Projects
In the Mexican elections, the quick count consists in selecting a random sample of polling stations to forecast the election results. Its main challenge is that the estimation is done with incomplete samples, where the missingness is not at random. We developed several statistical models used in the quick count of the 2018, 2021, 2022, 2023, 2024 and 2025. Here we explain in Nexos the problem in detail, here is an R package with the model implementation. To find out more visit Talks and Research.
%2014_41_19.png)
We developed a prediction model that aggregates results from more than 40 polling firms and incorporates their historical performance. The model simulates possible electoral scenarios to produce projections for election day. It accounts for the track record of individual polling houses, the dynamics of Mexican electoral races, and has been validated against past elections to ensure it properly reflects the uncertainty of potential outcomes.
We produced poll-of-polls models for the presidential race and gubernatorial elections in five states. You can explore one example here: Presidential Poll of Polls 2024.
Our Electoral Predictor used data from the Preliminary Electoral Results Program (PREP) to estimate voting trends in the 2024 presidential election, as well as in the Mexico City mayoral race and the elections for both the Chamber of Deputies and the Senate.
The model was presented live on election night starting at 8 PM. The official electoral results released later were fully consistent with all the intervals we had presented.
Expansión Política
This model ranked potential presidential candidates for the 2024 elections using a wide array of surveys, each including a different subset of the roster of plausible candidates. The model corrects for polling house biases, and dynamically updates preferences. Throughout 2023, the Power Ranking provided insights into public opinion and media influence before the announcement of the official candidate list.
COVID-19 State-Level estimates for Mexico
We fit the semi-mechanistic Bayesian hierarchical model found here to describe the Mexican COVID-19 epidemic. We obtain two epidemiological measures: the number of infections and the reproduction number. Here we explain the model in detail and the assumptions made for Mexico and here we show state-level estimates (latest update is August 2020). We also published an article in the Mexican newspaper Nexos explaining the model results here.
%2014_46_38.png)
