The Federal Republic of Nigeria held its latest quadrennial presidential election. Most polls showed the challenger, Muhammadu Buhari, ahead of incumbent Goodluck Jonathan. None of the polls, however, were believable in terms of sampling method, age, and weighting. But, broken clocks and poor polls can be right once in a while. Buhari won and Jonathan willingly stepped down. This is the biggest step a State can make on its way to full democracy.
What can we learn about the election?
Now that the election is over, and no one is claiming widespread fraud, we can examine the reported results to see if there are any interesting patterns. The following sections will deal with the regions of Nigeria supporting each candidate, the spatial distribution of the invalidation rate, and a look at the invalidation plot for this election.
Recall that Nigeria is a strongly divided country. This actually goes beyond the Christian-Muslim (South-North) split common to African States at that latitude. Nigerian society remains divided based on ethnicity and on tribes. This helps explain why the Jonathan-Buhari split did not echo the religious split. The following map illustrates the level of support at the state level for Buhari.
Figure 1: Map of the level of support for Buhari at the
state level. Darker blues indicate higher support.
As expected, Buhari found his greatest support in the Muslim north. He did receive a lot of support in the Christian south, however. His lowest levels are in the delta region, where the Igbo and Ijaw ethnicities are strongest. It is in the extreme delta region that Goodluck Jonathan found his greatest level of support, as evidenced by the following map.
Figure 2: Map of the level of support for Jonathan at the
state level. Darker reds indicate higher support.
Note that the pattern in this election is reminiscent of the patterns in the 2011 election (article: The epilogue?):
Figure 3: Map of the level of support for Jonathan at the state
level in the 2011 election. Darker blues indicate higher support.
In this 2015 election, however, Jonathan lost votes outside his homeland. This was the difference between re-election and not.
Candidate support is only a part of the equation. The invalidation rate is consistent with the probability of a person's vote counting. This map shows the invalidation rate at the state level.
Figure 4: Map of the invalidation rate reported at the state level. Darker greens indicate higher levels of invalidation. The dark spot in the south of Nigeria is Ebonyi state.
Note that there is variation across Nigeria. However, there is no discernible pattern to it. Had the north a higher invalidation rate than the south, then one could conclude that there is evidence votes are invalidated for different reasons in the north than in the south. Here, I do not see any spatial clustering of the invalidation rate.
Finally, let us examine the invalidation plot. The graphic plots each state according to the invalidation rate and the level of support for Buhari. If the election is free and fair, then we would expect the regression line to be horizontal, or nearly so.
Figure 5: The invalidation plot with a 95% Working-Hotelling confidence band. Note that a horizontal line can be fit inside the band, suggesting that the pattern seen is nothing more than randomness.
Figure 5 is the invalidation plot for this election. Note that Ebonyi is an outlier state; its invalidation rate is significantly different from that of the other states. With or without Ebonyi in the model, a horizontal line (the null hypothesis) is able to fully fit within the confidence band. This suggests no evidence of differential invalidation, of invalidating ballots simply because they were for the wrong person.
Last week, our projection (article: Goodluck on Saturday, Jonathan!) was
With all of this said, we predict this close election will be won by Goodluck Jonathan. Unfortunately, there will be severe riots as a result.
We were wrong. Sometimes, it feels great to be wrong!
Not only were we wrong about the outcome of the election, but we found no evidence that the election outcome was anything other than free and fair.
Well done, Nigeria! Now it is time to get your daughters back.
|Official electoral body:||Independent National Electoral Commission (INEC)|
|Election date:||March 28, 2015 (postponed from February 14, 2015)|