By Dr Ebrima Ceesay
A closer reading of the CepRass latest survey methodology raises several concerns that do not necessarily invalidate the poll but do place clear limits on how confidently its findings can be interpreted.
The most immediate red flag is the claim of a perfect response rate, no refusals and no missing data.
In real-world survey research, that is exceptionally rare. Even in tightly managed fieldwork environments, some respondents decline participation, skip questions, or provide incomplete answers.
The absence of any such cases suggests that something unusual has happened in the data collection or processing stage. It may reflect strong interviewer control, but it could just as easily point to subtle pressures on respondents, overly directive field practices, or post-fieldwork data cleaning that removed or normalised non-response.
Whatever the explanation, it reduces transparency and makes it harder to assess how representative the responses truly are.
This concern becomes more significant when considered alongside the sample composition. Nearly half of all respondents are drawn from a single area, Brikama, which is an unusually large concentration for a national survey.
Even if weighting adjustments have been applied to correct for population distribution, such a heavy regional skew can still shape the overall results in meaningful ways. Weighting can rebalance proportions mathematically, but it cannot fully eliminate the influence of a dominant cluster in the raw data, especially when that cluster exhibits strong, consistent views.
Given that Brikama appears among the most dissatisfied regions in the dataset, there is a real possibility that the national picture is being pulled in a more negative direction than a more evenly distributed sample might produce.
The choice of face-to-face interviews adds another layer of complexity. On paper, this method has clear advantages: it improves response rates, allows for better inclusion of respondents with lower literacy levels, and enables clarification of questions in real time.
But it also introduces well-known risks. Respondents may tailor their answers to what they believe is socially acceptable or safe to express in front of an interviewer. In politically sensitive contexts, this can suppress criticism.
What is interesting here is that respondents were, in fact, quite willing to express negative views about government performance. That could indicate a genuine openness in public opinion, but it could also reflect variability in how questions were asked or understood across interviewers. Without detailed information on interviewer training and supervision, it is difficult to separate authentic sentiment from potential field effects.
Timing further complicates matters. The data was collected several months before the report was published, in a context where economic conditions are likely to be fluid. In such environments, public opinion can shift quickly in response to changes in prices, availability of goods, or government interventions.
A five-month lag does not render the findings irrelevant, but it does mean they should be treated as a snapshot of a particular moment rather than a current reading of the public mood. Any attempt to draw immediate political conclusions from them needs to account for that temporary gap.
There are also hints within the report itself that not all parts of the dataset are equally robust. The acknowledgement of inconsistencies in certain disaggregated results, such as regional variations in trust in the military, suggests uneven data quality.
When breakdowns do not align cleanly across categories, it can indicate issues with sample balance, question interpretation, or data recording at the field level. These kinds of irregularities don’t necessarily undermine the entire survey, but they do signal that some caution is needed when interpreting more granular findings.
Taken together, these methodological issues point to a dataset that is directionally informative but not tightly precise. The broad patterns it reveals, particularly around economic dissatisfaction and perceptions of government performance, are likely capturing something real.
However, the exact magnitudes of those sentiments, and the extent to which they can be generalised across the entire population, are less certain than the headline figures might suggest.
What remains persuasive, even after accounting for these limitations, is the overall direction of public sentiment. The poll consistently indicates that economic concerns, especially the cost of living, dominate how people evaluate government performance.
At the same time, it shows that democratic values remain intact, with strong support for participation, criticism, and protest. This combination is significant. It suggests a political environment marked not by disengagement or systemic rejection, but by pressure – pressure on institutions to deliver more effectively and on leaders to respond to immediate material concerns.
In that sense, the methodological weaknesses do not erase the core message, but they do refine how it should be understood. Therefore, this latest CepRass poll should not be read as a precise measurement of public opinion or a definitive guide to electoral outcomes. Rather, it is better seen as an indicative signal of a broader mood: one of economic strain, political awareness, and continued commitment to democratic engagement.
Ebrima Ceesay is an academic based in the UK. He was a former editor of The Daily Observer newspaper.
