Writer wrong on opinion pollsters

 

Yet again, I was dismayed at the many falsehoods and errors in a recent piece about Kenya’s survey firms (‘The inconsistencies in opinion polling’, The Standard, September 9, by Abel Oyieyo, especially as he claims to speak with such authority on the subject.

He claims that first, “pollsters in Kenya have formed the habit of sharing very unlikely figures”, and that “unknowns include the methodology Kenyan pollsters use for sampling in such a tribal country with non-homogeneous tribes...”

What makes any particular survey result “unlikely”? Guesswork? Mr Oyieyo, while referring to Ipsos, presents no contrary evidence to the results that he questions. Worse, he ignores the critical importance of sampling demographics: the total ‘identity-portrait’ of the respondents. It includes especially: geographic region, gender, age, education level, employment status, religious affiliation, income and political party or coalition orientation.

All of Ipsos’ survey releases include such ‘portraits’. If the media (for whatever reasons) fail to report or publish these details, anyone can check with the source to obtain them. Moreover, is he aware that the Publication of Electoral Opinion Polls Act (2012) requires that all such methodological details accompany the results of any voter-intention survey published within the 12-month period preceding an actual election?

Secondly, that recent polls must be wrong because "Jubilee and CORD have strongholds for each (potential presidential candidate) not likely to fall below 30 per cent". Faulty arithmetic. In Ipsos’ most recent survey (June 2016), this is exactly what was found regarding expressed likely voting intentions: 53 per cent for Jubilee (Uhuru 51 per cent + Ruto 2 per cent), and 32 per cent for CORD (Raila 28 per cent + Kalonzo 2 per cent + Wetang'ula 2 per cent). And given that CORD has yet to name its presidential flag-bearer, it is unsurprising that 12 per cent remain “undecided”. So what is the problem?

Thirdly, “Kenyan voting is informed by tribe and a tribal kingpin can sway an entire tribal vote to someone from another tribe”. Another fuzzy over-generalisation. Who, in fact, is a “tribal kingpin”, and what determines this? Not Musalia Mudavadi, who in 2013 garnered only about one-third of the Luhya vote? Not Kalonzo Musyoka, CORD’s DP running-mate, among whose fellow-Kamba about 15 per cent voted for Jubilee in that election? And what is “tribal voting” for groups lacking any (credible) presidential or deputy presidential candidate?

Fourthly, that “concerted effort by pollsters, the media, and members of the public... will help make opinion polling in Kenya good enough to be used for planning and strategising". Kenya’s survey firms have been conducting such surveys for both state and private sector clients for years.

For example, Ipsos (first as Steadman and subsequently as Synovate) conducted national household surveys for the Ministry of Justice and Constitutional Affairs in both 2006 and 2011, first, to establish baseline-targets for its GJLOS reform programme, and subsequently, to ascertain the extent to which they had been met, while identifying the factors involved.

And if clients sometimes attempt to analyse the results using sub-samples that are too small for statistical significance, that is no fault of the survey firms who collect the data.

Fifth, that “we need to make the field competitive by opening it to competition". The field is already open. The Market Survey Research Association (MSRA) of Kenya has 15 corporate and nearly 50 individual members. Membership imposes adherence to international best practice in the field. Moreover, no law prevents the entrance of any new firm either to MSRA or to survey work (pending the granting of the appropriate Government annual licence), though the media (and the public) are advised to be wary of new entrants without a proven track-record and who shy away from MSRA membership.

In this connection, how can Mr Oyieyo make comparisons with the US in terms of the number of firms or frequency of polls, whose population is more than seven times that of Kenya, and whose economy is 330 times larger? As such, the "failings" which he cites cannot be attributed to any lack of imagination by Kenya’s pollsters or media houses.

Sixth, that because “opinion polls can help swing public opinion... it is imperative that they be reliable and as realistic as possible". What evidence does Mr Oyieyo offer to support this claim in Kenya? None, because there isn’t any.

Finally, that “we are still making the mistakes some of the developed countries made 100 or 200 years ago". Which countries even existed that long ago? And scientific survey research began only in the 1930s. More important, nearly all of Kenya’s market research firms that undertake polling have international partners; Ipsos is itself the world’s third largest, with a particularly high profile in this field (eg, the Reuters/Ipsos partnership in the US, and Ipsos/Mori polls in the UK).

But after all my complaints, I end by agreeing with Mr Oyieyo that “the media must ask questions” about pollsters’ methods and results. It is thus especially regrettable that he failed to do this himself before subjecting readers to so many misplaced criticisms.