Thursday, November 15, 2007

Kenyan Opinion Polling 101

As Kenya ramps up for the national election in December, polling companies have released a slew of survey results about parties, candidates and nominees. That's got me thinking about the roll of opinion polling in established and emerging democracies. Last week I got a chance to sit down and talk with a pollster and political analyst who has a lot of experience in Kenya.


Tom Wolf came to Kenya as a Peace Corps Volunteer in 1967 and has since spend 30 years in the East Africa. He did research  in comparative politics in Coast Province, was a lecturer at Nairobi University and was Democracy/Governance Advisor for USAID/Kenya. As an independent consultant, he has been designing, conducting and analyzing survery research for Steadman Group, the Kenya affiliate of Gallup International.

You can read the entire transcript of the conversation here, complete with examples, mathematical problems and two bonus questions about early political polling in the U.S.

Q: What is the utility of opinion polling before an election, particularly for the Kenyan electorate? 

[long pause] 

You have a lot of room on your tape recorder or whatever, right? 

I�m hoping we�ll use all of it.

One of the reasons I�m hesitating, it�s such an important question. There are so many caveats one would have to make�   

What�s the utility of polling to public, as you posed it? That depends how the media� how good a job they do in reporting, what the political response is.   

Tryptich_for_webBut I think we can say, in general, we know now that more than half of adult Kenyans are aware of the polls, because we track that: a higher percentage in urban than in rural areas.   

They may not know whether they are fantasy, fabricated or factual but there�s been a huge increase in the number of polls that have been conducted in the last couple of years and, I believe, also in their credibility. 

I think [public opinion polling] is empowering to the public. I get congratulated by people wherever I go: cashiers in stores, security guards in buildings, taxi drivers.

Everybody loves them. Even, I might say, when people are not happy with the results because they show their preferred candidate, party or whatever [was not in the lead]. 

Kenyans definitely appreciate the fact that this is giving a kind of a voice to the public, more often than once every five years at an election. And that is almost intoxicatingly empowering to ordinary people. 

The media love it because it fills up space in the papers. [Poll results] make good headlines.   

Whatever complaints I might have, and others in the business have, about capacity problems in the media and deliberate political distortion and sometimes just exclusion� 

Q: Exclusion? Really? Complete exclusion of poll results? 

Putting them in funny places in the paper. Taking parts of our poll presentation which they would view as politically unfavorable and just not reporting some. 

Q: What I�ve been wondering is� there must be research into how much early polling effects election outcomes. Does it sway voters? 

As far as I know, we have no research in sub-Saharan Africa on this topic. 

Q: What about elsewhere in the world? 

A book on public opinion in America � I�m forgetting the author�s name right now � written more than a decade ago, the conclusion was: there�s no proof to show it makes any difference. 

In the [European] continental political science literature there is, I think, a fairly strong basis for findings that there is more of an impact in systems with proportional representation. Unlike the Kenyan, British, American systems of winner-take-all-votes-for-losers-are-wasted, minor shifts in the percentage distribution will make an actual difference in parliamentary representation. 

Proportional systems are more likely to have three or more viable parties with a face in parliament, unlike the U.S. system. Voters who can see that their party is not doing well enough in the polls to be, say, part of a coalition in the government, can switch to a party that is a little more viable and know that that vote won�t be lost. And also [they] can find something on the ideological spectrum that�s just a matter of degrees away from where their preferred party would be, rather than the black-or-white contrast in the American system, and to an extent the British system. Germany last year� with Angela Merkel. The polls, one month before the election, were showing her winning by around ten percent. Polls were coming out by multiple firms every week. She almost lost! She won by less than one half of one percent. So those polls certainly didn�t create any bandwagon effect.   

Q: Does polling have the potential to stoke the fire of conflict over difference, or to dampen it? Depending on how they are presented and how people are educated about them? 

You know, those are assumptions that would have to be tested. 

Q: Spoken like a true sociologist!! 

Well, my mother was a sociologist. I�m a political scientist� 

The one example I always like to give, I got from a gentleman who works for Radio France International.

He told me about the wonderfully sobering example of Congo Brazzaville, where in 1995 there was an election coming. President Lissouba had a secret poll done, which showed he was going to lose. He brought in the military to overthrow his own government, and that led to a three year civil war, in which hundreds of thousands of people died.   

Now if he hadn�t had that poll done and they�d had the elections, would he have just nullified the results and would there have been even more civil war? Would he have been so shocked by the results that he would have peacefully walked out of State House? I don�t know. 

It would be very na� to think that you could take such an integral part of public discourse and the democratic process [such as opinion polling], from a society where the institutions upon which such a democracy is based have had a long and often torturous history of evolution but are basically grounded in the framework of that society, and parachute them into a completely different context without having a lot of things go wrong. Like multi-party democracy for that matter. 

I�ll tell you one other thing, though. You can talk about the pros and cons of survey research or polling but, in a way, the starting point is, how methodologically sound are these surveys to begin with? 

Barometer_for_webQ: Clearly you have confidence in the surveys you write? 

I would not be associated with the Steadman Group if I had one shred of evidence that there was any rigging, manipulation, catering to clients or anything like that. I wouldn�t. 

Q: The competition on the other hand�? 

I have had some concerns with them at times. Beginning with what I consider an insufficient degree of transparency with regard to the sampling frames that they use. Even more so, the sloppiness with which some of them report their very own findings. 

There are many aspects of the quality of these polls that have to be seriously considered before we can talk about whether they are good or bad for Kenya. 

Q: With all the complexities and potential pitfalls, you are still choosing to do this particular work in Kenya right now. I�m assuming you wouldn�t do this work if you didn�t think it had a net benefit for Kenyans. What tips the balance for you?   

There has been a very top-down political culture in this country going back to colonialism. The kind of concentration of power, abuse of power, autocratic control, intolerance of dissenters that we saw in the first 30 years of Kenyan independence: it�s not a surprise to a student of comparative politics. If I thought that the contribution that this kind of works makes to opening up Kenyan society - particularly the lower levels of it through the voice of the people - if I thought that would lead to national disintegration, I think I�d be able to restrain myself and do something else.   

But I really do think that it is useful to those in government. It is useful to those who would like to be in government. It is useful to prospective voters and non-voters, to be able to express their views on public issues more often than once every five years.   

I don�t think that leaders should be slaves to opinion polls. A great leader sometimes has to make very unpopular decisions. I also know some leaders in many countries would do anything to wish these polls away. But what�s the alternative: for somebody to decide when you can do a poll and when you can�t? The alternative is so draconian...

Q: It sounds like there is also a real potential for polling to have a very constructive roll in development Cover_for_webof a multiparty system here. Do you think that holds true? 

Potentially, yes. But it could have another effect also.   

In most places in Africa, where parties are perhaps only marginally distinct in terms of their policy face, and where the control of the state is so important, they could have a rather negative impact.   

If one candidate or party is far more popular than the others, and this became known to the public through polls, they might just say, �If you can�t just beat them, join them. We might as well just have a one-party system.� Or, �If so many people think this is such a good President, let�s get rid of term limits and let him be a President for life.�   

We really don�t know and it�s not unproblematic

Just because you�ve got the freedom to do polling - even if you have the technical capacity of doing accurate ones, starting with an accurate national census frame, even if you have the cooperation of the public in agreeing to be interviewed and then telling the truth - I think there is great need for humility.

Taking everything into account, it may be a bit difficult to praise this type of toy in all respects, but it�s much more difficult to say it shouldn�t be there

No comments:

Post a Comment