Rating the Performance of Politicians in Kenya: Why Pollsters Need to Style Up
It has become a common phenomenon, pollsters coming up with percentages and creating stories that have profound effects on the psyche of the nation. Social science research often involves collecting data in order to capture the views of individuals. However, social sciences distinguish themselves from common sense and pure rumor mongering because of adhering to scientific principles. Science as a way of knowing is unique because of effort exerted to ensure replicability of findings, objectivity, logical reasoning, purposefulness, empirical evidence, clearly defined methodology, systematic study process and ethical consideration throughout the inquiry process. For any research to be worthy of the scientific label, it has to be thoroughly and systematically thought through in such a manner that others doing a similar study can get similar results subject to a given margin of error.
Pollsters in Kenya continue to baffle me; they failed in getting presidential polls right but that did not jostle them into changing their ways. In the recent past, they focused on rating politicians in the most haphazard manner possible. It really begs the question whether these people appreciate the complexity of delivering a scientific study.
An opinion poll, it may be argued, is relatively simple and straight forward. What we basically need is opinions of a representative sample. The real challenge when it comes to opinion polls is getting the representative sample. The challenge is compounded when the influence of politics has to be considered in identifying the representative sample. In ordinary circumstances, multistage sampling that involves creating clusters and by simple random picking a sample from the identified clusters would suffice. The challenge is that in politics, representativeness is affected by difficulty in choosing a cluster that would not deliver biased results. For instance, how does one ensure that the cluster chosen is not an area dominated by people who are politically biased? This can only be addressed by taking on huge samples. Flawed as the process maybe, opinion polls can give indication in terms of who is the preferred presidential candidate; as long as the sample is big enough and clusters are carefully considered.
The sickest thing to do is to use the polling strategies in trying to measure the performance of senators, governors or to rank counties. I am amused because such a methodology was used recently by Infotrak to rank governors and senators in the country. What is wrong with using opinion poll methodology to measure performance of politicians? First, the question of sampling, which is a challenge for all opinion polls, arises. But that is not the real problem; the real problem is reducing performance to a matter of opinion or perception. This is how these pollsters get it all wrong; performance is not a matter of opinion or perception.
A few months ago, the World Bank released a report in which Bungoma County was ranked among counties spending least funds on development. A few weeks ago, infortrack released a report ranking Governor Lusaka second among the best performing governors in Kenya. The question is “who is fooling Kenyans?” This week, infortrack released findings and ratings of senators based on their performance in the counties and in the senate. To the extent, infortrack is merely publishing opinions then their findings may be entertained. However, questions in terms of sampling procedure and sample size would have to be answered to determine whether their findings actually capture the opinions and perceptions of majority of the people. What is wrong is that infortrack findings are not presented as views, opinions of some people interviewed. They are presented with absolute certainty that they are providing an indication of levels of performance of politicians.
What variables are the pollsters using to measure performance? People’s perceptions or opinions are not valid measures of performance. For instance, my governor Lusaka and Wetangula my senator have been ranked highly as performers based on opinion poll. However, we know that people’s opinions are influenced by political processes rather than actual performance. To measure actual performance of governors, pollsters or researchers need objective measures. Such quantitative measures can include expenditure on development against recurrent expenditure, local revenue generation, change in livelihoods of locals, cost-benefit ratios for projects implemented, level of citizen participation in projects, Investment volumes attracted or enabled etc. Such measures would give clear indications in terms of initial conditions, the interventions, the cost of interventions and actual results from projects undertaken by the governors.
If we want to measure the performance of MPs, the constitutional roles in parliament and the use of CDF is the only objective measure. As for senators and women reps, we can only do popularity polls because they do not manage development funds. The only way of ranking senators is when one focuses on their contribution in the senate in terms of contribution to debates and the bills they sponsor and participation in senate committees. Pollsters in Kenya have to do a little more and carefully consider the variables they are using as well as their sampling techniques. I believe they have successfully managed to render themselves unreliable by churning out baseless percentages and graphs that real do not reflect reality. If done properly, opinion polls are an invaluable tool that shapes public perceptions and thought processes.
For any research or monitoring and evaluation queries, send your email to firstname.lastname@example.org or email@example.com