Prepared By Ipsos Synovate, 15th March 2013 4 th March IEBC Election Results Vis a Vis Ipsos Synovate February SPEC Poll: Voter Turnout Explains (nearly all) the Difference The Polls vs. The IEBC Results It has been 6 days since the IEBC announced Uhuru Kenyatta and William Ruto of Jubilee as winners of the hotly contested presidential elections. According to the IEBC, they obtained 50.07% of the total votes cast. Yet the last opinion poll released by Ipsos Synovate on 22nd February, based upon fieldwork conducted between 15th to 19th February 2013 (just 13 days to the general election) showed a quite different result: a slim lead for the Jubileee team of 44.8% over Raila Odinga and Kalonzo Musyoka of at 44.4%. The difference with announced results of 50.07% was 5.25%. Recall also that only the Ipsos Synovate poll put Jubilee ahead of, even if this lead was within the survey's margin-of-error of +/- 1.26%. What can explain this difference? Key assumption: That results are accurate the IEBC figures with regard to both turnout and Both of the polls conducted by Ipsos Synovate before the election (mid-january; mid- a run-off would be February) utilizing the IEBC's total of registered voters indicated that necessary unless the situation changed significantly. Aside from any irregularities, three factors were identified as being able to cause such changes: 1) 'last-minute' swings in voting intentions, including (by necessity) the disappearance of the "undecided lot" (and those who declined to reveal their intentions), who (combined) comprised about 5% in January and 3% in February. Recall that the Publication of Electoral Opinion Poll Act (2012) prohibits the publication of any election-related poll within five days of an election. With a sample size of 6,000, that meant that the fieldwork for the February Ipsos Synovate survey had to be completed 2 weeks before March 4 th, a period during which both campaigns were extremely active, with the Jubilee ticket apparently riding a modest upward wave. 2) the decline of support for any or all of the six 'minor' candidates, all of whom the polls showed as having no realistic chance of winning 3) voter turnout, that could likely vary across different parts of the country In this connection, it should be recalled that in releasing its January SPEC poll results, Ipsos Synovate concluded by stating that: Together with such other unknown factors as turn-out rates in various parts of the country on Election Day, there is no solid basis for predicting the actual outcome as of now. Page 1
The Actual Outcome The outcome-results released by Ipsos Synovate for Raila Odinga and the six smaller candidates on February 222 were within the margin-of-error (+/-1.26%). Yet the 5.25% 'gain' for Kenyatta is well outside it. To try and explain this gap, Ipsos Synovate has done an analysis of the results provided by the IEBC. It concludes that: Higher voter turnout in Jubilee Strongholds gave Uhuru Kenyatta a clear advantage over his main rival, Raila Odinga. Specifically, in the 20 counties where Kenyatta won most votes, the averagee turnout was 88%, while in the 20 counties where Odinga obtained most votes, it was 84%. A modest bandwagonn effect also appeared to have been at work. Whereas the six 'minor' candidates received about 8% of the 'votes' in Ipsos Synovate's February poll, the IEBC's results gave them only 5.64% - a 'loss' of nearly 2.5%. (In fact, while Musalia Mudavadi, Peter Kenneth and Martha Karua 'lost' 2.75%, Mohamed Dida, James ole Kiyapi and Paul Muite 'gained' almost 0.5%.) Based on a regional analysis, it can be assumed that more than half of these 'lost' votes were switched to Kenyatta. Page 2
Ipsos Synovate Synovate poll results vis-a-vis IEBC results A comparison between the Ipsos Synovate poll results and the IEBC results shows that apart from Uhuru Kenyatta, the results were within the poll s margin of error (+/- 1.26%). Ipsos Synovate Error margin Polls Presidential Candidate Results for February 2013 IEBC Results Variance Higher than error Uhuru Kenyatta 44.82% 50.07% 5.25% margin Raila Odinga 44.36% 43.31% -1.05% Within error margin Musalia Mudavadi 5.18% 3.93% -1.25% Within error margin Peter Kenneth Martha Karua 1.61% 0.84% 0.59% 0.36% -1.02% -0.48% Within error margin Within error margin Mohamed Dida 0.20% 0.43% 0.23% Within error margin James Ole Kiyapi 0.05% 0.33% 0.28% Within error margin Paul Muite 0.00% 0.10% 0.10% Within error margin Rejected Votes (IEBC) 0.00% 0.88% 0.88% N/A Undecided (Polls) 2.95% 0.00% -2.95% Notes: The total positive variance amount to +6.74% The total negative variances amount to -6.75% N/A Conclusions: Support levels predicted by Ipsos Synovate for Raila Odinga and the other six 'minor' candidates were within the margin-of-error. The total positive variance amount to +6.74% and the total negative variances amount to -6.75%. A majority of those who had indicated support for any of the six minor candidates in the Ipsos Synovate polll decided to vote for one of the two major candidates; most voted for Kenyatta. Campaign activity in the final two weeks apparently benefited Jubilee. Page 3
Voter registration higher in Jubilee strongholds The voter registration in the various Strongholds shows that there voters in Jubilee Strongholds compared to Strongholds (a voters). Category Jubilee Strongholds (19) Strongholds (18) Amani Stronghold (1) Battlegrounds (5) Contested (4) Diaspora (1) Registered Voters Votes garnered by Kenyatta as per IEBC were more registered difference of 247,811 Votes Garnered by Odinga as per IEBC 5,730,281 4,545,774 456,041 5,482,470 505,309 3,611,998 202,822 2,542 77,825 2,515,703 991,832 1,016,202 418,620 127,025 177,256 2,637 951 1,224 Total Note: 14,352,533 6,173,433 5,340,546 The Stronghold analysis is as per the Ipsos Synovate classification: a vote-intention difference between the top two candidates of more than 20%. Conclusion: 79% of the registered voters in Kenyatta s 19 Strongholds voted for him and 9% did so in Odinga s 18 Strongholds. By contrast, only 66% of the registered voters in Odinga s Strongholds voted for him, with 10% of them voting for Kenyatta Page 4
Voter turnout rnout & votes cast analysis in Uhuru and Raila s Raila s top 2o counties The analysis on this section focuses on the top 20 counties for the each of the top two presidential candidates. i.e. Uhuru Kenyatta and Raila Odinga. According to EBC results, these are as follows: I Page 5
Uhuru vs. Raila:: Total Registered Voters and Total Votes Cast as a Function of Average Turnout in Their 20 Biggest Vote-Basket Vote Basket 12,000,000 Uhuru (88%) Raila (84%) 10,000,000 8,142,896 8,535,663 8,000,000 6,000,000 4,000,000 7,160,872 Uhuru s Top 20 Raila s Top 20 Uhuru s Top 20 7,162,710 Raila s Top 20 2,000,000 0 Total Registered Total Votes Cast Total voters registered in top 2o counties: Raila Odinga had a higher number of total voters oters registered in his Strongholds Strongholds (8,535,663) as compared to those registered in Uhuru s Strongholds (8,142,896). (8,142,896) Voter turnout in top 20 counties: In these counties, the voter tturnout for Uhuru Strongholds was 88% as compared to 84% in Raila s counties. This translates to 7,160,872 votes cast in Uhuru s and 7,162,710 in Raila s top 20 counties there was no significant difference in the number of votes cast in both top 20 counties. Votes won in top 20 counties: Uhuru was able to win 75% of the votes cast in his top 20 counties (i.e. 5,348,518 votes), but Raila won only 63% of the votes cast in his top 20 counties was (i.e. 4,540,495). Conclusion There was as a differential voter turnout especially especially in each of their strongholds, with Uhuru having an advantage over Raila on this. Page 6
Uhuru vs. Raila: Total Votes Won by Each Candidate in Their Own/ Their Rival s 20 Biggest Vote-Basket Vote 5,500,000 5,348,518 5,500,000 4,540,495 4,500,000 4,500,000 3,500,000 3,500,000 2,500,000 2,500,000 2,080,041 1,577,071 1,500,000 1,500,000 500,000 500,000-500,000-500,000 Uhuru Votes in His Raila Votes in His Own 'Vote-Baskets' Own 'Vote-Baskets' Uhuru Votes in Raila's 'Vote'Vote Baskets' Raila Votes in Uhuru's 'VoteBaskets' Uhuru s performance in Raila s vote basket: In the areas that were consi considered to be Raila s vote basket,, Uhuru Kenyatta garnered 2,080,041 votes. Raila s performance in Uhuru s vote basket: In n the areas that were considered to be Uhuru s ru s vote baskets, Raila garnered just 1,577,071 votes. Conclusion Uhuru was able to attract more voters in Raila s vote baskets relative to Raila s performance in his. Page 7
Ipsos Synovate predictions and outcomes at the county level Ipsos Synovate in partnership with key local media houses conducted 7 county polls whose outcome came to pass. The polls had shown that the candidates would emerge as winners on 4th March or had an advantage over the others. This is summarised below. County 1. Nairobi Predicted winner by Ipsos IEBC outcome - winner Synovate Polls Governor Dr. Evans Kidero Dr. Evans Kidero Senator Mike Mbuvi Sonko Mike Mbuvi Sonko Women Rep Rachael Shebesh Rachael el Shebesh 2. Kiambu Governor William Kabogo Senator Paul Wamatangi Women Rep Ann Nyokabi Gathecha William Kabogo Paul Wamatangi Ann Nyokabi Gathecha 3. Kakamega Governor Wycliffe Oparanya Senator Bonny Khalwale Women Rep Elsie Lihanda Wycliffe Oparanya Bonny Khalwale Rachel Amesso 4. Mombasa Governor Ali Hassan Joho Senator Hassan Omar Women Rep Missi Juma Hamisi Ali Hassan Joho Hassan Omar Missi Juma Hamisi 5. Machakos Governor Dr. Alfred Mutua Senator Johnstone Muthama Women Rep Dr. Susan Musyoka Dr. Alfred Mutua Johnstone Muthama Dr. Susan Musyoka 6. Nakuru Governor Kinuthia Mbugua Senator James Kiarie Mungai Women Rep Grace Kibuku Kinuthia Mbugua James Kiarie Mungai Mbugua Mary Njoki 7. Kisumu Governor Jackson Ranguma Senator Prof. Anyang' Nyong'o Women Rep Rose Nyamunga Jackson Ranguma Prof. Anyang' Nyong'o Rose Nyamunga In Kakamega and Nakuru, those indicating that they were uncertain on whom they would vote for as women s representative at the time if the survey was 54% and 68% respectively. Page 8
Ipsos Synovate Classification of based on the IEBC outcome 1 Mandera 2 Meru 3 Muranga 4 Nakuru 5 Nandi 6 Nyandarua 7 Nyeri 8 Tharaka 9 Uasin-Gishu 10 West-Pokot 11 Baringo 12 Bomet Elgeyo13 Marakwet 14 Embu 15 Isiolo* 16 Kericho 17 Kiambu 18 Kirinyaga 19 Laikipia 1 Bungoma 2 Busia 3 Homabay 4 Kakamega 5 Kilifi 6 Kisii 7 Kisumu 8 Kitui 9 Kwale 10 Machakos 11 Makueni 12 Migori 13 Mombasa 14 Nyamira 15 Siaya 16 Taita-Taveta* 17 Tana-River* 18 Turkana* 1 Lamu* CONTESTED 2 Wajir CONTESTED 3 Samburu* CONTESTED 4 Trans-Nzoia CONTESTED 1 Garissa 2 Kajiado 3 Marsabit* 4 Nairobi 5 Narok 1 Vihiga AMANI stronghold 1 Diaspora - For further details on this press release please contact: Dr Tom Wolf Research Analyst twolf@wananchi.com Victor Rateng Opinion Polls Project Manager victor.rateng@ipsos Ipsos Synovate Synovate.com Page 9