This article was made in a hurry and with little data; the definitive one is at Who voted for each candidate?
After the first round of the Chilean presidential elections, it is traditional for each candidate to begin "auctioning off" the votes of their followers to the candidate they feel most aligned with. In reality, votes cannot be "transferred" and since the vote is secret, no one knows who everyone voted for. That is why I decided to use data science to decipher how the votes from the first round were distributed to the second round.
As with any model, assumptions must be made, and in this case the key assumption we will make is that the votes from the first round were distributed in the same way across all regions in the second round (I should possibly do it by polling station or commune, but the complete database is not yet available).
What we will do is perform a linear regression with constraints of the type beta*X = y, where:
- X: are the votes per region for each candidate in the first round and the number of extra votes in the second round.
- beta: are the parameters to be estimated, which represent the percentage of votes from each of the X that went to Piñera. These values can be between 0 and 1.
- y: The votes that Piñera obtained in the second round.
When running the model, we obtain an R^2 of 0.99 with the following betas (which represent percentages transferred from the first round to Piñera in the second round):
We can see that, as expected, all those who voted for Kast and Piñera in the first round voted for Piñera in the second round; also, 75% of new voters voted for Piñera. Among other analyses, Guillier transferred 6% of his votes to Piñera, while voters for Artes and Goic did not transfer any. EDIT: I consider Navarro an error because he had very few votes.
In the following graph we present the votes transferred by candidates to Piñera in the second round:
Let us remember that the advantage Piñera had over Guillier was 635,671 votes. Therefore, if the voters of Sánchez and MEO had not supported him, Piñera would have lost the election. Now, why people who voted for them in the first round voted for Piñera in the second, only the voter knows.
ADDED: I think the assumption that votes move homogeneously could be better fulfilled by table at each polling station. I’ll see if I can get the complete database.
I would like to receive comments on this article to get ideas for making a better estimate when the database by polling station is released.
Thank you very much!
Leave a Reply