A guide for the electorate or an attempt to influence outcomes?

Most of the polls have predicted that the ANC will get below 50%. Picture Leon Lestrade/Independent Newspapers

Most of the polls have predicted that the ANC will get below 50%. Picture Leon Lestrade/Independent Newspapers

Published Mar 30, 2024


Prof Dirk Kotzé

Recently, the South African National Editors’ Forum (Sanef) issued a statement in which it addressed the use of opinion polls and cautioned that some of them left “a lot to be desired when it comes to their depth, audience reach, and what exactly they want to achieve”.

The statement appeared in the context of the 2024 national and provincial elections which show new contestations among the parties and significant changes in public support for the parties.

The potential impact of external forms of influencing the public opinion is therefore more pronounced than in the past.

Most members of the public or voters are dependent, in the first instance, on the media for their information relevant for the elections. The results of opinion polls are also communicated mainly in the media, and electoral analysts depend on the polls for their interpretations or comments. It means that polls have the potential to be used or abused by parties and for the media to promote partisan ideas or opinions. The media is not expected to be non-partisan but the information they use should be authentic.

Therefore, it requires ethical conduct and professionalism by the institutions conducting opinion polls, such as private companies, foundations or research institutes. In South Africa, no statutory requirements are in place for the polling bodies to be accredited or registered to a professional body or adhering to a code of conduct. No guidelines or legislation that regulate them are available. Few limitations are in place, such as that exit polls are not allowed in South Africa on the day of the election.

The only oversight mechanism available for such a purpose but not applicable to all these bodies, is Esomar (originally the European Society for Opinion and Marketing Research, established in 1947). It is the world’s largest network of organisations and businesses involved in marketing, opinion and data analytics. In South Africa, Ipsos Markinor, is a member. Esomar provides its members with guidelines on global standards, ethics and best practices.

In South Africa, membership of such organisations is by choice and therefore cannot serve as a professional oversight body for opinion polling in the country. Membership of such an organisation implies that the member must comply with the organisation’s standards and that will enhance the member’s reputation or integrity in its own environment. But it is not enforceable by South African authorities or market research associations. What is left is the media’s own ethical and professional standards and general legal prescriptions about use of information.

Public polling for election purposes seldom stands on its own feet. It is often part of a broader market research infrastructure which is primarily directed towards the private sector. The information, therefore, remains private and is not meant for public consumption. The potential impact of that information is therefore better controlled and also more limited.

Three bodies are engaged in electoral opinion polling in South Africa in recent times, namely the Social Research Foundation (using Victory Research as is polling agent), Ipsos Markinor and the Brenthurst Foundation. The Human Sciences Research Council (HSRC) conducts opinion polls on matters related to elections but not on party support per se. The Institute for Justice and Reconciliation and Afrobarometer are also players in the field but follow the same approach as the HSRC. The Council for Scientific and Industrial Research (CSIR) in Pretoria is not engaged in polling but in accurate projections of what the final elections results will be once about 10% of the results have become public.

Why are the matters important?

The polling results made public in recent times by the three polling agencies diverge. They are used by parties in their campaigns to support a particular message. They are also used by analysts in the media in support of their arguments, such as whether the ANC will retain or lose its 50%+1 majority or the growth of parties, like the EFF and MK party.

The results of opinion polls require good knowledge of the requirements of quantitative research. Knowledge of the methodology used by polling agencies are therefore important, such as the size and composition of the population sample, the methods used in the interviews, the language used and the type of questions that were asked. The media and their analysts must have the knowledge and must understand how that will influence their interpretation of the results. In South Africa, that is not a standard practice.

Sanef’s cautionary note therefore appears to be in response to concerns that the polling results can be misrepresented in the media and, therefore, present misinformation to the public. It might even be part of a strategy of some parties in their campaigning.

Sanef’s call is justified. It does not only apply to the media but also their choice of who they use as analysts. It, therefore, implies a call for professionalisation of academics, researchers and media persons who play the role of analysts.

Prof. Kotzé is from the Department of Political Sciences at Unisa

**The views expressed do not necessarily reflect the views of Independent Media or IOL