Kenya Today

Kenya News, analysis & in-depth reports




Shock: Ipsos Synovate Admits Poll Sample Was Skewed Towards Uhuru’s Stonghold

128 Comments

By: Kenya Today News Team

[wpsocialite]

Ipsos Synovate Managing Director Maggie Ireri has admitted that the sample used in the controversial poll on presidential debate was skewed towards Central and Rift Valley provinces. The two provinces are the bedrock of Mutahi Ngunyi’s tyranny of numbers. The damning revelation has left many in shock and now questioning the purpose of such biased poll.

Tom Wolf the firm’s lead research analyst confirmed that the findings of the survey were not nationally representative.

“This is not a national poll because it is limited to those who watched or listened to the debate and those whose phone contacts are in our data base,” said Wolf.

In the sample, Rift valley had 24 per cent of the respondents, Central 17 and Nairobi 13. Central is perceived as Uhuru’s strong hold while Rift Valley is perceived as under the grip of his running mate.

Raila’s Nyanza backyard had only 14 per cent of the sample being interviewed same as Eastern, which is perceived to be partially under the control of his running mate, Vice President Kalonzo Musyoka.

“In fact more than half (54%=24+17+13) of the respondents were from Central, Rift Valley and Nairobi respectively” Frankline Onchiri Magaki a research fellow at Washington University said

“Very highly skewed sampling like this, besides other inherent selection biases make this polling worthy of a gutter press.  Anyone who has attended basic research methods class can at glance thrush all its findings. What do you make of a study where authors spend half the time explaining the limitations/biases?” Wondered Magaki

The question Maggie Ireri could not answer however is why she released a biased poll results knowingly?

Many Kenyans were shocked at the findings that even Uhuru Kenyatta’s supporters could not believe it.

“I watched the debate, Uhuru did a good job but I did not expect him to win, I ranked him 2nd or 3rd, now the Synovate guys have just given as a bonus, hii ni mungu bwana!” Mosese Mwaura a Limuru based accountant said

The first presidential debate was held on Monday 11th Feb 2013; Raila Odinga was touted by many as the winner of the debate closely followed by Martha Karua. However released polls placed him second prompting many to question the motive of polling firms that knowingly released biased or “cooked” poll results.

Comments

  1. This is how they caused mayhem in Kenya Kenya in 2007. Because was leading with 55% in the polls and then all over a sudden they started saying that Raila was loosing popularity is this a repeat of the same stratetgy? Kenyan beware of this fake organisations. How can release a report that you know very well that is schewed in favour of a candidate? Shame on them.

    Reply Report comment

    • Polls are always realistic when carried out scientifically.The mayhem in 2007 was caused by cooking election results not the polls. There is no way Kibaki was neck to neck with Raila.President KIbaki was overwhelmingly voted by all tribes of Kenya in 2002 and if there could have been a president who would have shunned tribalism and favouritism MUST have been him.However, when he took power all the important ministries and parastatals were filled by Mt. Kenya Region at the expense of the entire country. This infuriated the electorate and that is why we rebelled against him in 2007 otherwise he could have won reelection with ease if he embraced regional balance

      Reply Report comment

  19. That shows how incompetent you are. It doesn’t matter how many people watched or listened to the debate at what place. You should have balanced then. No matter what the polls say, we will not be swayed. I will never trust your polls and those of Infotrak whether you say Uhuru is leading or not.

    Reply Report comment

  22. Hahahaha…. Just gotta love Kenya. So what happened overnight? Hahahahahaha……. I keep saying, all opinion polls should be banned – especially in Kenya! They contributed in a very big way to PEV last elections, and they are well on their way to instigating more animosity. Anyone who listened to the debate made up their own minds….polls notwithstanding. So Shut the &*#% up!

    Reply Report comment

  32. Raila should take that as a political challenge and work extra hard during the second debate which very crucial as its just few days to the elections.
    Kenyan electorate are easy swayed by such events.
    RAO dont slip work, hard on public address to ensure u catch each and every person eye…
    suxes and happy valentine

    Reply Report comment

  35. The poll was scientific, and bloggers should stop skewing the mind of Kenyans. Maggie did not comment as this blogger purports, they haven’t even quoted her. secondly the caveats on the poll were evident from the beginning and were clearly stated; the poll was limited to those who watched the debate, those who had mobile phones (since it was done through CATI) and also since the debate was held in English there may be a large section of those who watched who did not understand. Is terms of distribution the report is on Ipsos Synovate’s site and this blogger should stop lying the difference between Central and Nyanza was 1% and this is because sampling is done according to IEBC voter registration on National Census, so you don’t expect each area to have equal number of people sampled, that is sample science. a region with 3.5 million people cannot have a sample similar to a region with 1 million people. The pollsters are not the problem, rather it’s our lack of understanding of the poll science… and this bloggers are obviously not making it any easier by posting lies and non verified information… http://www.ipsos.co.ke/home/index.php/downloads link to the full poll.

    Reply Report comment

  37. Oops! before you pass judgement on whether the sampling was skewed or not please check that your facts are correct. I have checked that the sample shares in this poll corresponds to the registered voters in each region. so the sample is not skewed. Do the math out of 14,337,399 registered voters nationally, these are the shares by region Nyanza: 1,954,756 (14%), R. Valley 3,373,853 (24%), Central: 2,190,477 (15%), Eastern: 2,092,883 (15%), Nairobi: 1,778,903 (12%), Western: 1,434,987 (10%), Coast: 1,164,083 (8%), North Eastern 347,457 (2%). This shows that the Ipsos sample approximates the regional weights 'Rift valley had 24 per cent of the respondents, Central 17 and Nairobi 13 and Nyanza had only 14 per cent. so your your allegation of skewed polling is factually wrong.

    The caveat the pollster gave was that the sample was for only those who watched the debate and since not every Kenyan has a T. V or a Radio. therefore the sample is not fully representative as it left out those who did not participate. therefore if you say it is skewed in favor of Uhuru can only be true if those who listened or watched are Uhuru's supporters and those who did not watch or listened (therefore excluded from the sample) were Raila Supporters.

    Reply Report comment

    • Agreed Ben but the fact that the survey depended hugely on ownership of TVs and radios is reason enough not to trust the results…me and you know that socioeconomic profiles differ across regions, the basis of your justification for the poll

      Reply Report comment

    • @ Ben to respond to you the report is available on the synovate site… http://www.ipsos.co.ke/home/index.php/downloads… you will notice on slide 34 that the poll was conducted among debate watchers 37% of whom were already supporting Uhuru as opposed to 35% for Raila. In addition Tom wolf actually cautioned that the poll wasn't scientific as it was only conducted among those who watched the debate and should therefore not be used to make a conclusion on the candidate who is leading nationally.

      Reply Report comment

    • @Mike Well, you make the observation that 35% already supported Jubilee candidate and 35% Cord candidate, from what i gather from observation is that Jubilee candidate gained from this debate while cord candidate lost some points, which is consistent with the overall conclusion from the poll. nevertheless, if this sample was randomly drawn and 35% was found to support candidate1 and 37% supports candidate2, as long as its randomly drawn the sample cannot be said to be skewed in favor of candidate2 since it was not purposively made to be as such. and thus we can conclude that the debate swung perception in favor of candidate2.

      Your second observation on the fact that the poll was conducted on those who watched or listened to the debate, and therefore it can not be used to make conclusions on entire population. It is not entirely true to say that nor is it correct to simply discard the poll as unable to estimate the national opinion. and therefore the caution given by Ipsos. my reasons are given in the my first post, in addition i seek to aver the following:

      If you have believe that that –debate audience(both watching & listening) and phone ownership-was systematically skewed in favor of candidateX a and against candidateY then in this case you would be correct not to generalize on the population but in as long as this systematic bias is absent i do not see why you cannot generalize on the population.

      However am open discussion please prove how the –debate audience(both watching & listening) and phone ownership- is biased against/infavor of any particular candidate. i will take the poll with a pinch of salt

      Reply Report comment

      • Ben, what are you trying to insinuate? You’ve got only one vote. Propaganda & tribal arithmetic wouldn’t help you. We are all Kenyans, there must be a winner and a looser. Stop propagating such and be responsible Kenyan. Remember, statistics also has its limitations.

        Reply Report comment

    • I get your point Ben; to be sincere, these results should worry other presidential aspirants. I say this because I expect those who watch TV to be the elite who can are hypothesized to be least tribal and knowledgeable to understand the consequences of electing a multiple and grave crimes suspect.

      Reply Report comment

    • The fact that the poll was conducted among those who watched the debate DOES NOT make it unscientific. Someone need to educate Tom Wolf on data sampling. What makes data scientific is the way the sample is weighed. It is silly to argue that the poll is not scientific because it only reflected those who washed the debate. I have done polling for a long time with organization that did polling for CBS, ABC, Washington Post and many other US networks. What is needed is comparing aple for aples and oranges for oranges. That took place in this data. The issue here is that some people are so hell bent of opposing Uhuru and they can not accept the fact that Uhuru has plety much won the race. The fact is given what happened in 2007, many people saw and understood the game that Raila played and they made their minds a long time ago. The establishment mat want Raila but people have made up their minds. It happened here in US when the democratic establishment wanted Hilary Clinton but the more the establishment rooted for her, the more the average person (mwananchi) pooled for Obama.

      Reply Report comment

    • Milu Muyanga I agree it is true that candidates should be concerned but again that's democracy. I just hope that the info coming out helps to expand the spectrum of possible outcomes from this election and therefore examine each-ones' expectations accordingly

      Reply Report comment

  38. Oops! before you pass judgement on whether the sampling was skewed or not please check that your facts are correct. I have checked that the sample shares in this poll corresponds to the registered voters in each region. so the sample is not skewed. Do the math out of 14,337,399 registered voters nationally, these are the shares by region Nyanza: 1,954,756 (14%), R. Valley 3,373,853 (24%), Central: 2,190,477 (15%), Eastern: 2,092,883 (15%), Nairobi: 1,778,903 (12%), Western: 1,434,987 (10%), Coast: 1,164,083 (8%), North Eastern 347,457 (2%). This shows that the Ipsos sample approximates the regional weights ‘Rift valley had 24 per cent of the respondents, Central 17 and Nairobi 13 and Nyanza had only 14 per cent. so your your allegation of skewed polling is factually wrong.

    The caveat the pollster gave was that the sample was for only those who watched the debate and since not every Kenyan has a T. V or a Radio. therefore the sample is not fully representative as it left out those who did not participate. therefore if you say it is skewed in favor of Uhuru can only be true if those who listened or watched are Uhuru’s supporters and those who did not watch or listened (therefore excluded from the sample) were Raila Supporters.

    Reply Report comment

  42. Let me Help clarify….there is a link to the report on the synovate site http://www.ipsos.co.ke/home/index.php/downloads…firstly Tom wolf cautioned that the poll wasn't scientific as it was only conducted among those who watched the debate and should therefore not be used to make a conclusion on the candidate who is leading nationally. Secondly if you download the report itself on slide 34 you will notice that among those being polled 36.9% were already supporting Uhuru before the debate and that only 35.3% were supporting Raila….According to the poll Uhuru got an increment of 3% points and Raila lost 2% points among those polled…its just the way the information is getting sent out that is causing confusion…Kenyans…read for yourself.

    Reply Report comment

    • William…. Makora…. I see a big tribalist in one William Makora ; I had different thoughts..wrong again; …. disappointed ; and now i go back to my in laws house; and my vote for that house of Mumbi at least they have helped us in the past..walinipatia bwana…… disappointed…..

      Reply Report comment

  50. There is no need to speculate. We all have a candidate whom we support with a reason. If you reason is selfish, evil and backed by tribal bigotry, then be informed; God will not allow evil to triumph over good. We place the destiny of this nation in the hands of Almighty God the Creator of the universe.

    Reply Report comment

  55. To me it seems the sampling was proportionate. If you look at the percentage of registered voters and percentage taken from each region then it was proportionate and not skewed at all unless some one whats to change the meaning of skewed. This was a standard sampling. where standard error limits apply. take a sample and test the degree of confidence.

    Reply Report comment

  56. Have read this article and this guy is an intellectually irresponsible, he gives percent like they were supposed to be he need to learn about sampling before amplifying his ignorance. And in your judgment I quote "The first presidential debate was held on Monday 11th Feb 2013; Raila Odinga was touted by many as the winner of the debate closely followed by Martha Karua" and which criteria did you use to come to such a conclusion. This shows shoddy work of your research and in your article. Learn to do research on articles you publish. Am not pro Uhuru and Raila.

    Reply Report comment

  64. I thought kenya was above such and i think the presidential debate was like an interview for our next president and for someone to destroy evry persons dreams because of opinion polls! Learn from mistakes kenyans let the best win and we ought to back him/her.

    Reply Report comment

  65. These are the people/organisations Kimaiyo should be sanctioning not gagging debate on genuine issues! This is very suspicious and mischivious behaviour. Inspector General Kimaiyo can you do the job Kenyan taxpayers are paying you for!

    Reply Report comment

  67. Fellow Kenyans, Statistics or not, We should all preach Peace! It is the only thing that matters for our future and the future of our children. Without it none of the presidential candidates can deliver the good they are aspiring to bring to this Great Nation, KENYA.. With PEACE though, all aspirants have good intentions for our country. Let’s poll for peace, tolerance and nationhood. Kenya is GREATER than all of us and I will think, act, speak and DO everything to vote for KENYA!

    Reply Report comment

  68. The pollsters in Kenya are very unreliable going by his confession. This is why Kenyans must not reliable on their grossly biased predictions and instead waiting for the actual election outcome.

    Reply Report comment

  70. Responsibility? Guilt? Shame? Conscience? or Truth? You should come out publicly and simply earn back the trust of Kenyans Wolf and Team….all that is needed is Peace and proper structural functions…help build PEACE.

    Reply Report comment

Leave a Reply

Your email address will not be published.

Disclaimer

Contact

Privacy and cookies