Opinion polls cannot help predict electoral winners

Ipsos Synovate lead researcher Tom Wolf addresses journalists during release of findings based on opinion polls touching on economic conditions between Cord and Jubilee supporters at their Nairobi offices on July 16, 2016 PHOTO | ANTHONY OMUYA | NATION MEDIA GROUP

What you need to know:

  • In analysis of social, psychological and cultural dynamics, opinion polls are shallow and cannot unearth deep motives, innuendos and hidden feelings.
  • The method assumes that human motives can be measured in bites and binary combinations.
  • Opinion polls simply produce numerical data amenable to the fallacy that bites and binary permutations can tell you the truth about human motives.
  • The polls end up giving a fraction of the truth and cannot present a holistic picture.

President Donald Trump’s win in the American elections last November may have surprised many, but it has been used to fundamentally reveal the deficiencies in research paradigms used in opinion polls. It showed that the simplistic approaches employed are not reliable and scientific enough for an holistic trustworthy picture. 

Opinion polls rely on quantitative approaches that simply consider the number of people who said this or the other, period. In analysis of social, psychological and cultural dynamics, this approach is shallow and cannot unearth deep motives, innuendos and hidden feelings or the “non-dit” — the veiled, the sublime and perhaps the shameful and unconventional. It assumes that human motives can be measured in bites and binary combinations. That is balderdash.

It is important to note that opinion polls are just one of the many forms of research surveys whose findings get to the public domain. Unfortunately, without exception, they all utilise quantitative research methods.

The challenge, especially in trying to understand human motives, is that they fail to visualise the social dynamics and treat society, not as a social system, but a techno-economic system. From this premise, their generalisations are small specks scratched from the tip of the iceberg. At best, quantitative research methods are suitable for natural sciences. Human beings are not atoms or plants.

Advocates of quantitative methodologies including my MBA teacher, Dr Mutua Mbithi of Jomo Kenyatta University of Agriculture and Technology, and his comrade in arms, Prof Joseph Rotich of Masinde Muliro University of Science and Technology, believe that quantitative researches are controlled, objective and generalisable in revealing truth about phenomena.

They operate from the premise that facts can be obtained by asking simple questions! This approach simply produces numerical data amenable to the fallacy that bites and binary permutations can tell you the truth about human motives.

They end up giving a fraction of the truth and cannot present a holistic picture. They do not take cognisance of the character of human creatures and the processes through which they create meaning that is either overtly expressed or hidden. Bluntly said, it is a mirage to imagine that human perceptions can be measured in terms of quantity, amount, intensity or frequency.

I agree with Charles Ochieng’ Ong’ondo and Julius Ouma Jwan, the famous duo of qualitative research methodologies, when they argue that we have to rethink use of quantitative research methodologies in humanities, cultural studies and social science research.

QUANTITATIVE RESEARCH

Recent cases show that the results from such studies are untrustworthy.

As an active operator in the academy in Kenya, I am aware of various schools of thought on research paradigms. The first group subscribes to quantitative research methodologies, the second is persuaded by qualitative approaches while a third group subscribes to an hybrid approach that combines the two.

The wars between these groups are often fought in boardrooms especially during defences of both Masters and PhD theses. Often, those who suffer as the titans lock horns are the students. But this should really never happen. The truth is that some research areas do not render themselves to quantitative methods.

In management theory, quantitative methods have failed to capture the human aspects of the social and work environment. I have taken time to look at quantitative researches meant to assess and improve the 8-4-4 system.

The researchers kept churning out figures and graphs on performance of the system as the ground burnt underneath our feet. Quantitative researches flopped miserably because they did not focus on what was not said. Opinion polls suffer the same scourge.

RELIABLE INTERPRETATION

In human science research, we need to problematise the socially constructed nature of reality from a multiplicity of viewpoints. We can only arrive at reliable interpretation when we generate and analyse holistic data rather rigorously.

Sometimes it may mean we interpret and describe the voice, tone, reflexive movements and examine patterns of words. We have to pay critical attention to unique circumstances of the context and participants to derive meaning.

To achieve reliable results, researchers have to establish connections through application of multiplicity of interconnected interpretive practices for a cumulative view of the subject matter. For opinion polls to be reliable they have to interrogate experiences of human beings in a variety of situations.

Prof Kabaji is the Founding Principal of Turkana University College [email protected]