2016 UgandaDecides Infographics.001

After discovering 20,000 ghost voters in the 2016 voter registry, we spent yesterday deep-diving into the reported number of registered voters against the number of citizens eligible to register as voters in the country.

You can find Evelyn Namara’s analysis here. And Javie Ssozi’s analysis here.

Part of the difficulty in ascertaining the accuracy of the data presented by the Electoral Commission (EC), is that there isn’t a standard agreed upon dataset that accurately breaks down the country’s population count. For example, the population data presented by Uganda’s Bureau of Standards (UBOS) differs from that present by the World Bank. In fact, every data set we came across differs. All in all, with all data sets compared, Uganda’s population falls somewhere between 36 million and 41 million.

For the purposes of comparing oranges to oranges, we decided to compare the EC’s data against the country’s own data set (UBOS).

According to UBOS data, 56.7% (19,874,000) of the population is under age of 18. This assertion is inconsistent against their projected 2015 population of 35.8 million. The correct total should be 20,298,600 as presented in the infographic above. Accordingly, this means (an estimated) 43.3% of the population over 18 is eligible to register to vote.

Screenshot 2016-02-12 12.32.15

The EC’s (corrected) voter register report shows that 15,277,198 people registered to vote. What’s glaringly obvious in comparing these numbers is that it only 1.5% of Ugandans did not register to vote. According to the disambiguated data, that 224,202 did not register to vote. If you were to allow for a meager 1-2% statistical margin of error, this means that greater than 100% of the eligible citizens registered to vote!

How is this possible?

Without the appropriate  breaking down the voter register, it is going to be difficult to find the errors in the voter register. But if the numbers presented are to be believed, it seems impossible to have 100% of the eligible population registered to vote. Just an anecdotal tally of office mates here shows that 25% didn’t register to vote (15-person sample size).

So either UBOS is wrong in its population estimation or the EC itself has a seriously flawed database, far worse than what we found yesterday. Either way, we need better data, and better data stewards.