distribution of voting intention.


NewImage

Our simulations allow to predict some interesting parameters:




  1. The probability that Zuluaga defeats Santos in the first run is 43.64 %.

  2. The probability that Santos defeats Zuluaga is 52.26 %.


Now, this is very interesting and just for Colombians: if Musa, Ñoño, and caribbean congressmen move the people into buses :) then, the forecast of the election would be as the following chart shows.  The credibility interval for blank vote would be (4.0 %, 10.5 %), for Santos would be (43.7 %, 53.6 %) and for Zuluaga would be (39.0 %, 48.9 %).


NewImage

 Under that scenario, this would be the posterior distribution of voting intention. 


NewImage

3 comments:

  1. Manoel Da Silva MeloJune 7, 2014 at 4:11 PM

    Doctor Andrés: Independiente de las opiniones políticas y de las preferencias personales -pero como observador que he visto guerra desde mi niñez en los campos colombianos- espero sinceramente que al igual que en sus predicciones anteriores para la primera vuelta no falle. Espero que los congresistas costeños se les dé la suficiente mermelada para que muevan a los electores, aunque esto no es para nada de mi agrado. Éxitos en sus proyecciones

    ReplyDelete
  2. Hola Andres, una pregunta, los votos nulos pueden hacer la diferencia en esta elecciones, de pronto la estas cuantificando?

    ReplyDelete
  3. Efectivamente pueden hacer la diferencia, pero no las tuve en cuenta puesto que las encuestadoras no reportan una estimación para los votos nulos!

    ReplyDelete