Last Saturday, Niger’s electoral commission released provisional results from the December 27, 2020 general election for President and National Assembly. Mohammed Bazoum, Interior Minister for incumbent President Issoufou, secured 39.33% of the vote for President while former President Mahamane Ousmane received 16.99%.
- In 2016, President Issoufou won 48.43% of the vote in the first round of voting in 2016 versus 17.73% for the second place candidate.
- In 2020, voter turnout was slightly up over 2016: 69.67% versus 66.82%.
In the battle for Niger’s National Assembly, Bazoum’s party, the Nigerien Party for Democracy and Socialism (PNDS) won 80 out of 166 seats–five more than the party’s 2016 haul (and an even larger percentage of seats since there were more overall seats in 2016).
At the same time on Saturday, 100 civilians were reported killed between two towns, Tchoma Bangou and Zaroumadarey, near Niger’s capital, Niamey The attack was one of the deadliest in the country’s history and demonstrated the security issues that have been a focus of this year’s election.
- For context, here are where attacks by violent non-state actor groups have taken place in the Sahel region over the last 3 years:
- Two French soldiers were also killed by an improvised explosive device (IED) in neighboring Mali on Saturday, putting pressure on greater efforts to stymie violence in the Sahel region.
Since no presidential candidate received 50%+1 of the vote in the first round of voting, the election will head to a second round vote where Bazoum and Ousmane will directly face-off on February 21st, 2021. As a result, coalition building will become crucial.
- Both candidates have already begun speaking with other contenders from the first round to form electoral blocs, aiming to build a sufficient enough vote share to win in February. Ousmane began this process before the first round of voting, which helped ensure a second round vote.
In the second-round vote, Bazoum is the clear favorite: We gave him a 91.69% chance of winning it originally.
For caution!: Our pre-election prediction gave Bazoum a 10.55% (89% CI at 9-15%) chance of winning the election in the first round of voting. Although we were in some respects correct since the election is heading to a second round vote, it is clear looking at where our model thought Bazoum would perform that our first-round prediction missed the mark.
What went wrong with our prediction?
To be truthfully honest, we aren't entirely sure yet, but there are three key factors we've been able to isolate so far:
Calculating Bazoum's Current Position
In our original prediction, we calculated Bazoum's current polling position based on an aggregate of different adjusted polling in Niger. However, when looking at the different models to blend, we over-weighed the ones which had Bazoum over-performing while under-weighing models which factored in youth turnout or how the capital, Niamey, would vote.
Calculating Historical Base Rate
When calculating the round 1 base-rate (but not when calculating the round 2 base-rate), we forgot to adjust for the lack of incumbent in this election. This is actually a pretty big deal, especially because Issoufou, who was the incumbent in 2016, had been a top-two presidential candidate in 4 elections since 1999. Bazoum had not. This factor seems to be particularly important given the relative performance of the PNDS over Bazoum.
Factoring in Local Election Results
Although not a focus of our original article, Niger also held long-delayed local municipal elections earlier in December, 2020. The results from that election were strong for the PNDS, winning roughly 42.5% of all council seats available. However, such a result was both a far cry from the 50% + 1 needed to win in the round 1 vote, and the numbers included in our prediction model.
Predicting the Second Round Outcome
Based on our reflections of our round-one prediction, we believe that our second-round vote prediction is in need of some fine tuning and possibly more than that. We'll release a full, re-adjusted prediction a few days prior to the February vote. For now, there are two opposite ways in which we are viewing the round 1 vote to inform our updated round 2 prediction.
The first approach is to retrofit our model to "make sense" of the results from the first round of voting in different ways, evaluate the different approaches to make sense of what happened and determine which changes fit best, and then port those changes to our round 2 prediction. We can approach this either as being off in round 1 by:
- A certain percentage change; or
- A certain percentage point amount
The latter is more bullish than the former on Bazoum's chances of winning, although both options paint a second-round vote which is, more-or-less, a toss-up.
The second approach is to adjust the round 1 results based on the top-two results, either by candidate (Bazoum vs Ousmane) or by power coalition (in-power vs opposition). Conversely to the first approach, the former option here is more bullish than the latter–although also conversely to the first approach, both options for the second approach paint a race where Bazoum remains the favorite.
|Bazoum Adjusted||Ousmane Adjusted||Type|
|69.83%||30.17%||Top 2 Share|
|60.26%||38.74%||Top 3 Share|