Election:
A Summary-dive into the methodological failures, arithmetic errors, and political misrepresentations of Gambia Participates’ controversial survey – and why you should question everything you have read.
As a data scientist, I was drawn into this discourse not only by professional curiosity but also by a sense of civic duty following numerous public appeals for my analytical verdict. When public trust demands an evidence-based interpretation, one cannot abdicate responsibility. It is, therefore, in this context that I provide a summary of my rigorous dissection of the said opinion poll.
Below are my summary findings already published by the Standard Newspaper and Alkamba Times.
As The Gambia marches toward what is anticipated to be one of the most consequential elections in its democratic history, public discourse is being increasingly shaped by polls and surveys claiming to reveal the nation’s political pulse. Yet, not all polls are created equal. Some illuminate, others obscure. Some inform, others mislead. A recent survey released by Gambia Participates, titled “Public Perceptions and Leadership Prospects Ahead of the 2026 Presidential Election,” falls squarely into the latter category. Touted as a snapshot of national opinion, this poll has sparked heated debate and uncritical headlines. But a forensic, data-driven examination by DataEd experts under the leadership of Touba Marrie reveals a deeply flawed document – one riddled with methodological shortcomings, arithmetic sloppiness, and interpretive overreach that render its conclusions not just unreliable, but dangerously misleading.
This is not a matter of mere academic disagreement or political banter. It is about the integrity of public information on the eve of a crucial electoral contest. Here are the most salient summary points from a detailed technical dissection of the poll:
1. A Statistically Irrelevant Sample Masquerading as National Representation:
The entire edifice of the survey rests on a shockingly small sample size of just 1,556 respondents. To put that in perspective, The Gambia has over 962,000 registered voters. This means the poll claims to speak for the nation after listening to a mere 0.16% of its electorate. Even more absurd is the distribution of this tiny sample across 53 constituencies, resulting in an average of only 29 people per constituency being asked to represent the views of tens of thousands. Would you trust 29 people in Brikama to speak for over 350,000? This is not a representative sample; it is a statistical echo chamber.
At the heart of any opinion poll lies its sample size and representativeness. Gambia Participates claims to have surveyed 1,556 respondents across 53 constituencies. On the surface, this appears comprehensive. Yet when subjected to analytical scrutiny, this sample represents a mere 0.16% of the 962,157 registered voters as per the IEC’s most recent data.
From a statistical standpoint, the law of large numbers suggests that reliability in predictive surveys emerges when sample sizes adequately mirror population heterogeneity. A 0.16% penetration rate in a diverse polity such as The Gambia is not only insufficient but risks systematic bias, particularly when the electorate is highly fragmented along regional, generational, and partisan lines. To assert that such a minuscule cohort can stand in as the “voice of the Gambian people” is analytically reckless and politically misleading.
2. A Track Record of Predictive Failure:
The poll’s association with CepRass invites scrutiny of past performance, and the historical record is damning. Previous predictions have repeatedly and significantly missed the mark on scale, even when occasionally getting the winner right. In the 2021 presidential election, CepRass underestimated President Barrow’s victory margin by a staggering 22 percentage points. In the 2023 local elections, errors exceeded more than 28 points in key areas like Banjul. A poll that is consistently wrong by such vast margins has no business claiming predictive authority for 2026.
3. Basic Arithmetic Errors That Undermine Credibility:
Perhaps the most glaring red flag is the presence of fundamental arithmetic mistakes. On the central question of perceived frontrunners, the reported percentages do not add up to 100% – they sum to 99%. Similar summation errors appear in regional breakdowns, with totals veering to 101% in Brikama, Kerewan, and Banjul, and falling short in others. In the world of data science, this is not a minor oversight; it is a cardinal sin that indicates sloppy data handling, poor validation, and a lack of basic quality control. If they cannot add simple percentages, why should we trust them to model complex voter behavior?
4. A Skewed and Unrepresentative Respondent Pool:
The demographics of the respondents further erode the survey’s credibility. About 62.85% of those polled were men, compared to 37.15% women, despite women making up roughly half the population. Furthermore, about 38.56% of respondents were reported as unemployed, and 33.87% had a tertiary education – a likely overrepresentation of urban, educated males whose views may not reflect the broader electorate. A family dinner where only the men are asked for their opinion does not produce a family consensus – it produces a biased narrative.
5. When Recalculated, the Poll’s Own Data Tells a Completely Different Story:
When the survey’s percentages are applied to the actual number of registered voters and voter turnout from the 2021 Elections by the Independent Electoral Commission (IEC), the narrative flips dramatically. The recalculated numbers show:
i. Essa Faal and Adama Barrow in a virtual statistical tie, where Faal edged him in first place.
ii. Mama Kandeh and Ousainou Darboe separated by a mere few hundred votes, while Talib sits last at 5th position nationally.
iii. Most importantly, the combined “Undecided” and “Other” categories emerge as the largest bloc by far – a whopping 270,623 voters, more than any single candidate. This proves the election is wide open, and the so-called “frontrunner” is a mirage. The real story is the vast pool of persuadable voters yet to make up their minds.
6. Biased Question Design That Primed Respondents:
The survey’s questions were not neutral. They disproportionately focused on Adama Barrow, Ousainou Darboe, and Talib Bensouda, giving them undue prominence while sidelining other potential candidates like Yankuba Darboe, Essa Faal, Mama Kandeh, and Rohey Malick Lowe. This “priming effect” is a well-known flaw in survey design, by repeatedly mentioning certain names, you artificially inflate their perceived viability in the minds of respondents. The poll did not measure public opinion; it subtly shaped it.
7. The Illusion of “Public Opinion”:
Labeling the views of 1,556 people as “public opinion” is perhaps the poll’s greatest misrepresentation. This tiny group’s views are presented as the national will, creating a misleading illusion of consensus or trend. This is not public opinion; it is, at best, a handful of respondents’ opinions. The Gambian public deserves to know the difference.
8. Dangerous Implications for Democracy:
Flawed polls are not only harmless. They can manipulate public perception, influence campaign strategies, shape media narratives, and even demotivate voters whose candidates are falsely portrayed as unelectable. In a young democracy like The Gambia, still consolidating after decades of authoritarian rule, the irresponsible use of data is not just misleading – it is a threat to the integrity of the electoral process itself. True progress demands polls grounded in unassailable data, transparent methods, and humble interpretations – lest they erode the very trust they seek to foster, leaving voters to navigate a fog of uncertainty toward an uncertain future like a long rainy day in Nyambia Forest without moonlight.
9. UDP Leadership Prospects:
Gambia Participates’ take the UDP leadership succession query: 44% favor Bensouda as Darboe’s heir, per the poll. Yet, this intrudes on internal party mechanics, decided solely by UDP members under Gambian political norms, not public whim. Without disclosing respondent affiliations, the results lack substance; if few or none are UDP cardholders, it is mere speculation. Empirically, this conflicts with binding decisions, risking mischaracterization. A non-member expressing a preference on UDP succession is devoid of political legitimacy and analytically insignificant. For everyday understanding, it is like asking neighbors who should run your family business – entertaining, but irrelevant to the actual choice.
10. A Legacy of Predictive Inaccuracy on Scale – Ignoring Historical Precedent:
GP’s association with CepRass, whether official or not, invites scrutiny of past predictive performance. The historical data provided reveal a persistent and systemic failure in scale prediction.
Error Margin Analysis: While past polls may have correctly identified winners (a binary outcome), they consistently misjudged the magnitude of victory by staggering margins. For instance, The 2021 Presidential Election (PE) underpredicted the NPP’s actual vote share by 22 percentage points. The 2023 Local Government Area (LGA) elections showed even wider chasms, with errors exceeding 28 and 23 percentage points for major parties in key areas like Banjul and Brikama, respectively.
Implication: This historical pattern demonstrates a chronic inability to model voter behavior accurately. A poll that gets the winner right but is off by over 20 points is not a successful forecast; it is a lucky guess that fundamentally misreads the political landscape.
11. Historical Precedent — Polling Versus Political Reality:
It is not unprecedented for opinion polls to diverge significantly from electoral outcomes. Indeed, even in advanced democracies like the United States, polling has frequently faltered, most notably in 2016 and 2024 when the intricacies of the electoral college and underreported voter blocs upended projections. CepRass itself has historically erred in The Gambia: polls correctly identified victors but consistently failed to predict margins, betraying the fragility of their inferential assumptions.
Unlike the U.S. electoral college system, where polls may misestimate localized distributions but still approximate the national narrative, The Gambia operates under a simple-majority system. Here, miscalculations in popular support cannot be statistically absorbed; they become existential flaws in predictive accuracy.
12. The Fundamental Flaw, Mischaracterizing Public Opinion:
Perhaps the most troubling dimension of this survey is the misuse of terminology. Branding the document as a representation of “Public Perceptions” is disingenuous when the reality is that it reflects only a minuscule, arbitrarily selected subset of citizens. A more accurate descriptor would be “Respondents’ Perceptions”. The former suggests a national mandate, the latter acknowledges methodological limitations. This careless mischaracterization transforms what should be a bounded dataset into a weaponized political narrative.
Conclusion: The 2026 election will be decided by a complex mix of governance records, economic realities, coalition building, and grassroots campaigning – not by a deeply flawed survey of 1,556 individuals. True democracy requires transparency, rigor, and intellectual honesty. As citizens, we must demand better. We must question the numbers pushed before us and ask: Who produced them? How? And for what purpose?
DON’T BE MISLED! KNOW THE FACTS!
This article is only a summary. The full, detailed 10-page forensic dissection – complete with data tables, methodological deep dives, and point-by-point debunking – is available for politicians, civil society organizations, and political analysts who seek to truly understand the manipulation of data in the ongoing electoral discourse.
Don’t be misled. Know the facts!
For access to the full report, “The Empirical Mirage: A Forensic Dissection of Misrepresentations in Gambia Participates’ 2026 Presidential Opinion Poll,” please get in touch with the DataEd CEO:
Touba Marrie (Data Scientist & Mathematician)
Email: tm2448@nyu.edu
AboutTheAuthor: Touba Marrie is a trailblazing scholar, a Data Scientist and Mathematician who captivates the world as a BI Developer and Data Architect at Keystone Industries in the USA. As the audacious founder and CEO of Data Analytics & Education Consulting Group (DataEd) in The
Gambia, he’s revolutionizing how data shapes futures. Armed with a Master of Business and Science in Industrial and Applied Mathematics (Distinction) from Rutgers University, a Master of Science in Data Science and Analytics from Cardiff University, and a Bachelor of Science in Mathematics, Economics & Finance from New York University, Touba’s academic and research journey spans continents from decoding The Gambia’s election data to spearheading transformative big data projects across the USA, UK, UAE, and beyond. Touba’s expertise electrifies industries. His relentless pursuit of innovation in data analytics and industrial research is setting the global stage ablaze, positioning him as a leading global voice in the application of data for decision-making and innovation.
