With the second round of the French presidential election just days away, what can election forecasts tell us about the likely result? Drawing on a citizen forecasting model, Andreas Murr, Yannick Dufresne, Justin Savoie, Bruno Jérôme and Michael S. Lewis-Beck write that Emmanuel Macron looks set to win a comfortable victory over Marine Le Pen.
As the Ukrainian war drags on, the “rally around the flag” effect is fading for Emmanuel Macron. His lead over Marine Le Pen in polling for the French presidential election has faded during the campaign. In addition, as the final vote draws closer, citizens who supported minor candidates eliminated in the first round must move their support towards either Macron or Le Pen in the second round. Who then should we expect to become the next French president?
We asked French citizens this question in late November 2021, late March 2022, and mid-April 2022. Specifically, we asked them on an 11-point scale how likely it is that each in a series of candidates was going to win the presidential election (where 0 = “very unlikely” and 10 = “very likely”). In November 2021, citizens predicted Macron would win with a mean “likely” score of 6.3, followed by Le Pen at 4.6. These numbers translated into a decisive winning vote share for Macron in the first round. This first model, four months before the election, predicted first-round vote shares of 25.4% for Macron (the actual result was 27.9%) and 18.7% for Le Pen (the actual result was (23.2%). The results for other candidates and more details are reported in an accompanying (open access) paper.
In March 2022, citizens updated their expectations, according to our second survey fielded on 30-31 March. We again asked citizens how likely they thought the election would be won by a selection of candidates. Compared to our November survey, Macron’s chances had increased on our 11-point scale (from 6.3 to 7.2) and Le Pen was also up a full point (4.6 to 5.6). This was still not a close race according to the likelihood scores.
Of the other candidates, Jean-Luc Mélenchon had risen to third place, increasing his score by more than a point from 2.5 to 3.6. Valérie Pécresse and Éric Zemmour were below Mélenchon, respectively, at 3.5 and 3.3. None of the other candidates scored over 2.0. These findings suggested a clear victory for Macron. If we translate the “likely” numbers into vote shares, they predicted a first-round vote share of 21.5% for Macron (actual result 27.9%), 16.7% for Le Pen (actual result 23.2%), 10.8% for Mélenchon (actual result 22%), 10.5% for Pécresse (actual result 4.8%), and 9.8% for Zemmour (actual result 7.1%). While the forecast order and magnitude are correlated with the true results, the most successful candidates were underestimated while the less successful candidates were overestimated.
In our third and final survey (fielded 13-14 April), we asked the same citizen forecasting question as above, but restricted it to the two candidates in the second round. We found that in terms of average likelihood, Macron obtained a score of 7.2 and Le Pen a score of 4.9. This translates into a prediction that Macron will prevail with a vote share of 59% to Marine Le Pen’s 41% in the second round. For the sake of comparison, as of 21 April, a simple average of the last few days of official polling would predict a 56-44 victory for Macron.
How much should we trust the forecasts of citizens? The evidence suggests quite a bit, though caution with forecasts is always advisable. As with any other forecasting method, sometimes citizen forecast models get it wrong. Nevertheless, it turns out that citizen forecasts tend to be more accurate than voting intentions when it comes to predicting elections. For instance, in the 2019 British general election, citizen forecasting produced the second most accurate forecast for the Conservative Party among 18 forecasting approaches.
Given the stellar performance of citizen forecasting in the 2017 French presidential election, coupled with the continued comfortable lead of Macron across our three surveys – two before the first round and one before the second round – we have some confidence in our forecast of a Macron victory in 2022.
Note: This article gives the views of the authors, not the position of EUROPP – European Politics and Policy or the London School of Economics. Featured image credit: European Council