Før valget i Uganda linket vi til Elliot Green, og nå er han tilbake med analyse etter valget er ferdig.
The graph clearly shows that Museveni’s support overwhelmingly comes from rural areas and that Uganda’s incredibly low level of urbanization – which is the third lowest in the world after Burundi and Papua New Guinea – could be one reason why Museveni continues to win elections.
For alle STATA fans der ute: grafene er lett gjenkjennelig, sant?
We are pleased to welcome back Elliott Green, a Lecturer in Development Studies at the London School of Economics, with a post-election report on last Friday’s Ugandan presidential elections. See here for the pre-election report.
As expected, Uganda’s election last Friday resulted in a win for President Yoweri Museveni, who managed to up his take of the vote from 59% in 2006 to 66% of the vote (or 68% of the total number of valid votes, of which more in a minute). This marked the first time Museveni won a higher percentage of the vote than he did in a previous election, which in part is due to large-scale bribing of voters. But there are three additional factors which could explain Museveni’s support, and the release of the election data on Sunday (available here) allows us to test these theories.
First, in a pre-election article in the Ugandan Independent Melina Platas suggested that Museveni would be able to exploit his incumbency advantage as he had in the past. In particular she noted a link in the 2006 election between both higher voter turnout and higher support for Museveni – which suggested the suppression of turnout in opposition areas – as well as a link between higher levels of invalid votes and lower support for Museveni, which is more reminiscent of outright fraud.
The 2011 election data clearly shows that both turnout and invalid votes are strongly correlated with Museveni’s support across Uganda’s 112 districts, as seen in the two graphs below. Just as before, Museveni’s support increased with turnout and decreased with the percentage of votes declared invalid. (I’ve also listed the coefficient, t-statistic and R2 obtained by regressing turnout and invalid votes on Museveni’s support.)
A second theory which I wrote about in my previous post was the role of patronage, specifically the role of new districts. The results again suggest higher support for Museveni within newer districts than in older ones, as seen below:
Percentage Voting for Museveni in
New Districts (2009-2010) (n=32) 68.3%
New Districts (2006-2010) (n=42) 67.8%
All Districts (n=112) 65.6%
Districts created before 2006 (n=70) 65.0%
Districts created before 2005 (n=56) 63.2%
Districts created before 1997 (n=39) 62.5%
Districts created before 1986 (n=33) 60.1%
While voters in new districts receive the most benefit from their creation, those in the ‘mother’ districts also receive indirect benefits when their district is split, inasmuch as there is a fixed set of district administrative jobs which are then distributed across a smaller population after the split. Thus we should also expect to see an electoral effect among the 33 older districts (i.e., those created before Museveni came to power in 1986) according to when they were split up, which the data again supports:
Percentage Voting for Museveni in
Old Districts not split after 2006 (n=16) 59.8%
Old Districts not split after 2000 (n=8) 58.8%
Old Districts not split after 1997 (n=6) 56.5%
Old Districts never split (n=3) 48.9%
The three districts which have never been split are Jinja, Kampala and Kasese. Interestingly, all three districts are highly urbanized, inasmuch as they contain the 7th, 1st and 11th largest cities in the country, respectively. This then leads us to the third aspect of Museveni’s victory, namely his support in rural areas. Below I’ve plotted the relationship between Museveni’s support and urbanization levels per district – which is unfortunately only available from the last census in 2002, when Uganda had only 56 districts. (Museveni’s support is now on the vertical axis inasmuch as I’m regressing support for Museveni on urbanization.) I’ve left out Kampala as it is an obvious outlier (100% urbanization); it was also one of four districts where the main challenger Kizza Besigye beat Museveni into second place. The second-most urbanized district in 2002 was Gulu, which was subsequently broken up into Amuru, Gulu, and Nwoya districts; in last week’s election all three districts were carried by Gulu district chairman Norbert Mao.
The graph clearly shows that Museveni’s support overwhelmingly comes from rural areas and that Uganda’s incredibly low level of urbanization – which is the third lowest in the world after Burundi and Papua New Guinea – could be one reason why Museveni continues to win elections. Museveni’s rural vote could be tied up with levels of education and development more generally, but district-level Human Development Index data from the 2007 Uganda Human Development Report is not significantly associated with Museveni’s support. I suspect that his support comes in part from the greater importance of patronage in rural areas and in part from greater access to media in urban areas, but this remains something for further analysis.
In conclusion we can see how Museveni was successfully able to employ a stick and carrot approach to win the election, using both the suppression of voter turnout and valid votes in opposition areas alongside the creation of new districts and patronage more generally in rural areas as means to win him yet another 5-year term in office.
(Via The Monkey Cage.)