Kenyan pollsters eat humble pie

No one in the mainstream Kenyan media, at least not yet, wants to talk about the failure of opinion polls to predict the outcome of last Monday’s election (For some thoughts on the election check out my post on the monkey cage blog here).

A week to the election the three main polls showed the race to be neck and neck between Messrs Raila Odinga and Uhuru Kenyatta, with a slight advantage to Mr. Odinga, on average.

But in the end it was not even close. Mr. Kenyatta handily beat Mr. Odinga by almost 7 percentage points (50.07% – 43.28%).

So what went wrong?

In my view, the pollsters missed the mark both by not taking turnout into account (despite my unwarranted advice!) and perhaps poorly weighting the results from the different regions of the country (Kenyans largely vote along ethnic lines for various instrumental reasons; ethnic groups are geographically concentrated, with variance in size and population density).

The polling firms ought to have done a better job of basing their results on likely voters as opposed to self-declared registered voters. Kenyatta’s strongholds registered voters at higher rates, and based on past elsctions, were likely to register higher turnout rates than  Odinga’s strongholds- and they did (88.6% to 84%).

My own pre-election predictions a month before the election highlighted this fact. In my estimation the polls consistently overestimated Mr. Odinga’s support.

I eagerly await the polling firms’ own rationalization of what happened. Hopefully the misses this time won’t permanently damage the public’s perception of opinion polls.

Despite the difficulties in forecasting political outcomes, it is better to have polls than fly blind into an election.

Will turnout disappoint Odinga tomorrow?

Travel, conference and more travel have kept me from blogging in the last few days. I am back online. Kenyans go to the polls tomorrow. This is a post from a few days ago.

I am in New Haven (great to be back!) for a conference and visiting with friends before going back to Nairobi for Monday’s election and wanted to make a quick post on the latest in the upcoming Kenyan elections.

1. The Second Presidential Debate:

The second presidential debate was rather dull, to be honest.

The entire first half was supposed to be dedicated on the economy but mostly dwelt on the specific subject of corruption at the expense of other more pressing concerns like jobs and the modernization and formalization of the Kenyan economy (Not to trivialize Kenya’s obscene levels of corruption, but apparently the moderators do not read Blattman’s blog. The problem is huge, but there were other pressing economic issues that could have been addressed.)

The candidates were, as expected evasive over the matter – mostly giving vague answers to pointed questions on scandals they were rumored to have been involved in; from Goldenberg to Anglo Leasing to Maize to typos at the treasury.

The jester Candidate Dida provided a light moment when he asked the moderator if she expected thieves to admit that they had stolen.

The second half was better. It dealt with the issue of land. Again, no real answers emerged but it put Mr. Kenyatta – who’s family owns large tracts of land – on the hot seat. Land is an issue that has been the cause of ethnic clashes in every Kenyan election since 1992. Voters may just have got one more thing to think about before casting their votes on Monday.

Overall, I cannot say that any one candidate won the debate. However, Mr. Kenyatta lost a few points due to the amount of time spent on the land question. His main opponent Mr. Odinga even came to his rescue at one point, reminding viewers to cut Mr. Kenyatta some slack since he only inherited land that may have been dubiously appropriated by his father while he was president.

2. Election Day’s Big Unknown: TURNOUT

As I have pointed out before, the outcome of the presidential election in Kenya next Monday will hinge on turnout. Whether we shall wake up on March 5th with a president-elect or have to go for a runoff will depend on regional turnout rates.

As it is most public opinion polls point to a runoff.

However, the polls do not give us a sense of what proportion of registered voters are likely to vote.

One private poll that I have seen suggests that 98.3% of those polled said they would show up to vote. Such turnout numbers belong in Belarus. I expect average turnout to be between 75-85%.

Below is a table with the turnout rates in the last three presidential elections. (The 2007 figures should taken with a pinch of salt, for obvious reasons). As you can see history seems to be on Mr. Kenyatta’s side. Central Kenya and the Rift Valley, the two most populous regions of the country both support him overwhelmingly (according to the latest Ipsos Synovate poll) and have the highest average historical turnout rates.

Region/Turnout 1997 2002 2007 Kenyatta Support Odinga Support
Central 74.1 66.1 82.1 88.1 6.4
Rift Valley 75.9 60.8 72.8 69 23.3
Eastern 72.6 60.9 65.9 41.8 52.7
Nyanza 67.2 55.6 76.2 9.8 83.8
Western 68.1 57.1 62 3.4 53.6
North Eastern 55.9 57.8 61.3 37.9 44.1
Coast 50.6 42.1 57 18.3 73
Nairobi 50.2 42 51.5 39.6 51.9

Could turnout rates be different this time?

The answer is maybe, due to the following new variables:

  • New positions created in the constitution – governor, senator, and women rep – might attract new voters in addition to those who have in the past voted for the three other posts – president, MP and councillor (now county rep). This might increase turnout across the board, but since Mr. Kenyatta is already “maxing” his turnout rates Odinga might benefit from an overall increase in turnout rates.
  • The dropping of calls for “six piece” vote. Earlier in the cycle both candidates had insisted that voters should vote one straight ticket for the party for all six posts. However, since the nominations for both leading parties/coalitions were a total mess many popular candidates did not get nominated on the “right parties.” Insisting on a six piece vote would have lowered turnout. However, without it everyone has a candidate to vote for in their preferred presidential candidate’s stronghold. This may increase turnout since it gives incentive for say someone running in an Odinga stronghold, who supports Odinga, but is not in Odinga’s party to still mobilize his supporters to the polls in competition with the candidate in Odinga’s party. Again, a higher average turnout rate will benefit Mr. Odinga.

3. The Numbers:

The last polls before the election (about five of them) show a slim national lead for Mr. Odinga over Mr. Kenyatta, 45% to 43% on average. But since we do not know the make up of likely voters I would be reluctant to declare Mr. Odinga a favorite going into Monday. Mr. Kenyatta’s strongholds have historically had better turnout rates than Mr. Odinga’s.

And on the matter of polls, the following factors may lead to surprises on Monday night:

  • Desirability biases in the survey: We do not know to what extent those interviewed lied about who they support. And on this count Mr. Kenyatta runs the highest risk. Because of the ICC cases he and his running mate face, many higher income and educated Kenyans have had concerns about their ability to run the country “via Skype.” If such people residing in the Rift Valley and Central regions lied to pollsters because they were expected to like Kenyatta then we may be in for a surprise on Monday night. If Mr. Odinga gets more than 35% of the votes in Rift Valley and about 15% in Central Kenya he will win the presidential election in the first round. 
  • Strategic voting: One of Mr. Odinga’s many challenges has been the Western Kenya region where Musalia Mudavadi comes from. Mr. Mudavadi is the third candidate, getting 6% on most polls nationally and about 34% in Western region. Mr. Odinga gets 53% in the region and Kenyatta 3%. In the event of a runoff, most of Mudavadi’s voters will break for Odinga. Mr. Mudavadi’s candidature centres on becoming a King maker and elevating his stature as a serious contender in the next elections (2017). But if Western voters who would support Odinga in the second round decide to do so in the first round Odinga will have an outright victory in the first round.

Barring any surprises on Monday (esp with regard to turnout) the state of play is that the first round will be a close contest between Kenyatta and Odinga, with one or two percentage points separating them.

Mr. Kenyatta is a slight favorite to win the first round, but will be short of the 50% required for an outright win. In the second round, however, I expect Mr. Odinga to be a clear frontrunner since most of Mudavadi’s voters in Western Kenya will break for him. I also expect a little bit of anti-Ruto (Kenyatta’s running mate) votes in the Rift Valley going to Mr. Odinga. I am very curious as to what President Moi’s promised major announcement on March 5th will be…. perhaps an endorsement?

All in all I would put my money on Mr. Odinga becoming Kenya’s fourth president.

Latest polls, Kenya Decides 2013

Ipsos Synovate just released another poll on the upcoming presidential election. The poll shows Prime Minister Raila Odinga leading his deputy Uhuru Kenyatta by 46% to 40% in approval rating. 10% of of those polled were undecided. More on the latest polls this weekend after I get a regional breakdown of the data.

For the historical trend in the polls, including today’s, see below.


On a related note, the Guardian Africa Network has a piece on the threat of violence leading up to and after the March 4th poll. Some call such warnings fear-mongering, but I’d rather have everyone freaking out about violence and thus sort of prepared than a repeat of 2007 in which everyone – including Kenya’s intelligence services – was caught flatfooted.

The Presidential Race in Kenya’s 4th of March 2013 Election

The race to succeed President Kibaki promises to be an interesting one. All the pointers indicate that it will be a close race between the Raila-Kalonzo-Wetangula and co. faction vs. the Uhuru-Ruto-Mudavadi and co. faction. Prime Minister Raila is expected to be at the top of the ticket under an umbrella special purpose vehicle called CORD (Coalition for Reforms and Democracy). Deputy Prime Minister Uhuru Kenyatta leads his own SPV called the Jubilee Alliance. Below I give a little background information before delving into the state of the race.

Institutions Matter:

The 2011 Kenyan constitution mandates that the winning presidential candidate garner 50% + 1 of the votes cast and at least 25% of the votes in at least half of Kenya’s 47 counties. In addition, presidential candidates and their veep candidates must run on a joint ticket. Previously, the presidential candidate could promise the veep slot to any number of ethnic chiefs. The constitution also limits the president’s ability to buy support by limiting the number of cabinet slots to 22 (necessitating the creation of minimum winning coalitions).

This situation has forced Kenya’s politicians to form alliances that cross ethnic lines, a change from the past when nearly all the major ethnic groups produced their own presidential candidates. The logic of minimum winning coalitions has set in, with two main camps forming ahead of the polls – Mr. Odinga has insisted all year on the stump that this is a two horse race between him and a straw man non-reformer, and that any other candidates are mere donkeys.

In this cycle the big five (Kikuyu, Luhya, Luo, Kamba and Kalenjin, together making over 70%) are all in either CORD or the Jubilee Alliance. Continuing the Luo-Kikuyu feud that has characterized Kenya’s political history since 1966, this election will pit the son of Kenya’s first president (Kenyatta) vs. that of the first vice president (Odinga). CORD’s formateur is Odinga (a Luo) who leads the Orange Democratic Movement (ODM). Jubilee’s is Kenyatta (a Kikuyu) who is leader of The National Alliance (TNA).

It is likely that CORD will field Odinga as the presidential candidate and Vice President Kalonzo Musyoka (a Kamba) as his running mate. Jubilee is likely to field Kenyatta as the the presidential candidate and William Ruto (a Kalenjin) as his running mate. This would leave the Luhya as a big five swing group. Such a scenario favors Odinga, who is already widely popular in Western Province (Luhya-land) and sections of Rift Valley Province (Kalenjin-land). This scenario is likely, but not set in stone.

The ICC Question:

Mr. Kenyatta and Mr. Ruto are facing charges at the ICC over the post election violence that rocked Kenya in 2007-08. Back then they were in opposing parties – Mr. Ruto with Odinga in ODM and Mr. Kenyatta with Kibaki in PNU. They are both suspected to have funded gangs of rival ethnic groups (Kalenjin and Kikuyu) that committed heinous crimes including murder, rape, and arson. The international community has sent a strong signal – through Kofi Annan’s statements, threats of sanctions and the EU’s travel ban on the duo – that the two should not run for office.

The Kikuyu business elite (including cash crop farmers and horticulturalists who would be hardest hit by international sanctions) have thus been trying to prevail on Mr. Kenyatta to forgo his presidential run in favor of Mr. Musalia Mudavadi (a Luhya) – evidence suggests that this was the carrot that Uhuru and Ruto used to lure Mr. Mudavadi into their coalition. If Uhuru steps down for Mudavadi then a good chunk of the Luhya vote would depart CORD for Jubilee. But it may create room for the lesser presidential candidates from Kenyatta’s region – Martha Karua and Peter Kenneth – to get a sizable chunk of the Kikuyu vote. Plus it is unclear if Mudavadi can weather the accusations of being Kenyatta’s project on top of campaigning against the formidable Mr. Odinga. Kenyatta’s last presidential bid faltered partly because he was seen as Moi’s project.

The State of the Race:

Ipsos Synovate, a local polling company, just released a poll of adults 18+ on their preferred candidates for president. Mr. Odinga leads the pack with 34% of the respondents saying he is their preferred candidate. Mr. Kenyatta is second with 27%. Mudavadi, the other likely Jubilee candidate polls at 5%. Musyoka and Ruto poll at 3% and 2% respectively. It is important to note that it is only Mr. Odinga and Mr. Kenyatta who are presently outperforming their ethnic group size in the polls (by 21% and 2% respectively). 22% of Kenyans remain undecided. Notice that the number of undecideds is highest in provinces that lack a presidential front-runner, i.e. all except Central (Kenyatta) and Nyanza (Odinga).

click on image to enlarge.

opinion polls

If Odinga eventually faces off with Kenyatta the key swing region that will determine the outcome of the election will be the Rift Valley Province. Mr. Odinga will have Nyanza, North Eastern, Western, Coast and half of Nairobi in the bag. Mr. Kenyatta will have Central, (possibly) the Rift Valley and half of Nairobi locked in. In this scenario (let’s call it scenario 1), for Mr. Kenyatta to win he would have to run the numbers in both the Mt. Kenya region (which as a whole has about 24% of voters, according to the 2009 census) and the Rift Valley Province (with 25% of voters) and get a good showing in Nairobi.  For Mr. Odinga to stop him he would need to have a respectable showing in the Rift Valley – something that he can given the fact that he has managed to keep key leaders from the region in his party, ODM. In Scenario 1 Odinga will be the favorite to win.

If Jubilee nominates Mudavadi to face Odinga, then things will get interesting (It would also potentially make for a de-ethnicized presidenital race). Key questions will be:

  1. Whether the Kikuyu would vote for Mudavadi, given that the ticket would not have a Kikuyu (with Ruto as running mate). Would they opt for other Kikuyu presidential aspirants in Martha Karua and Peter Kenneth?
  2. Whether Mudavadi would manage to unite the Luhya bloc behind him. Luhya leaders have traditionally had a hard time uniting the region to vote as a bloc. Can Mudavadi overcome the sub-ethnic divisions of the Luhya?
  3. Whether Mudavadi will be able to effectively fight the inevitable portrayal of his candidacy by the opposition as Uhuru Kenyatta’s (or Kibaki’s) project. Can Mudavadi be his own man?

If the Rift Valley, Mt. Kenya (Central and sections of Eastern) and Western vote go to Mudavadi en masse, it is hard to see how Odinga can make it to State House. In this scenario, turnout would be key. Jubilee would win by a landslide. But while this situation is likely, it’ll still be a huge gamble for Jubilee to nominate Mudavadi.

Mr. Odinga has more national appeal than his former ODM assistant Mudavadi. The latter lacks a strong political base in his home region of Western Province. Add to that the fact that he will leave the gates with the imprint of “project” on his forehead, not to mention the uncertainty over how Mt. Kenya region would vote and the election becomes a real tossup ex ante. Nominating Mudavadi to head the Jubilee ticket would be a high risk gamble for Uhuru and Ruto that would either pay off big come March 4th or hand Odinga victory on a silver platter.

The Jubilee Alliance will nominate its candidate (either Uhuru or Mudavadi) next week. CORD will name its presidential candidate (very likely to be Mr. Odinga) on the 22nd. I expect minor defections and realignments that will have a non-trivial impact on the race before then. All in all right now Odinga is a slight favorite to become the fourth president of Kenya.


Kenyan Elections 2013 and The High Potential for Violence

In an excellent piece over at African Arguments Sheekh and Mosley give a comprehensive discussion of the recent outbreak of violence in the Tana Delta region of Coast Province, Kenya. According to the authors:

Long-standing competition and conflict over access to pasture and water resources were important factors, but did not alone provide the trigger for violence.  A range of political and economic factors have fed into the local dynamics in Tana Delta. These include longer-term trends related to alienation of local people from land due to large-scale government and private sector purchases, and shorter-term impacts related to the process of delineating electoral constituency boundaries and county districts in line with Kenya’s new constitution. The ready availability of small arms has also seen such conflicts intensify in recent decades. Lack of livelihood opportunities for the youth is also a major factor.

As such, the recent clashes are emblematic of wider trends.  Although the Tana Delta (along with the rest of Coast Province) has tended to be politically marginalised, tensions in other areas – such as Mt Elgon and parts of the Rift Valley including Eldoret, Nakaru and Naivasha, and counties in northern Kenya – could also be exacerbated by the same political factors.  Some of these areas were flash-points in the post-poll violence of late 2007 and early 2008, with major national and regional ramifications.”

Spot on.

The Tana Delta conflict is symptomatic of a larger dynamic that will play out in anticipation of the March 4th 2013 elections in Kenya. The new constitution has created 47 county governments, many of them multiethnic or otherwise diverse, that will each have three county-wide elected officials (a governor, deputy governor, and a senator). Ethnic and communal rivalries will inevitably surface in these county contests, with potentially disastrous outcomes such as what we’ve seen so far in Tana River County.

The potential for decentralized violence in Kenya’s 47 counties is a real cause for concern.

In order to limit the potential for violence, the national commission charged with policing ethnic harmony has initiated talks in potential flash-points to broker inter-ethnic power-sharing deals with the hope of avoiding a situation in which certain communities are totally excluded from county-wide elected offices. Sadly, so far there is no sign that these initiatives will work (Not to mention how un-democratic such back room arrangements will be). Plus the violence will not necessarily be exclusively of the inter-ethnic variety (which is what the commission is fixated on at the moment). Even ethnically homogenous counties might experience inter-clan violence.

While most of the attention in the next few months will be on how to avoid a repeat of the aftermath 2007-08 election, Kenya watchers should be warned that the problem will be much more complicated. If nothing is done, many counties will experience inter-communal violence. The new county governments will have real resources (about a third of national revenue) that will generate real patronage networks worth fighting for.

Given the nature of Kenyan politics, the race for the presidency (more blog posts on this soon) will inevitably hog all the attention in the next five months. I hope the contest for State House and its own risks for violence will not overshadow the county-level contests which will also be just as intense and likely to result in violence.