Patterns of Turnout in the Ethiopian Election of May 15, 2005 and Implications for Validity of the Results. Part 3 - Zonal Trends

Draft Report,
Ethiopian News and Views
June 28, 2005
(updated July 1 with information from the investigation list posted by the NEBE on its website)


Note: The discussion below refers to average turnout by zone. The average was calculated after first excluding unopposed candidates and clearly erroneous turnout (greater than 100% or less than 40%). Unopposed candidates were excluded because the EPRDF won all these weredas with vey high reported turnout (average over 93% nationwide) and these weredas would mask the patterns in the contested weredas. If these weredas were included they would further increase the discrepancy in average turnout between opposition-won weredas and those won by the EPRDF.

AFAR

The EPRDF-allied ANDP won all 8 parliamentary seats in the AFAR region with an average winning percentage of 86%. ARDUF candidates finished second in 5 of the 9 races while the CUD was second in the remaining three.

The results from the Afar Region are curious for several reasons. Afar Region averaged 95,955 voters per wereda, whereas the national average was 49,907 registered voters per wereda. Afar region has five of the eleven weredas nationwide with more than 100,000 voters - either the region is badly underrepresented or the voter rolls have been inflated.

The Afar Region population was given as 1,206,338 in 1996 (Parliament website). The UNFPA lists the Ethiopian median age as 17.2 years and the population growth rate as 2.5%. Thus the Afar population should be around 1,381,665 in 2005 and the number of people older than 17.2 years would be 690,832. So the number of eligible voters in the Afar Region is actually less than the number of registered voters (767,636). This in itself should invalidate the election, particularly in five of eight weredas where the number of registered voters was vastly inflated. These five weredas cover mostly the thinly-populated desert areas of Afar including the Dallol area.

It is interesting to note that the two weredas where the ANDP winning percentage was relatively moderate (below 70%) were also the two weredas with the lowest number of registered voters (average of 28,000 compared to average of 119,000 in the other six). These weredas are in the more-populated upper Awash valley (Awash Fentale and Amibara).

The result in Gulina is highly suspect - 98% turnout with only 18 of 111,600 ballots ruled invalid, and the ANDP won 97.9% of the vote. This is the third highest winning percentage in the country outside of Tigray (excluding unopposed candidates). In fact four of the eight weredas in Afar had win percentages above 90% - in no other Region other than Tigray were the results so one-sided. Even the prominent ARDUF leader Mahomooda Gaas could not get more than 6% of the vote in Alaba wereda.

Another standout for the Afar Region is that three of the eight weredas had invalid ballot percent of zero or virtually zero. The average and the median invalid percent (5.7%, 2.5%) was actually lower than that of Addis Abeba (7.6%, 6.4%) - a dubious outcome that is inconsistent with comparative socioeconomic parameters (e.g. literacy).

Update: As of July 1, 2005, the NEBE will not be investigating any constituencies in the Afar Region although complaints were presented for all eight weredas.

TIGRAY

The people of Tigray experienced the least open election of any region in the nation. Fully 16 of 38 weredas (42%) were single candidate elections where the population had no alternative to the EPRDF. Despite the large proportion of uncompetitive races, the reported regional turnout of 93.9% was the highest in the nation.

The EPRDF/TPLF win margin did not drop below 85% except in Erob Liyu wereda where it was only 53%. The population of Irob was directly affected by the Eritrea war, and discontent with Meles Zenawi's handling of the arbitration no doubt contributed to the near upset victory of the independent candidate running in opposition to the TPLF.

BENCH MAJI

The SMPDUO (Sheka and Mezenger Peoples Democratic Unity Organization) won Sheka Liyu wereda in Bench-Maji zone while the EPRDF took the other six weredas. The SMPDUO is not listed as belonging to one of the two opposition coalitions and it won running against an EPRDF (SEPDC) candidate. Turnout was 87% in Sheka-Liyu wereda, which is much higher than the 79.9% average elsewhere in the zone. This was the largest positive difference between opposition and EPRDF turnout nationwide.

DAWRO

There are only three weredas in Dawro zone - all won by the EPRDF - but one of them, Loma Bosa, reported both the highest turnout (287%) and the highest inferred invalid ballot percent (82%) in the entire nation. This may be due to a typographical error on the reported number of votes cast. But it should also be noted that Loma Bosa was the closest result in the nation: the EPRDF won by 48.65% to 48.32% for the opposition, and the margin of victory was only 75 votes.

Update: The NEBE edited the result from Loma Bosa wereda to change the number of votes cast from 126,875 to 26,875. This is an improvement but the turnout is now abnormally low at 61%. The NEBE has decided to investigate all three weredas in Dawro zone in response to complaints by the UEDF.

GAMO GOFA

Twelve of eighteen weredas in Gamo Gofa zone were won with winning percentages of less than 60%. As usual, turnout in the EPRDF-won weredas exceeded that of the opposition-won weredas (75.3% vs 73.8%). The opposition won Arba Minch, Chencha, and three other weredas, while the EPRDF won 15 weredas. Arba Minch and Chencha both recorded 75% turnout, and only Zela Deremalo wereda had higher turnout (77%) among the opposition weredas. Most EPRDF weredas had turnout that exceeded that of Arba Minch, although a subset, (Selamber and Kemba) had relatively low turnout (58% and 66%). These two weredas were won by the EPRDF with 50.7% and 49.7% of the vote respectively.

Update: Ten of the eighteen weredas will be investigated by NEBE as well as two addtional weredas that did not have elections (Basketo Liyu and Basketo Regular). Selamber and Kemba will be investigated, but Melo Koza (37% invalid ballots) and Arba Minch Zeyse Liyu (0% invlalid ballots) will not be investigated. Eight EPRDF and two CUD wins will be investigated.

GEDEO

Gedeo zone is located directly south of Sidama zone and contains nine weredas. The opposition won only one: Wenago 2, which contains the large town and zonal capital of Dilla.

Gedeo is one of the few zones nationwide where opposition turnout exceeded that of the EPRDF. This is due to the influence of Dilla on the turnout (79%) in Wenago 2. This turnout is typical of large towns across Ethiopia, and the invalid ballot percentage (11%) is also normal.

But the EPRDF managed to win three rural weredas with even higher turnout. Bule wereda in particular is a standout with 93% turnout and only 3.4% invalid ballots. Two other weredas produced interesting results, Wenago 1 with only 56% turnout (7th lowest in the nation) and Wenago 2 with 33.6% invalid ballots (7th highest in the nation).

Update: Seven weredas are scheduled for investigation, including the only win by the oppositon (Wenago 2).

GURAGE

Gurage zone was won by the opposition with 10 weredas to 3 for the EPRDF. Two of the EPRDF weredas are anomalous outliers due to turnout (19% in Gumer 1) and invalid ballots (0% in Meskan & Mareko 2). The third wereda (Yem Liyu) is not anomalous as such, but is interesting because it recorded the highest turnout in the entire zone (86%).

Update: Emdibir (won by CUD), and Meskan & Mareko 2, and Yem Liyu (won by EPRDF) are scheduled for investigation by NEBE while Gumer 1 is not. Gumer 1 continues to show 19% turnout and 0 votes for the opposition in the NEBE database on the web.

HADIYA

The opposition won Hadiya zone, taking 8 of the 10 weredas, but as usual, the EPRDF achieved higher average turnout (74.0% vs 68.1%). The EPRDF also averaged lower invalid ballot percentage (13.4% to 17.8%), which is also a typical pattern. Beyene Petros won in Sike 2 wereda where turnout was 75% - second highest for the opposition.

Update: Lemo 2 (won by EPRDF) and Konteb 2 (won by UEDF) will be investigated by the NEBE.

KEFFA

The EPRDF swept all seven weredas in Keffa, winning on average 82% of the vote. One of the most lopsided results in the country occurred in Chiri Liyu wereda where the opposing candidate got only 10 votes (out of 1704 cast). In fact three of the eleven most lopsided contests nationwide (by winning margin) took place in Keffa zone (excluding Tigray and Addis Abeba).

The closest contest took place in Gimbo wereda which contains the largest town, Bonga. Here the EPRDF barely scraped through with 50.6% of the vote. Turnout in Gimbo was 89% but four other weredas had turnout ranging from 92%-99% and the EPRDF won overwhelmingly in those weredas.

Update: Three of the seven weredas will be investigated by the NEBE, including Gimbo wereda.

BALE

Eight of eleven weredas in Bale zone were won by the EPRDF. As usual EPRDF weredas had higher average turnout (90.5% vs 88.5%) and lower invalid ballots (12.5% vs. 18.9%). Goba (largest town) had turnout of 94% and was taken by the EPRDF with only 45.9% of the vote. Six weredas had turnout over 90% and all were won by the EPRDF. Three of these weredas also had less than 10% invalid ballots - an improbable combination in rural Ethiopia.

Update: Three weredas won by the EPRDF (Goba, Kokosa, and Siraro 2) will be investigated, but some of the weredas with a combination of improbably high turnout and low invalid ballot percentage will not be investigated currently (Ginir, Mena, Jara).

BORENA

The lowest zonal turnout nationwide occurred in Borena zone where the opposition won five weredas with an average turnout of 56%, while the EPRDF won one wereda (Yabelo) with an 83.4% turnout. Low turnout in Borena zone is almost certainly related to the disruption caused by ethnic conflict in the region.

    NEW DISPLACEMENTS DUE TO CONFLICTS IN OROMIYA REGION
    April, 2005. Source: http://www.uneue.org/

    The ongoing tribal conflict between Guji and Gabbra in southern Oromiya has resulted in massive displacement of people. In April an NGO reports as many as 50,000 people in Hagere Mariam, Yabello and Arero have been displaced and several thousand huts burnt. The DPPC has sent an assessment team to the affected zone but was unable to verify the figures because not all affected areas were accessible due to poor road conditions and security concerns. There is also a report of an additional 7,000 displaced people in the border area of Moyale and Borena zones. The displacement in Moyale and Borena is reported to be a result of conflict between Oromos and Somalis. According to Somali regional authorities the displaced are in need of food, shelter and medical attention.

Yabelo wereda was reported as also affected so it is not clear how the EPRDF managed to generate 83% turnout there while the other weredas couldn't exceed 66% and two weredas had only 48% turnout.

Having said that, one must also note two serious anomalies in the weredas won by the opposition: invalid ballot percentage is inferred to be only 1% in Moyale and an impossible -6% in Kercha. One needs to examine the individual polling station data to track the source of these anomalies.

Update: No investigations in Borena zone are scheduled.

KEMBATA TEMBARO

The six weredas of Kembata/Tembaro zone were closely contested with an average winning percentage of only 54%. The opposition won five of the weredas with an average turnout of 72.3%, and an average invalid ballot percentage of 20.8%. The only wereda won by the EPRDF (Tembaro) had the second highest turnout (77%) and the lowest invalid ballot percentage (15%) in the zone.

Update: Tembaro wereda will be investigated.

SIDAMA

The opposition won 3 of 16 weredas in Sidama zone, including the largest town of Awasa. Although turnout levels were moderate overall, as usual, the EPRDF had higher average turnout (77.4% to 74.1%).

EPRDF turnout was lowered by the three easternmost weredas (Belela, Bensa, and Aroresa) where turnout was less than 60% - substantially (more than 2 standard deviations) below the average for the zone. It is unknown why the turnout was abnormally low here. The Sidama Liberation Movement (SLM) was in second place in two of these weredas and two of these were won by the EPRDF with less than 50% of the vote. The SLM did manage to win a seat in Guguma wereda.

Surprisingly, aside from the three weredas discussed above, Awasa had the lowest turnout in the zone (72%).

Update: The SLM won a second wereda (Arbegona) which was announced after a month delay. The NEBE will investigate eight weredas including Arbegona and the other SLM win in Guguma. Two other weredas that didn't have elections are included as well (Hagere Selam and Bursa). Bensa and Aroresa will be investigated but Belela will not. In addition, a wereda won by the CUD with normal turnout (Tula) will be investigated.

WELAYTA

Only two of the 13 weredas in Welayta zone were won by the opposition. Six of the weredas have no data listed other than the winner and winning vote %. The opposition won the two weredas of Damot Gale averaging turnout of 73.4%. This exceeded the EPRDF wereda turnout average of 72.3%. The reason for this is the low turnout in Bolosa Sore 1 and 2 weredas (57%, and 58%). Why the turnout was so low in these two weredas is unclear.

Update: Five weredas won by the EPRDF will be investigated - four of these had no turnout data released. Bolosa Sore (1 and 2) will not be investigated and neither will be Ofa, which reported a highly unlikely invalid ballot percentage of only 1%.

WEST SHEWA

The opposition won 14 of 15 weredas in West Shewa with an average turnout of 87.9%. Turnout in these 14 weredas was fairly similar, ranging from 85.3% to 91.4% with a standard deviation of 1.9%.

The one wereda won by the EPRDF (Addis Alem) had turnout of 95% - more than three standard deviations above the opposition average.

The only other notable feature of this zone is that it contains the only two opposition weredas nationwide that had the anomalous combination of turnout > 90% and invalid < 10%. These weredas, Ambo 2 (90.9/8.2) and Kachisi/Gindeberet (90.1/9.2) are barely in the anomalous range. The result in Ambo 2 is reasonable given that Merera Gudina was the opposition candidate. Polling station level data should be looked at in Kachisis/Gindeberet to check whether any of the polling stations are anomalous.

Update: Addis Alem has been scheduled for investigation by the NEBE along with Welmera (won by CUD with normal turnout).

SOUTHWEST SHEWA

The EPRDF won only two of nine weredas in this zone. This is one of the few zones where the opposition-won weredas had turnout exceeding that of the EPRDF-won weredas (87.3% vs 83.4%). One wereda won by the opposition has turnout listed as 100% (the only such instance for the opposition nationwide).

Update: The two weredas won by the EPRDF (Teje and Weliso 2) have been scheduled for investigation.

EAST SHEWA

East Shewa zone was won by the opposition (10 weredas to 4). Comparison of turnout is made difficult because of the four EPRDF weredas, one (Adama 3) has no turnout information, and another, (Alem Tena) has impossible turnout of 143% (EPRDF barely beat the OFDM here). The other two, (Siraro 1, and Ziway) averaged 68% turnout, and anomalous invalid ballot percentages of 58% and 39% respectively. Turnout in the opposition weredas averaged 85.6%, so the EPRDF weredas were either 4 standard deviations above or below the turnout in the opposition weredas. Opposition turnout was elevated (average 88%) in the weredas adjacent to Addis Abeba such as Akaki , and Ada 1 and 2 (Debre Zeit).

It would be hard to argue that the opposition rigged the election in this zone because the CUD, UEDF, OFDM, several smaller parties and independent candidates all competed against each other in many of the races. In two weredas, Adea 2 and Welenchiti, the EPRDF was dropped to third place - the only such occurrences nationwide.

Adama 3 was one of the closest results in the nation with a winning margin of only 0.8% (213) votes so it is critical to have the turnout data.

Update: Six weredas will be investigated: 2 won by the EPRDF (Adama 3 and Zeway) and four won by the opposition (Adama 2, Ada1, Ada 2, and Akaki). Alem Tena will not be investigated: - the NEBE recently changed the number of registered voters in the wereda from 27,688 to 46,759, so the turnout is now 85% instead of 143%. Siraro 1 (58% of ballots were invalid) will not be investigated.

NORTH SHEWA (OROMIA)

The average winning percentage in North Shewa zone was only 52%, indicating that the zone was highly competitive. Four different parties won seats, but the EPRDF took six of the ten weredas and averaged much higher turnout (82.1% vs 73.7%). Opposition weredas did not exceed 79.5% turnout while the EPRDF had a maximum of 92% in Wichelena Jida wereda.

Update: Four weredas will be investigated: 3 won by the EPRDF (Degem, Sululta & Mulo, Wuchalena Jida) and one won by the UEDF (Kuyu).

ARSI

Of the eighteen weredas in Arsi zone, the opposition won 7. As usual, average turnout for the opposition weredas (84.5%) was lower than that for the EPRDF weredas (90.1%). Maximum zonal turnout (96%) was in the EPRDF-won wereda of Aseko, while the lowest turnout (75%) was in the zonal capital town of Asela, which the opposition won.

Elevated turnout throughout the zone is indicative of systematic manipulation, but it would be hard to suggest that the opposition was responsible because three separate opposition parties won seats and competed against each other.

Update: Five weredas (four of them EPRDF wins) will be investigated but not Aseko, probably because the EPRDF win margin was so high.

WELLEGA

The election in Wellega exhibited a number of contradictory features. By one standard, it was among the most open contest in the nation, with three opposition parties and one independent candidate winning seats in addition to the EPRDF. But it was also a very uncompetitive election by another measure: the EPRDF was unopposed in 12 of the 31 weredas (39%) - the highest percentage of noncompetitive weredas outside of Tigray.

The highest turnout for opposition wins was in Enango/Bila wereda where the leader of the OFDM, Bulcha Demeksa won election with 95% turnout. The opposition achieved an 89.9% turnout in the 17 weredas they won in Wellega (east and west combined) - the highest for any area outside of Addis Abeba. But the EPRDF managed to find even more votes in Wellega; it achieved an average 95.4% turnout in the 14 weredas it won, even though in 12 of these 14 weredas, the EPRDF candidate ran unopposed. The two contested weredas won by the EPRDF had a combined turnout of 94.9%.

It would be hard to argue that the opposition rigged the vote in Wellega, as opposition wins were split amongst the OFDM (7), UEDF (6), CUD (3), and one independent (Dr. Negaso Gidada).

Update: Two weredas (both EPRDF wins) will be investigated in East Wellega and none in West Wellega. The EPRDF was unopposed in Alibo wereda but it has been selected for investigation (likely related to ballot access for the opposition - a problem widespread in Wellega). Gelila wereda has also been selected (99% turnout) .

DIRE DAWA

The two weredas in Dire Dawa were split between the opposition and the EPRDF-allied SPDP. Higher turnout was recorded in the SPDP wereda (87% vs 84%). The SPDP won with only 47% of the vote while the opposition took 60%.

Update: The NEBE is currently not planning to investigate Dire Dawa.

EAST HARERGE

The EPRDF won 12 of 14 weredas in East Harerge, and as usual, turnout was higher for the EPRDF weredas (86.6% to 83.9%). The most interesting result is from Alemaya, which was divided into two weredas and exhibits a large discrepancy between the weredas.

Alemaya 1 Alemaya 2
Winner UEDF EPRDF
Turnout 82% 65%
Invalid Ballots 17% 68%
Winning% 81% 58%

Update: The NEBE is currently not planning to investigate any result in East Harerge.

JIMMA

Jimma zone was swept by the EPRDF with turnout averaging 91%. Two of the 17 weredas were sole candidate elections and one has incomplete data. A striking feature of the results is that 5 of the weredas had a combination of turnout greater than 90% and invalid ballots less than 10%. Yet the wereda containing Jimma town (Jima Kersa) reported turnout of only 82.6% - more than two standard deviations below the zonal average and the lowest turnout in the entire zone. The invalid ballot percentage was also relatively high at 13.2%. Why would voters in the largest town in the zone vote at much lower rates than voters in the rest of the zone?

Update: Jimma town is a separate wereda that was announced as won by the CUD only in the past 2 weeks - more than a month after the election. No turnout details were provided. The NEBE will investigate 11 weredas, including the only opposition win in Jimma town.

GUJI

Of the four weredas in Guji zone, the opposition took one (Bore) with 78% turnout. The EPRDF-won weredas exceeded the turnout in Bore, averaging 79%. Kibre Mengist was won by the EPRDF with 49% of the vote but no other details were made available for this town.

Update: The NEBE will investigate Kibre Mengist and Negele (both EPRDF wins).

NORTH-SHEWA (AMHARA REGION)

The 18 weredas of North Shewa zone went 12-6 for the opposition. This was one of the few zones in the entire country where opposition weredas matched the turnout in the EPRDF weredas (86.0%).

Overall this was a "clean" zone with few outliers other than Alem Ketema (EPRDF), where the results have been changed three times in the NEBE website (the turnout was variably reported as 18%, 100% and then 85%).

The other outlier of interest is Ancober (opposition) where the 94% turnout was highest in the zone. The opposition won only 54% of the vote in the wereda, so it is likely that the EPRDF won some of the individual polling stations.

Update: Five weredas (all EPRDF wins) will be investigated, including Alem Ketema.

BENSHANGUL-GUMUZ

Turnout in the Benshangul-Gumuz Region was very high (90.5%) and even exceeded that of Addis Abeba - only Tigray had higher turnout. Turnout is particularly important here because four of the nine weredas were won with less than 50% of the vote. The opposition only won one wereda (Bambasi), and as usual, this wereda had lower turnout as compared to the EPRDF weredas and higher invalid ballot percentage.

Update: The NEBE is currently not planning to investigate any result in Benishangul Gumuz Region.

AGEW AWI

The opposition won Agew Awi zone, taking four of seven weredas, but weredas won by the EPRDF averaged significantly higher turnout (88.1% vs. 80.9%).

The only anomalous result occurred in Yechereka wereda (won by the opposition) where the reported number of votes cast is less than the sum of votes for individual candidates (invalid < 0%). Why this happened needs to be investigated at the polling station level, because the opposition won only 57% of the vote - some of the polling stations were likely won by the EPRDF.

Update: All three EPRDF wins will be investigated.

GOJJAM

Gojjam is fairly homogenous area that is divided into two zones containing 18 weredas in East Gojjam and 16 weredas in West Gojjam. The large town of Bahir Dar is separated into its own zone.

Despite its homogenous characteristics, the Gojjam results exhibited one of the most striking divergences of any neighboring zones nationwide. East Gojjam was swept by the EPRDF with the exception of the zonal town Debre Markos and another smaller town wereda - Mota. West Gojjam by contrast, was swept by the Opposition by an average winning margin of 36%.

The relationship between adjusted turnout and winning percentage in East and West Gojjam was explored in the previous article from May 25. After that article was published, two additional results from East Gojjam were made available (Bichena and Sede), and these results strengthened the correlation (R-squared increased to 0.78 from 0.77). In fact the linear regression equation would have accurately predicted the winning EPRDF percentage simply based on the adjusted turnout.

When using winning vote percent the correlation goes up further to R-squared= 0.81. The winning vote percentage is the better variable to use because the winning margin can be affected by the presence of multiple candidates.

The reason the relationship exists is because the amount of vote tampering (if it occurred) was variable. More votes were added in some weredas than others. More legitimate votes were discarded in some weredas than others. Thus the adjusted turnout varies and the winning vote percentage becomes closely linked to how many votes were added.

Simulation
A simulation exercise was performed to model what happened on and after election day. West Gojjam was assumed to be representative of a free election. The results were then modified by (a) removing 35% of the votes cast for the opposition, and (b) randomly adding votes. The votes added were based on a percentage of the unused ballots in each wereda. These parameters were selected by trial and error to increase the turnout in West Gojjam to match that of East Gojjam and decrease the invalid ballot percentage.

A set of ten trials produced median increased turnout of 83%, and reduced invalid ballot median of 8%. The median linear regression correlation (R-squared) between adjusted turnout and winning vote percentage was 0.41 and ranged from 0.32 to 0.65. [The original, un-tampered results from West Gojjam had no correlation (R-squared=0.06) and median turnout of 74% with invalid ballots =17%].

This simulation experiment demonstrates that vote tampering (depressing real turnout followed by artificially inflating the turnout) can explain the observed turnout patterns in East Gojjam where the EPRDF swept all but two weredas.

Update: One wereda in West Gojjam (Gonj -won by CUD with 47.6% of the vote) will be investigated as well as seven EPRDF-won weredas in East Gojjam. Outlier results from Digotsion, Merto Lemariam, Rebu Gebeya, and Yeidwuha are not included.

NORTH GONDER

North Gonder zone was split with 10 weredas going to the opposition and 11 to the EPRDF. As usual, weredas won by the EPRDF averaged higher turnout (76.7% vs. 74.5%). The EPRDF turnout margin would have been even higher were it not for the two Gonder town weredas that the opposition won with turnout of 78% and 81%. Gonder town is where the opposition-won weredas exhibited the highest turnout. Turnout greater than what was recorded in Gonder town occurred only in EPRDF-won rural weredas, such as 93% in Beyeda Dilybeza where the EPRDF was unopposed, 88% in Janamora and 87% in Debark.

Update: Nine weredas are scheduled for investigation. Four of these are CUD wins outside of Gondar town. Janamora will not be investigated.

SOUTH GONDER

The opposition won only 2 of 19 weredas in South Gonder zone. Turnout in the two opposition weredas was 76.3% compared to the 82.1% average for the EPRDF weredas. The EPRDF also achieved a 9% invalid ballot percentage in weredas it won in South Gonder zone, which is very low for a rural zone.

Kemikem 1 wereda was a standout with 94% turnout and only 3% invalid ballots. The EPRDF won 86% of the vote here but it would appear improbable that the rural voters would turn out in higher percentages than in Addis Abeba, and mark their ballots with much greater accuracy.

Update: The only wereda scheduled for investigation in South Gonder zone is Farta 2 (EPRDF win).

WELLO

Wello is divided into two zones: North Wello and South Wello. The main feature governing the election results was proximity to Dessie. The opposition won 11 weredas, 10 of which were clustered around Dessie. The eleventh, Bugna, is located in the northwest corner of Wello and is a remote area aside from the religious center of Lalibela.

The highest turnout for opposition wins (87%) occurred in the largest town, Dessie, while the lowest turnout (71%) took place in Bugna. Dessie also had the lowest percentage of invalid ballots - a result that makes sense when one considers the high rate of illiteracy in the region and the correlation between literacy and urbanization.

Three weredas won by the EPRDF equaled the turnout achieved in Dessie, while another three weredas exceeded that mark. Kelela and Kutaber weredas achieved 89% and 96% turnout respectively. These same two weredas reported invalid percentages of 9% and 8% - the lowest percentages in all Wello. This co-occurrence between lowest invalid percentage and highest turnout does not seem to be explained by literacy or any other socio-demographic parameters. It is notable that the most prominent EPRDF contestant in Wello, Information Minister Bereket Simon, was only able to attract a 71% turnout in Bugna wereda.

Thus it appears likely that this co-occurrence is due to other factors. For example, large scale extra-legal addition of ballots artificially inflates turnout and simultaneously decreases the percentage of invalid ballots - precisely the effect observed in Kelela and Kutaber.

Inflated turnout could have been of special importance because in six of the weredas won by the EPRDF the winning percentage was less than 60%.

Update: Four weredas in North Wello will be investigated. Three are EPRDF wins, while one (Bugna) was won by the opposition with normal turnout and normal invalid ballot statistics. Three weredas in South Wello will be investigated. Two are EPRDF wins (Debresina 2 and Sayint 2) and one is a CUD win (Albiko). Kelela and Kutaber will not be investigated.



Ethiopian News and Views