An in-house analysis by Europe Elects has shown that the international team of data analysts had the most precise forecast for the European election 2019. It compared the projection of Europe Elects, the Financial Times, and POLITICO Europe, which were the only major media which regularly published projections in the run-up to the European election 2019.
The European Parliament election took place on 23-26 May 2019. POLITICO Europe, Europe Elects, and the Financial Times had attempted to forecast the number of seats of each political group in the European Parliament. All projections for the European Parliament are based on publicly available data, i.e. polling or election results. In evaluating its own projection, Europe Elects considered the projections published in the week before the election.
The in-house analysis evaluated its performance in two ways: Firstly, it checked if the number of seats for each national list were projected correctly. For example, the German centre-left party SPD (S&D) received 16 seats. That’s what the Europe Elects projection forecasted and hence the Europe Elects projections got one national list registered as “correct”. In the case of the right-wing SNS party in Slovakia, Europe Elects and the Financial Times projected one seat. The party ended up with no seat in the election. Although the projection was only one seat off, the party is counted as “wrong” in the evaluation. This makes the evaluation on the national level very rigorous.
In total, 209 national parties and independent candidates entered the EU Parliament or were mentioned by either the Europe Elects, the POLITICO Europe or the Financial Times projection, which means that the theoretical maximum number of national lists projected correctly was 209. Europe Elects projected the seat number of 101 national lists correctly, the Financial Times predicted 88 and POLITICO Europe only 87 national parties and lists of independents correctly. Europe Elects predicted the seat number correctly for all seats in Finland, Greece, and Malta.
Secondly, Europe Elects considered the mean deviation in the number of seats for each EU Parliament group. Europe Elects was off on average by 8.6 seats from the actual result, POLITICO Europe was off 14.0 seats, and the Financial Times was off on average by 21.1 seats. The advantage for Europe Elects was grounded in the fact that Europe Elects had interviewed all relevant parties about their intended future parliament group affiliation. While POLITICO showed a hypothetical NI/new group of 43 seats, Financial Times showed a hypothetical “new parties” group with 53 seats. Before election day, Europe Elects only showed 1 MEP (Volt Europa) in this “unaffiliated” national parties category, which made the projected result for the European Parliament groups more precise than what Financial Times and POLITICO projected.
Even if we artificially exclude the NI and “unaffiliated/new” category, Europe Elects had the most accurate projection. With the artificial exclusion of the two categories, Europe Elects was off on average by 8.5 seats. POLITICO Europe, in this case, was off by 9.3 seats, and the Financial Times was off on average by 12.8 seats in the EU Parliament level.
Filip van Laenen, who is in charge of the statistical part of the projection by Europe Elects, emphasizes that Europe Elects applied a mass probability model seat projection which enabled the organization to project the number of seats for the groups in the EU Parliament more precisely than other more simple models.
Julius Lehtinen, communications officer for the organization, underlined the importance of Europe Elects having representatives all around the EU, closely monitoring the probable alignments of the MEP candidates, in Europe Elects’s success over other media outlets when predicting the EU parliament group memberships.
Europe Elects projection was regularly published by the international broadcaster euronews.