French Election First Round Results versus SMI Forecasts
Social Media Influence (SMI) forecasts for the first round of voting of the 2017 French Presidential Election compares well with actual results. Taking the main SMI forecasts that ended up being correct:
- Macron will win the most votes in the first round. SMI forecasted that Macron would win the most votes. Polls showed Macron more or less tied with Le Pen but the betting markets (via predictit.org) on the eve of voting (April 22nd) showed Le Pen with a 60% chance of winning the most votes and Macron winning the most votes at 39%. Consensus opinion right before voting appears to have been for Le Pen to beat Macron in total first round votes. SMI performed very well on this most important metric.
- Le Pen will underperform. SMI forecasted that Le Pen, contrary to most analysis, would not perform as well as her poll figures. Many observers believed that voters could have been suffering from Social Desirability Bias (not wanting to admit that they would vote for her due to a belief that supporting her publically would be viewed negatively) and/or that she could benefit from the late terror-inspired attack in Paris. SMI forecasts included, multiple times starting in December and most recently on April 22nd, weakness for Le Pen and expectations that she would end up underperforming polls in the first round. The actual vote tally for the first round shows her, in percentage terms (using the average of the last week of poll data, 17 polls), underperforming to the greatest extent of any of the four main candidates. This was an out-of-consensus call for SMI, but ended up being correct.
- Melenchon will outperform. In early December, SMI analysis stated that Melenchon could become the ‘dark horse candidate’ due to his strong SMI rating and relatively low poll and betting market figures. This came true as he bounded higher close to the end of the campaign. Additionally, SMI forecasts included him outperforming polls in the first round of voting. In fact, of the main four candidates, he outperformed polls to the greatest extent. Again, the SMI forecasts were proven correct.
- Melenchon and Le Pen too close to call. SMI forecasted on April 22nd, right before the election, that the race for second place behind an expected Macron victory would be between Le Pen and Melenchon but that it was going to be a very tight race as Le Pen would underperform polls while Melenchon would outperform polls. In fact, Le Pen did underperform and Melenchon did outperform polls. The final differential amounted to +1.7 percentage points in favor of Le Pen which is within most margins of error (traditional political polls generally have a margin of error around 3%).
And, the SMI forecast that did not go according to plan:
- Fillon will underperform. Due to his meager SMI throughout the campaign and due to the fact that his SMI was the weakest of any of the four main candidates, Fillon was forecast to underperform polls in the first round of voting. His final votes amounted to just 0.38% percent more than his poll estimates (taken as an average from the last week before the election), which means he more or less performed in-line with polls. For this forecast, SMI got it incorrect, but with no significant consequences as this less than half a percentage point did not impact the outcome of the election.
In general, SMI forecasts performed very well and better than consensus expectations in the first round of voting on most measures.
Additionally, it is important to highlight that early SMI forecasts in December more or less laid out the terrain for the remainder of the first round. Though written while many of the candidates were still making public their decisions to run and before some of the main party primaries, the forecasts proved accurate. For instance, it stated that Macron, though considered a long-shot at the time by the consensus, would end up winning the election. It also identified Melenchon as being a dark horse candidate, Le Pen as eventually declining from the then-current high expectations of the consensus, and Fillon, though being the then-current front-runner, as a relatively weak candidate who would end up declining. At the time these forecasts were way out-of-consensus and likely seemed inaccurate as they were not backed up by then-current polls, traditional political analysis, or betting markets. SMI proved not only correct but well ahead of any other forecasting method or analysis.
For the second round of voting, it currently appears that Macron will win the election without much effort. This coincides with poll and betting market data which also call for Macron winning with relative ease.