Media should help the public to interrogate opinion poll results

A recent opinion poll indicating that neither Uhuru Kenyatta nor Raila Odinga would win in the first round of voting if elections were held was, unsurprisingly, dismissed by both sides of Kenya’s political divide.

The opposition termed them irrelevant and diversionary. The ruling coalition claimed that the findings meant nothing to it.

It is now common for politicians to dispute polls that show them in a weaker position than they imagine themselves to be. Indeed, politicians often attack pollsters as compromised hirelings working for their political competitors. Following the release of the recent polls, the company in question had to come out publicly to deny claims of impartiality as it had earlier consulted for the opposition.

It is not a crime for pollsters to work for politicians. Serious politicians routinely work with pollsters to make the most of their messaging and campaigns. For mwananchi, opinion polls are important for democracy as they stimulate public debate, encourage civic participation, and make elected officials more accountable by acting as a barometer on their performance. Thus, politicians lie when they claim that polls “mean nothing” to them. The truth is that both the public and politicians pay close attention to polls.

Currently, the stakes are heightened by the reality that opinion polls are also a crucial source of news discourse with copious amounts of space dedicated to their analysis and what they mean for a country’s political direction.

In playing out its powerful role as the link between the public and pollsters, the media must help the public to usefully interrogate the content and methods used by pollsters. As a researcher, I sometimes find some of the methods used by pollsters a little opaque and their attention to Kenya’s political culture a bit unserious.

While some opinion polls in Kenya have been fairly accurate, such as a recent one which predicted the outcome of the recent by-elections, others are designed to mislead. Many polls claim to be representative of public opinion than they really are while some are simply nicely told conjectures.

Requiring pollsters to release details of the politicians who commissioned the polls is important as politicians often come up with pre-election polls that are favourable to them.

However, it is also important to ask how the survey questions were phrased and sequenced. It is very easy to frame a question that elicits a desired response from respondents. For instance, if a pollster asked how increased corruption has personally affected them followed closely by a question on whether they think the president is doing a good job, there is a high chance that the response will be negative.

Opinion polls can be entertaining but if they are unscientific, they give us responses to a set of questions rather than public opinion.

Also, the media should interrogate the representativeness of surveys. While most Kenyan pollsters use between 1,500 to 2,500 respondents, a good sample for opinion polling, some have been known to have skewed representation. This poor weighting inevitably leads to questionable results.

In addition, some of the data collection methods preferred locally, like household surveys, are often done haphazardly and without sensitivity to how specific local cultures articulate political preference. As a result, local pollsters speak of non-responses and respondents’ indecisiveness as if they are the same thing.

Moreover, pollsters rarely tell us whether respondents’ are registered voters and how demographics such as gender and age affect political preferences. This means that the public is denied other critical aspects of these surveys.

The public also needs to know the confidence levels pollsters have of their surveys. Anything less than 95 per cent confidence level is unacceptable.

In polls that are considered to be too close, the implications of the margin of error must be thoroughly explained so that polls are understood as snapshots of opinion and not election predictors.

The reason opinion polls should be thoroughly interrogated is because of their perceived bandwagon effect, whereby voters identify with candidates who will most likely win. This undermines democracy as voters vote tactically and not conscientiously.

Dr Omanga is the head of department, Publishing and Media Studies at Moi University. [email protected]