The following processes were carried out to calibrate the weighted average for Pollwatch, a collaborative project between EUobserver and EM Analytics, in the context of the European elections in June 2024.
1. Reliability ranking of pollsters
The data from all the polls published for the last three elections in the member countries were compiled. The deviation in the estimation of each party has been calculated from the vote data of the last projection of each pollster, excluding exit polls. The pollster's deviation in each round is the sum of the absolute values of the vote deviations of the political parties. The average deviation, which results in the pollster ranking (Euro PollCheck), is the average of the deviations of the last three polls per pollster and country. [A link to the table with the Pollster-Country-ranking will be included].
2. Participation estimation
To calculate the turnout data, the electoral census has been used as a starting point, based on the most recent data on the population over 18 years of age available in Eurostat for each country. To estimate the turnout, an average of the turnout in the last three European elections was used for each country, assigning 50 percent weight to the 2019 figure and the remaining 50 percent to the average turnout in 2014 and 2019. [A link to the table with the Census - Participation ratio per country will be included].
3. Vote estimation
The vote estimate is calculated using the weighted average of the latest polls published for the European elections in 2024 by each pollster. Each pollster is assigned a specific weight per country, according to its position in the historical deviation ranking. A minimum of 5/10 is set for those with a deviation greater than two times the average error of pollsters in that country, and a linear ranking between five and 10 for deviations between 2xMAE and 0.
4. Estimation of MEPs
Once the percentage of the vote per party has been calculated, the number of voters who would support that political party is estimated (based on the census and the turnout) [census * turnout * percentage]. The estimation of deputies is made according to a proportional distribution using the D'Hondt system, considering the specific minimum thresholds necessary to obtain representation per country. To do this, the estimated vote for each party is divided by correlative numbers and those coefficients equal to or greater than the last coefficient that gives entry to the European Parliament are assigned as MEPs (provided that it is greater than or equal to the entry barrier, otherwise, 0 MPs are assigned).
5. Grouping of data
For the vote calculation by political families, the voters of the political parties belonging to each family per country are added up, and their percentage of the total census and estimated turnout is calculated. Similarly, the aggregation of voters by family in the different countries allows us to obtain the overall figure for support by family in the Union as a whole. The MEPs by political families are the result of the sum of the MEPs of the parties that are attached to these families.
Have questions? We're all ears! Feel free to reach out and delve deeper into our methods or insights. Send us an email at [email protected] or at [email protected]