Baselines and biases: The trouble with opinion polls in race to the local elections

The 2026 municipal elections in South Africa have sparked debate over the reliability and quality of opinion polls. While the polling industry is essential, concerns have arisen about its regulation, including transparency, representativity, margins of error, and conflicts of interest with candidates. Pollsters claim to simply reflect voter preferences, but their interpretations can influence public opinion and impact voter sentiment.

Parties typically rely on their own polling data to refine their strategies, but post-election analysis is rarely conducted to compare predictions with actual outcomes. The Social Research Foundation (SRF) survey has generated significant discussion, with figures showing declining support for the African National Congress (ANC) and increased support for alternative parties.

However, the SRF's interpretation of the data, influenced by close ties to the Democratic Alliance (DA), appears to align with their policy preferences. To assess the impact of baselines on the upcoming elections, analysts have calculated hypothetical starting points based on the 2024 election results compared to the 2019 and 2021 proportional representation outcomes.

This crude experiment reveals that the ANC's support in 2024 had declined significantly, with a potential baseline of 32% in local government elections. Meanwhile, the DA's baseline is stable in the mid-20s, while the EFF's support remains around 10%. The findings highlight the importance of evaluating the reliability and quality of polling data post-election, rather than solely focusing on the reliability of the data itself.