The campaigns have been bruising. Each of the two major camps has exuded confidence of winning the elections come Tuesday. Speaking to the same crowds but at different times, each camp has shown their ability to mobilise the numbers.
Amidst the campaigns have been two opinion polls: one by Infotrak and the other, IpsosSynovate. The two pollsters have put Raila and Uhuru as being neck to neck but with neither of them hitting the magical 50%+1 vote to enable them be declared the winner. So going by the polls, the Presidential vote could be headed to second round subject to effect of the undecided voters. Just how reliable are these opinion polls? Should we believe them?
Political opinions as we know them today have their genesis in the USA. In 1824, the ‘Harvisburg Pennsylvania’ newspaper ran a poll of its readers showing that Andrew Jackson was leading Quincy Adams in the Presidential campaign. The ‘Raleigh Star’ published a poll of 4, 000 voters conducted at political meetings-again the poll put Jackson ahead of Adams. As it turned out, both polls were dead wrong. In 1824, Quincy Adams won the contest!
In 1936, the Literary Digest mailed out 10 Million questionnaires- and had 2 million returned upon which they called the 1936 election for Alf Landon, the Republican candidate. On the other hand, George Gallup used a much smaller but a tightly controlled random sample, employing very well trained interviewers. He predicted that Franklin Roosevelt would win the election. The Literary Digest was wrong, George Gallup was right!
Opinion polls attempt to measure the public’s voting intentions at a given point of time. Since election campaigns are dynamic and opinion shifts significantly over time, it is imperative to look at polls conducted late on the campaign- just before the vote takes place. The latest poll by IpsosSynovate put Jubilee Candidate Uhuru Kenyatta at 47% and RailaOdinga of NASA at 44%. Infotrak on the other hand put RailaOdinga at 49% with Uhuru Kenyatta at 48%. How credible are these projections?
So what goes for a good opinion poll?For opinion polls to serve the intended purposes, there is need for pollsters to ensure a number of things: first, methodological issues are key- failure to use a random sample, interviewer effects (level of training and political inclinations), measurement problems among others. A sample will provide an accurate picture, as long as it is representative of the electorate.
A sample from Nyeri, Murang’a and Kiambu cannot be representative of the Kenyan voters. It will obviously be skewed in favour of Jubilee. Similarly, a sample from Kisumu, Homa Bay and Kisii will be biased towards NASA even before the data is collected and analysed! The Kenyan scene is quite tricky as voters’ decisions are not necessarily based on issues and matters manifestos but on tribe and regions. This makes voter representativeness near impossible!
Of concern is the fact that the two major pollsters divided the country into regions representing the old provinces of Coast, North Eastern, Eastern, Central, Rift Valley, Western, Nyanza, and Nairobi. Why use this approach when the current constitution provides for 47 Counties and that a winner must garner at least 25% of the votes cast in at least 24 Counties?
Second, a good opinion require that pollsters use proper wording and phrasing of the questions are vitally important to producing reliable, objective data. There is need to emphasize neutral wording of the items as this can make a big difference in the way people answer the question. A question such as “Why do you think the country is headed in the wrong direction & who is responsible?” by Infotrack makes it hard for the respondents. Why not split the two questions? And even then, the last bit is so leading, asking who is responsible focuses the answer on an individual rather than institutions. “What accounts for the state of affairs” would no doubt elicit different responses!
Public opinion are indeed great when they represent the will of the people and not the pollster. As such, survey questions should be carefully constructed, be clear and free of bias. Why didn’t pollsters carry out a survey on voter registration? On expected voter turnout in terms of gender, age level, level of education, religion?
A third consideration is the need for scientific polling to be accurate and thorough reporting of results. Pollsters need to include a statement of how the poll was conducted and what the limitations of the survey are. One weakness in the Kenyan polls has been the lack of clarity on how respondents were selected. The surveys’ margin of error and confidence levels have also not been adequately explained.
Given the neck-to-neck statistics, it would have been prudent to find out what the undecided voters were looking for in the candidates. Other issues such as the war in Somalia, unga, unemployment levels, the promise for free education could have formed excellent grounds for opinion polls. Just asking who you would vote for is not enough!
As we head to the all-important general elections, the question to ask is: can voters translate the opinion polls into action? Yes, and No…yes as some people may be influenced to vote for a candidate who appears the front runner…and no as most people do not vote based on one’s popularity but on ethnicity! But the best opinion poll is the actual vote by all the registered voters.
The turnout on Tuesday in NASA and Jubilee strongholds and the direction that the swing Counties will take will both combine to determine which coalition carries the day and by extension forms the next government! The outcome will tell which of the pollsters was right and which one was wrong! Voxpopulivoxdei!
Dr Masibo Lumala is a Senior Lecturer, School of Communication, Information and Media Studies, Moi University. Email: firstname.lastname@example.org