Analysis of Results Suggests Large-Scale Vote Tampering in
Ethiopian News and Views
May 26, 2005
On May 26, 2005 the NEBE released an additional three results from East Gojjam. Two were EPRDF wins, and one was a CUD victory. The result from Bichena, continues to be incomplete because the turnout was not reported.
The CUD won 15 of 16 seats in West Gojjam, the exception being Bure Wereda. The EPRDF won 15 of 17 seats in East Gojjam with the exception being the towns of Debre Markos and Mota. (note: Bahir Dar, which was won by the CUD is a separate zone and not included in West Gojjam).
The results are summarized in the Table below which is sorted by adjusted turnout. The adjusted turnout is simply the percent of votes cast that were actually counted. In any election, some votes are disqualified because the ballot is spoiled. Examples could be where the voter made no mark (under-voting) or where the voter marked more than one choice (over-voting).
The summary table below shows the large discrepancy between East Gojjam and West Gojjam in the adjusted turnout. In West Gojjam the median adjusted turnout was 60.3% while in East Gojjam it was 13.4 percentage points higher at 73.7%.
In the two largest towns, Debre Markos and Mota, the adjusted turnout averaged 73.4%. Both these towns are located in East Gojjam and were the only weredas in East Gojjam that were won by the CUD. Without these towns included, the median adjusted turnout in East Gojjam would rise to 74.6%. In other words, rural turnout in East Gojjam was higher than the turnout in the towns, and the rural voters made fewer mistakes on their ballots than the townspeople did (invalid % was lower in the rural areas than the towns).
In West Gojjam by contrast, the situation was more as expected, with lower turnout as compared to Debre Markos and Mota, and higher percentage of spoiled ballots.
Summary Table of Wereda Level Results from Gojjam, May 26, 2005
Another large discrepancy between East and West Gojjam was in the relationship between turnout and winning percentage. In West Gojjam, there was no relationship between adjusted turnout and the CUD margin of victory. In East Gojjam however, there is a strong relationship between the EPRDF margin of victory and the adjusted turnout (see figures below).
Why the difference? We would expect the margin of victory to vary with local factors at the Wereda level, such as the candidates popularity and the degree of publicity. As such, no clear trend would emerge between the turnout and margin of victory. If anything, we would expect a huge turnout to reduce the EPRDF margin of victory (or cause a CUD win). But East Gojjam produced the opposite trend.
In the absence of other explanations the most likely scenario was that the vote counting/reporting was tampered with. This would explain why turnout would increase mysteriously in rural areas and this increased turnout would be overwhelmingly pro-EPRDF. It would also explain why the percentage of spoiled ballots decreased: - if someone is manufacturing votes, they don't make many mistakes.