The Gambia Have Your Say

08/09/2026

30/08/2026

Why the Poll's Kanifing and Brikama Leads May Be Less Certain Than They Appear, Analyst Argues In the second part of his review of Gambia Participates' Round 2 election poll, data scientist Touba Marrie turns to the finding that has generated the most public debate: that Talib Bensouda leads current vote intention in both Kanifing and Brikama. His conclusion is that the statistics cannot yet support that claim with confidence — and that the underlying politics make the finding harder to explain, not easier. THE STATISTICAL CASE Marrie's core argument rests on sample size. The national poll of 1,556 respondents carries a margin of error of roughly plus or minus 2.5 percent, but splitting that sample into regional subsamples sharply increases the uncertainty around each individual region's numbers. In Brikama, where roughly 373 people were interviewed, he calculates the margin of error at close to plus or minus 5 points. In Kanifing, with roughly 265 respondents, it widens to around plus or minus 6 points. Against margins that size, he argues, the reported gaps between candidates in both regions — Bensouda at 24 percent, Barrow at 23 percent and Darboe at 20 percent in Brikama; Bensouda at 27 percent against Barrow's 24 percent and Darboe's 21 percent in Kanifing — fall inside or right at the edge of statistical noise. In his words, the poll can say the race in these regions is close, but cannot reliably say who is actually ahead. THE POLITICAL CASE Beyond the statistics, Marrie argues the finding lacks an obvious real-world explanation, particularly in Brikama. UNITE, Bensouda's party, did not exist as a formal political party until its national congress in May 2026, roughly three months before the poll was conducted, and has never appeared on a ballot anywhere in the country. Brikama's Area Council, by contrast, is held by Yankuba Darboe of the UDP, who won the seat in 2023 by one of the largest margins of that entire election cycle, with the NPP a clear second. Marrie says nothing has been publicly documented to explain how a three-month-old party with no prior electoral footprint would be leading in a region held by two established parties with recent, decisive wins there. He treats Kanifing somewhat differently, noting that Bensouda has a genuine personal electoral history there as the twice-elected sitting Mayor — though he points out he won both those elections as a UDP candidate, before UNITE existed, making it a reasonable but still untested question whether that personal popularity transfers to his new party. His conclusion is not that the finding is wrong, but that a razor-thin statistical lead sitting on top of a region with no visible party organisation to explain it deserves closer scrutiny — a larger regional sample and a second polling round to see if the result holds — before being treated as settled. A POLL IS A SNAPSHOT, NOT A DESTINATION Marrie illustrates how quickly conditions can shift with an example from the survey's own field period. One question in the poll tested whether a coalition of Bensouda, Essa Mbye Faal and Mama Kandeh could win the election. On July 29, two days into the poll's three-week data collection window, Kandeh signed a formal alliance bringing his GDC party into President Barrow's NPP. By the time the report was published on August 21, Faal had also withdrawn from the opposition coalition process, Bensouda had emerged as the flagbearer of a differently-composed Coalition 2026, and the UDP had formally exited those talks altogether. In other words, Marrie notes, the coalition tested in the poll no longer existed in its tested form by the time the results reached the public — not through any fault of the pollster, but because political events did not pause for the survey to finish. THE ESSA FAAL QUESTION Marrie devotes particular attention to a sharp shift in Essa Mbye Faal's numbers between the two survey rounds. A year earlier, he notes, he registered 18 percent in Brikama and 21 percent in Kanifing on the same "perceived chance of winning" measure. In the new round, those figures fall to roughly 2 percent and 0 percent respectively. He argues that scale of collapse is difficult to explain through normal political attrition, noting that Faal's Sobeya movement actually merged into the larger APP-Sobeya coalition during the intervening period — organisationally a broadening move, not a retreat — and that political support rarely simply evaporates without a clear triggering event such as a scandal, a major electoral defeat or a public withdrawal from politics, none of which occurred in Faal's case. He notes he recently received a warm reception at the UDP's 30th anniversary celebration, and cautions that writing him out of the political story entirely, on the strength of this one number, could prove premature. A companion piece summarises Marrie's broader message to voters and political parties about how to responsibly use — and not misuse — polling data ahead of December's election. selected

29/08/2026

29/08/2026

Data Scientist Says New Election Poll Is Better Than Last Year's — But Still Has Real Blind Spots A data scientist reviewing Gambia Participates' second-round presidential election poll says the survey represents a genuine improvement over its 2025 predecessor, while identifying several gaps he argues could be misleading the public if left unaddressed. Touba Marrie, a data scientist and mathematician originally from Foni Kandonku, published a detailed public analysis of the poll, which surveyed 1,556 registered voters across all eight Local Government Areas between July 27 and August 10, 2026. Marrie credited the pollster, Gambia Participates, for clearly separating "who would you vote for" from "who do you think will win," disclosing raw respondent counts rather than only percentages, and explicitly warning readers against misreading the results — practices he said reflect responsible polling. But he argued a well-conducted poll is still only "a photograph, taken on a handful of days," and laid out several areas where he believes that photograph is blurry. THE MISSING QUESTION: WHICH PARTY DO YOU SUPPORT? Marrie's central criticism is that the poll never asked respondents which political party, if any, they currently support or lean toward — a question he says most professional pollsters treat as standard alongside age and gender. Without it, he argues, nobody outside Gambia Participates can check whether the sample happened to lean toward one party's supporters before a single question was even asked, which could tilt every subsequent answer without anyone being able to detect or correct for it. He pointed to Round 1 of the same poll as a concrete example of the risk. That survey found 44 percent of respondents wanted Talib Bensouda to succeed Ousainou Darboe as the UDP's presidential candidate. Bensouda never received that succession — he left the UDP entirely in September 2025 to found his own party, UNITE. Marrie argues this illustrates how a question about internal party succession, answered largely by people outside that party's actual decision-making process, can tell the public very little about what will really happen inside it. He was careful to frame this as a transparency gap rather than an accusation of bias, and proposed a simple fix: adding a single party-identification question in future rounds, which he said would let researchers weight results the same way the survey already does for gender and geography. THE EMPLOYMENT DATA THAT WAS COLLECTED BUT NEVER USED Marrie also flagged that the poll recorded each respondent's employment sector — 19 percent public sector, 17 percent private sector, 2 percent NGO, 23 percent informal sector, and 39 percent unemployed — but never cross-referenced this data against any political question. Every major finding in the survey is broken down by gender, age and region, he noted, but never by employment sector. He called this a significant missed opportunity given that the same poll found "the economy and cost of living" to be the single largest factor shaping how Gambians say they will vote, cited by 76 percent of respondents. Unemployed and informally employed respondents together make up 62 percent of the sample, he pointed out, meaning the group most exposed to economic hardship is also the group whose views were never separately examined. Marrie said he does not believe this was a deliberate omission, but rather a case of data being collected and then left unanalysed. SAMPLE REPRESENTATION CONCERNS Marrie raised two further concerns about how well the sample reflects the country. He noted the poll's sample was 59 percent male, compared to a roughly even gender split in the national population, with no disclosed statistical correction applied. He also noted that within large, diverse regions like Brikama and Kerewan, the report does not specify how interviews were distributed between urban centres and rural settlements, making it difficult to judge whether a region's mix of interviews reflects its real political and geographic texture. He stressed he was not asking for the release of any respondent-identifying information, only an aggregate, ward-level breakdown of where interviews took place. WHAT SHE RECOMMENDS Marrie's overall assessment was that none of these gaps make the poll "wrong," but that they limit how confidently its findings — particularly at the regional level — should be read. He recommended that future rounds add a party-affiliation question, publish the employment-sector cross-tabs using data already collected, and disclose a ward-level breakdown of where interviews were conducted, changes he said would strengthen public confidence in the survey without requiring a larger sample. A companion piece examines Marrie's specific concerns about the poll's findings in Brikama and Kanifing, along with his broader argument about how much can change between a poll's field dates and election day. selected

29/08/2026

𝐒𝐔𝐑𝐕𝐄𝐘𝐒 𝐃𝐎𝐍'𝐓 𝐕𝐎𝐓𝐄 – 𝐘𝐎𝐔 𝐃𝐎

𝐁𝐲 𝐓𝐨𝐮𝐛𝐚 𝐌𝐚𝐫𝐫𝐢𝐞, 𝐃𝐚𝐭𝐚 𝐒𝐜𝐢𝐞𝐧𝐭𝐢𝐬𝐭 𝐚𝐧𝐝 𝐌𝐚𝐭𝐡𝐞𝐦𝐚𝐭𝐢𝐜𝐢𝐚𝐧

𝐀 𝐏𝐞𝐨𝐩𝐥𝐞'𝐬 𝐆𝐮𝐢𝐝𝐞 𝐭𝐨 𝐭𝐡𝐞 2026 𝐆𝐚𝐦𝐛𝐢𝐚 𝐏𝐚𝐫𝐭𝐢𝐜𝐢𝐩𝐚𝐭𝐞𝐬 𝐏𝐨𝐥𝐥: 𝐖𝐡𝐚𝐭 𝐈𝐭 𝐆𝐞𝐭𝐬 𝐑𝐢𝐠𝐡𝐭, 𝐖𝐡𝐞𝐫𝐞 𝐈𝐭 𝐅𝐚𝐥𝐥𝐬 𝐒𝐡𝐨𝐫𝐭, 𝐚𝐧𝐝 𝐖𝐡𝐲 𝐘𝐨𝐮𝐫 𝐕𝐨𝐭𝐞 𝐒𝐭𝐢𝐥𝐥 𝐃𝐞𝐜𝐢𝐝𝐞𝐬 𝐄𝐯𝐞𝐫𝐲𝐭𝐡𝐢𝐧𝐠.

I would like to begin, as I always do, by extending sincere gratitude to Gambia Participates for their continued commitment to opinion research in The Gambia. As The Gambia moves toward what many believe will be one of the most closely contested presidential elections in its history, public conversation is increasingly shaped by opinion polling and the predictive claims built on top of it. Gambia Participates' "2026 Presidential Election Opinion Poll – Round 2," released on 21st August 2026 and fielded between 27th July and 10th August 2026 among 1,556 registered voters across all eight Local Government Areas, is the latest and most detailed such contribution – and, like its predecessor a year earlier, it has already generated intense public debate.

Just to remind you, "When Mathematics Meets Politics, Something Doesn't Usually Add Up."

As a data scientist, I have been drawn into that debate again this year – not merely by professional curiosity, but by the same sense of civic duty that compelled me to examine Round 1. When members of the public repeatedly ask for an evidence-based reading of a document shaping how they think about their own election, silence is not a neutral option. It is, in that spirit, that I offer this public piece.

A fuller, more technical dissection of Round 2 – with every calculation shown, accompanies this piece as a companion report for readers who are technical with interest and ability in STEM. I will tell you at the end what you must do to see that technical document.

What follows here is written differently, and deliberately, in plain language: for the market trader, the taxi driver, the student, and the grandmother following this election as closely as anyone in a newsroom, because the right to understand what a poll does and does not say should never depend on a university degree.

Now, let's set the formalities aside for a moment and start with something simple. 𝐁𝐞𝐟𝐨𝐫𝐞 𝐚 𝐬𝐢𝐧𝐠𝐥𝐞 𝐧𝐮𝐦𝐛𝐞𝐫 𝐚𝐩𝐩𝐞𝐚𝐫𝐬 𝐢𝐧 𝐭𝐡𝐢𝐬 𝐩𝐢𝐞𝐜𝐞: 𝐰𝐡𝐚𝐭𝐞𝐯𝐞𝐫 𝐲𝐨𝐮 𝐟𝐞𝐥𝐭 𝐰𝐡𝐞𝐧 𝐲𝐨𝐮 𝐬𝐚𝐰 𝐭𝐡𝐨𝐬𝐞 𝐩𝐨𝐥𝐥 𝐫𝐞𝐬𝐮𝐥𝐭𝐬 – 𝐡𝐨𝐩𝐞, 𝐰𝐨𝐫𝐫𝐲, 𝐚𝐧𝐠𝐞𝐫, 𝐜𝐨𝐧𝐟𝐮𝐬𝐢𝐨𝐧 – 𝐢𝐬 𝐯𝐚𝐥𝐢𝐝, 𝐚𝐧𝐝 𝐢𝐭 𝐰𝐢𝐥𝐥 𝐧𝐨𝐭 𝐝𝐞𝐜𝐢𝐝𝐞 𝐭𝐡𝐞 𝐞𝐥𝐞𝐜𝐭𝐢𝐨𝐧. 𝐘𝐨𝐮 𝐰𝐢𝐥𝐥.

Gambia Participates has released Round 2 of its presidential opinion polls, and Gambians everywhere are asking me the same question they asked a year ago: "𝐓𝐨𝐮𝐛𝐚, 𝐢𝐬 𝐭𝐡𝐢𝐬 𝐭𝐫𝐮𝐞? 𝐒𝐡𝐨𝐮𝐥𝐝 𝐈 𝐛𝐞 𝐬𝐜𝐚𝐫𝐞𝐝? 𝐒𝐡𝐨𝐮𝐥𝐝 𝐈 𝐠𝐢𝐯𝐞 𝐮𝐩?"

So let's sit down, the way we would over Attaya on a slow afternoon or whatever time you read this, and go through it properly, not just the headline numbers, but the story behind them: what this poll got right, where it quietly fell short, what real history teaches us about polls like this one, and, most importantly, how you can use everything in it to your own advantage without losing a single night's sleep over it.

𝐅𝐢𝐫𝐬𝐭, 𝐂𝐫𝐞𝐝𝐢𝐭 𝐖𝐡𝐞𝐫𝐞 𝐈𝐭'𝐬 𝐃𝐮𝐞: Last year, I picked apart Round 1 of this survey with a fine-tooth comb, and I did not go easy on it. So it's only fair that I tell you plainly: Keep the results aside for a moment, round 2 is a more careful piece of work, and is a materially better piece of work than Round 1. This time, Gambia Participates separated "who would you vote for" from "who do you think will win. They told us how many people actually answered each question, not just tidy percentages. They even added a section warning readers not to twist the numbers. That is exactly what responsible pollsters are supposed to do. But a poll can be well-conducted and still not be the finish line. It is a photograph, taken on a handful of days in late July, of a country that has kept moving every single day since. I want to show you, with real numbers and real recent history, exactly where that photograph is sharp – and exactly where it's blurry.

1. 𝐀 𝐅𝐚𝐢𝐫 𝐇𝐢𝐬𝐭𝐨𝐫𝐲 𝐋𝐞𝐬𝐬𝐨𝐧: 𝐆𝐮𝐞𝐬𝐬𝐢𝐧𝐠 𝐑𝐢𝐠𝐡𝐭 𝐈𝐬 𝐍𝐨𝐭 𝐭𝐡𝐞 𝐒𝐚𝐦𝐞 𝐚𝐬 𝐏𝐫𝐞𝐝𝐢𝐜𝐭𝐢𝐧𝐠 𝐑𝐢𝐠𝐡𝐭
Being close, if not close enough on scale, could be anyone's guess.
Ahead of the 2023 Mayoral and Chairmanship elections, CepRass – the polling outfit mostly associated with opinion polls in Gambia went on record projecting by quality outputs that the UDP would win in Banjul, Kanifing Municipality, the West Coast Region, and the Lower River Region, while the NPP would take Basse, Kerewan, Janjanbureh, and Kuntaur. When the votes were actually counted, every single one of those eight calls landed correctly. Talib Bensouda kept Kanifing. Rohey Malick Lowe kept Banjul. Yankuba Darboe took Brikama for the UDP. The NPP held its rural strongholds exactly where CepRass said it would. That is a genuinely good track record on the question that matters most in a simple-majority system – who wins.

But here's the distinction that actually matters: Naming the winner correctly and predicting the result accurately on scale are not the same skill, and mixing them up is exactly how a country ends up trusting a coin that only ever seems to land the way it was already leaning. In a region with a long-dominant party, guessing that the dominant party will win again is not a hard trick – anyone following local politics casually could do it. The real test of whether a poll is doing its job is not whether the name matches. It's whether the number matches or close enough. Call it the difference between a correct guess and a correct prediction.

So let's look at the actual record, side by side, prediction (likely to win) against outcome, for every major candidate CepRass scored across the last two election cycles:
(i) 2021 Presidential Election, NPP (Barrow): predicted 31%, actual 53% (Miss: 22.0 points – Close Call: No)
(ii) 2021 Presidential Election, UDP (Darboe): predicted 15%, actual 28% (Miss: 13.0 points – Close Call: No)
(iii) 2021 Presidential Election, GDC (Kandeh): predicted 3%, actual 4% (Miss: 1.0 point – Close Call: Yes)
(iv) 2023 Banjul City Council, UDP (Lowe): predicted 26%, actual 54.3% (Miss: 28.3 points – Close Call: No)
(v) 2023 Banjul City Council, NPP (Faye): predicted 14%, actual 45.7% (Miss: 31.7 points – Close Call: No)
(vi) 2023 Brikama Area Council, UDP (Y. Darboe): predicted 23%, actual 46.3% (Miss: 23.3 points – Close Call: No)
(vii) 2023 Brikama Area Council, NPP (Ceesay): predicted 22%, actual 31.1% (Miss: 9.1 points – Close Call: No)
(viii) 2023 Brikama Area Council, Gitteh (Ind.): predicted 3.4%, actual 4% (Miss: 0.6 points – Close Call: Yes)
(ix) 2023 Kanifing Municipal Council, UDP (Bensouda): predicted 55%, actual 54.1% (Miss: 0.9 points – Close Call: Yes)
(x) 2023 Kanifing Municipal Council, NPP (Badjie): predicted 19%, actual 40.9% (Miss: 21.9 points – Close Call: No)
(xi) 2023 Kanifing Municipal Council, Pa Modou (Ind.): predicted 2%, actual 5.0% (Miss: 3.2 points – Close Call: Yes)

Of the eleven predictions (likely to win) tracked here, only four landed within a few points of the real result – and three of those four were for third-place or minor candidates, whose small vote shares are, mathematically, far easier to land close to by sheer chance than a genuine top-two contest is.
The one true standout is CepRass's 2023 Kanifing call for the UDP: 55% predicted against a 54.1% actual result – a miss of less than one point, on the single most consequential number in that entire election. That result proves, beyond doubt, that rigorous, accurate polling is achievable right here in The Gambia. It is simply not the norm. It is the exception that shows what's possible when the conditions line up – not the rule you should expect every time.
On the other seven predictions for the two leading parties, the misses run from 9 points all the way to nearly 32 points – several of them larger than the entire share of the vote separating first and second place in this year's GP poll.

A coin lands on the correct side about half the time, purely by chance. Nobody calls a coin toss a forecasting method – and a poll that names the right winner while missing the number by twenty points has earned exactly that same modesty, no more.
Keep every one of those numbers in mind the next time a headline tells you candidate x, "will win" Kanifing or the West Coast because a current poll shows him narrowly ahead there. History – CepRass's own history, on CepRass's own past predictions – says a Gambian polling body can name the exact right winner and still miss the actual number by the equivalent of an entire province's worth of votes. A snapshot that shows someone ahead today is not a guarantee of anything on 5th December. It has never been. Only the actual count is.

2. 𝐓𝐡𝐞 𝐏𝐡𝐨𝐭𝐨𝐠𝐫𝐚𝐩𝐡 𝐓𝐡𝐚𝐭 𝐌𝐨𝐯𝐞𝐝 𝐖𝐡𝐢𝐥𝐞 𝐈𝐭 𝐖𝐚𝐬 𝐁𝐞𝐢𝐧𝐠 𝐓𝐚𝐤𝐞𝐧
Here is something genuinely remarkable, and I want you to read this slowly, because it might be the single best real-world lesson in "why a poll is not a crystal ball" that I have come across in my whole career, anywhere.

Gambia Participates collected these interviews between 27th July and 10th August 2026. One of the questions they asked was whether a coalition of Talib Bensouda, Essa Mbye Faal, and Mama Kandeh could win the election.

On about 29 July – two days into that very same data-collection window – Mama Kandeh signed a formal alliance putting his GDC party into President Barrow's NPP, though Kandeh went on to say "join ak boka du benna". The man named as one-third of a hypothetical opposition coalition had, while the survey about him was still being conducted, walked across the aisle to the other side entirely.

It didn't stop there. By the time the report reached the public on 21st August, Essa Faal had also withdrawn from the opposition's own coalition process, and Talib Bensouda had emerged as the flagbearer of Coalition 2026 – a coalition that, in its actual final form, doesn't include either of the other two names the poll tested. The United Democratic Party, meanwhile, had formally walked out of those same coalition talks.

If a headline political figure can switch sides while the interviews are still being conducted, imagine how much more can shift between now and 5th December.

This is not Gambia Participates' fault. No survey, however well designed, can be redrawn hour by hour as political deals are signed behind closed doors. But it is the clearest, most concrete proof I can hand you that a poll captures a moment, not a destination.

3. "𝐎𝐧𝐞 𝐕𝐢𝐥𝐥𝐚𝐠𝐞 𝐒𝐩𝐞𝐚𝐤𝐢𝐧𝐠 𝐟𝐨𝐫 𝐚 𝐖𝐡𝐨𝐥𝐞 𝐑𝐞𝐠𝐢𝐨𝐧"?
Here's a question worth asking that hasn't gotten enough attention: within each of those eight regions, exactly where were people interviewed?

Brikama, as an LGA, stretches from dense, fast-growing towns like Sukuta's outskirts and Brusubi to quiet farming villages deep in the West Coast including my own Kandnku. Kerewan runs from busy roadside settlements to remote riverine communities. A region is not one place – it's dozens of very different places stitched together on a map, and they don't always vote the same way.

Gambia Participates tells us how many people it interviewed in each of the eight LGAs. It does not tell us how those interviews were spread across the towns, wards, or villages inside each LGA. With roughly 373 interviews across the whole of Brikama – a region containing dozens of settlements and well over three hundred and fifty-nine thousand registered voters, it matters enormously whether those 373 conversations happened mostly in a handful of accessible, easy-to-reach towns near the main road or were genuinely spread across the region's full urban-to-rural mix.

To be clear, I am not asking anyone to publish names or exact addresses – that would be a serious breach of respondent privacy, and no credible pollster should ever do that. But publishing a simple district- or ward-level breakdown – how many interviews happened in Busumbala versus Brikama Town versus a rural settlement, for instance, without naming a single person – would let political parties, journalists, and independent researchers like myself actually check whether a sample matches the real urban-rural and political texture of a region, instead of asking the public to simply trust an unlabeled number.

This is a fixable gap, and it's the kind of transparency that would strengthen every future round of this survey – for Gambia Participates and for any organization that follows in its footsteps.

4. 𝐓𝐡𝐞 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧 𝐍𝐨𝐛𝐨𝐝𝐲 𝐀𝐬𝐤𝐞𝐝: 𝐖𝐡𝐨𝐬𝐞 𝐉𝐞𝐫𝐬𝐞𝐲 𝐖𝐚𝐬 𝐭𝐡𝐞 𝐑𝐞𝐬𝐩𝐨𝐧𝐝𝐞𝐧𝐭 𝐖𝐞𝐚𝐫𝐢𝐧𝐠?
Picture two enumerators heading out to the same neighborhood on the same afternoon. One happens to knock on doors where the households are mostly UDP families. The other happens to knock on doors where the households mostly lean NPP. Neither enumerator did anything wrong – that's just the ordinary accident of who was home that day. But if nobody writes down which households were which, the final report can't tell anyone whether the neighborhood's 50-50 split is the real political mix of that street, or just the toss of a coin about whose door happened to be answered.

That, in essence, is the gap in both rounds of this survey. Nowhere does Gambia Participates ask a respondent the one question professional pollsters almost everywhere else in the world treat as standard alongside age and gender: which party, if any, do you currently support or lean toward?

Here's a simpler way to feel why that matters. Imagine conducting a poll about a football match – say, Real Madrid against Barcelona. Would it matter whether most of the people you asked support Real Madrid or Barcelona? Of course it would. Knowing the jerseys in the room would immediately make every other answer easier to read. Politics works exactly the same way. People don't walk into a survey as blank slates – they bring existing loyalties into every question they answer, whether the question is about a candidate, a coalition, or who should lead a party they may or may not belong to.

𝐖𝐡𝐲 𝐭𝐡𝐢𝐬 𝐦𝐚𝐭𝐭𝐞𝐫𝐬 𝐦𝐨𝐫𝐞 𝐭𝐡𝐚𝐧 𝐢𝐭 𝐬𝐨𝐮𝐧𝐝𝐬: Party affiliation isn't just another demographic box to tick. It's a diagnostic tool – it helps us understand the sample itself, not just the answers the sample gave. And understanding the sample is often just as important as understanding the answers. Without it, an independent analyst is effectively trying to inspect a car with the hood welded shut: you can watch how it moves, you can see the results roll past, but you can't see the engine that's actually producing them.

Right now, nobody outside Gambia Participates can answer basic questions that any serious election researcher would want answered before trusting a topline number: how many of the 1,556 respondents identify with the NPP? How many with the UDP? How many with the GDC, or APP-Sobeya, etc? How many are genuinely independent, and how many simply declined to say? A poll asking about candidate favorability, coalition viability, or succession within a party is, whether it means to be or not, also partly a poll of whoever happened to be reached – and if the people reached already lean heavily toward one party before a single question is asked, the results will lean that way too, regardless of what the rest of the country actually thinks. That isn't dishonesty on anyone's part. It's simply what happens when a sample's political make-up goes unmeasured and, therefore, uncorrected.

This is precisely why party affiliation is normally collected in the first place: not to score political points, but to weight the results – the same basic idea already used elsewhere in this report for gender and geography. A good poll doesn't just collect responses; a good poll checks whether those responses resemble reality. If a national sample happens to contain more supporters of one party than actually exist in the electorate, the final results may show strength for that party – not because public sentiment genuinely shifted, but because that party's supporters happened to speak a little louder in the room. Party affiliation data is what lets a pollster catch that and correct for it before broad national conclusions get drawn. Without it, nobody – Gambia Participates included can tell the difference between a real shift in the country's mood and one political tribe simply being overrepresented on the day the enumerators went out.

𝐖𝐡𝐞𝐫𝐞 𝐢𝐭 𝐛𝐢𝐭𝐞𝐬 𝐡𝐚𝐫𝐝𝐞𝐬𝐭 𝐢𝐧 𝐭𝐡𝐢𝐬 𝐬𝐮𝐫𝐯𝐞𝐲: The clearest example isn't hypothetical – it already happened, and the irony is worth sitting with. Round 1 asked exactly this kind of question: who should succeed Ousainou Darboe as the UDP's presidential candidate? Forty-four percent of respondents named Talib Bensouda, putting him well ahead of every other name on the list.

Here's the twist. Talib never received that succession. In September 2025, amid mounting internal tension, he withdrew from the UDP's actual flagbearer contest and resigned his post as the party's National Organising Secretary. Within weeks, he had broken away entirely to build his own political vehicle, the party now known as UNITE, under which he is contesting the presidency as an outsider to the UDP rather than as the man the party chose. Some commentators have linked that exit to frustrated national ambition; Bensouda himself has publicly rejected that reading. Whichever account persuades you more, the sequence of events makes the underlying point better than any hypothetical could: 44% of a general public sample wanted Bensouda leading the UDP into 2026. The UDP's own internal process did not produce that outcome, and the man the wider polls outcome preferred is "no longer in the party at all".

A question about internal succession, answered largely by people who may never have set foot inside that party's own decision-making process, told us almost nothing about what would actually happen inside it.

A survey that already leans toward one party before it asks a single question will keep leaning that way in every answer that follows – and without measuring the lean, nobody can tell how far it's tilting the floor, whether an angle of elevation or depression. Don't tell me you forgot basic math too soon.

𝐓𝐨 𝐛𝐞 𝐟𝐚𝐢𝐫 – 𝐭𝐡𝐢𝐬 𝐢𝐬 𝐚 𝐠𝐚𝐩, 𝐧𝐨𝐭 𝐚𝐧 𝐚𝐜𝐜𝐮𝐬𝐚𝐭𝐢𝐨𝐧!
Let me be absolutely clear about what this section is and isn't saying. The absence of party affiliation data does not mean this poll is wrong or right. It does not mean the survey was biased or fair. What it means is that a powerful layer of transparency is missing – and transparency is what builds confidence. If future rounds included a simple party-identification question, political parties, journalists, researchers, and ordinary citizens would all gain a clearer understanding of why certain results emerged, instead of arguing past each other about who's right. That is good for democracy, good for public trust, and good for polling itself.

One simple addition would strengthen every future round of this survey considerably: "Which political party, if any, do you currently identify with most closely?" The answer wouldn't need to determine the results. It would simply help explain them – because knowing what people think is only half the story. Knowing who's doing the thinking is the other half, and in political polling, that context is often what turns a column of percentages into a story anyone can actually trust.

5. 𝐖𝐡𝐲 "𝐓𝐚𝐥𝐢𝐛 𝐋𝐞𝐚𝐝𝐬 𝐕𝐨𝐭𝐞 𝐈𝐧𝐭𝐞𝐧𝐭 𝐢𝐧 𝐁𝐫𝐢𝐤𝐚𝐦𝐚 (𝐖𝐞𝐬𝐭 𝐂𝐨𝐚𝐬𝐭) 𝐚𝐧𝐝 𝐊𝐚𝐧𝐢𝐟𝐢𝐧𝐠" 𝐈𝐬𝐧'𝐭 𝐚𝐬 𝐂𝐞𝐫𝐭𝐚𝐢𝐧 𝐚𝐬 𝐈𝐭 𝐒𝐨𝐮𝐧𝐝𝐬.
When a Poll Tries to Measure a Giant Region Using a Handful of Interviews, Extraordinary Claims Require Extraordinary Evidence. Of all the findings in this poll, none has generated more public debate than the claim that Talib Bensouda leads vote intention in both Kanifing and Brikama. For supporters of UDP, NPP, APP-Sobeya, and even politically neutral observers, the reaction has been largely the same: "How exactly did we get here?"
So, let us look at it properly, from two directions – THE MATHS PROBLEM, and then THE POLITICS ON THE GROUND – because both point the same way, and together they say more than either one does alone.

𝐓𝐡𝐞 𝐌𝐚𝐭𝐡 𝐏𝐫𝐨𝐛𝐥𝐞𝐦:: Margin of error shrinks with the square root of sample size, not with sample size itself – which is why splitting a national sample into eight regional subsamples multiplies the uncertainty around every regional number far beyond the ±2.5% margin that applies to the national topline of 1,556. While the national sample sits within an acceptable margin of error, the uneven representation of regional numbers tells a different story. Once N is "large enough" relative to n, the population size stops being a margin-of-error problem and becomes something else entirely: a representativeness problem, which is really the more important question in this whole project, and exactly what the LGA-weighting, gender-skew, and party-affiliation sections are actually about. A sample can have a tiny, mathematically negligible margin of error and still be badly wrong if it isn't drawn evenly across the population it claims to describe.

In Brikama, the poll spoke to roughly 373 people. In Kanifing, roughly 265. When a sample gets that small or missed the quality representation, the built-in wobble every survey carries, statisticians call it the margin of error – gets noticeably bigger. In Brikama, that wobble is roughly plus-or-minus 5 points. In Kanifing, it's closer to plus-or-minus 6.

So, when the poll shows Bensouda at 24%, Barrow at 23%, Darboe at 20% in Brikama, and Bensouda at 27% against Barrow's 24% and Darboe's 21% in Kanifing, those gaps are sitting comfortably inside or right at the edge of a margin of error that size. In plain terms: the poll cannot actually tell you, with any confidence, who is really ahead in either region. It can tell you it's close. It cannot tell you who wins the coin toss. And these are also the two LGAs where the sample's geographic weighting drifts furthest from the real electorate – a second, independent reason the regional numbers here deserve more caution than most.

𝐓𝐡𝐞 𝐏𝐨𝐥𝐢𝐭𝐢𝐜𝐚𝐥 𝐏𝐫𝐨𝐛𝐥𝐞𝐦 – because the Math isn't the only reason to pause: Here's what the statistics alone don't tell you, and what deserves to be said plainly: Talib Bensouda's party has almost no track record in either of these two regions, and in one of them, none at all.

UNITE began life as a movement in "October" 2025, when Bensouda first broke from the UDP. It did not become a formally constituted political party until its national congress in "May" 2026 – roughly three months before Gambia Participates went into the field. In its entire existence, UNITE has never appeared on a single ballot, anywhere in the country. It holds no seats in the National Assembly, no chairmanships, no ward seats – nothing, because there has been no election for it to contest since it came into being.
Brikama, meanwhile, is not an unclaimed territory. Its Area Council is chaired by Yankuba Darboe of the UDP, who won that seat in 2023 with 77,946 votes against the NPP's 52,429 – one of the largest margins recorded in that entire election cycle.

The NPP, for its part, is the sitting national government and came a clear second in that same race. Two established, previously victorious political machines already occupy the ground this poll credits a three-month-old party with leading. What campaign infrastructure, ward-level organisation, or prior result does UNITE have in Brikama that would explain overtaking both of them? Nothing has been publicly documented that answers that question.

Kanifing deserves a more careful, separate look, because it isn't quite the same story. Bensouda himself does have a real personal electoral history there – he is the sitting Mayor, twice elected. But it's worth being precise about what that mandate actually is: he won it running as a UDP candidate in 2023, a full two years before UNITE existed. Whether his personal popularity in Kanifing genuinely transfers to his new party is a reasonable, testable question – plausible in a way it simply isn't in Brikama, where neither Bensouda nor UNITE has ever stood for anything at all.

A one-point statistical tie is easy to explain. A one-point statistical tie for a party that has never won a single seat, anywhere, in a region held by two rival machines with real track records and 2023 victories behind them, needs a better explanation than "margin of error" alone.

In theory, it is entirely possible for a new movement to grow fast; it has happened elsewhere in the world. But "possible" is not the same as "established," and right now this specific finding – Brikama especially doesn't yet have an obvious real-world explanation to sit alongside the statistical one. That combination – a razor-thin margin on a small subsample, sitting on top of a region with no visible party organization to back it up – is exactly the kind of result that deserves a second look, not a headline.

So, with all of this stacked up, how does anyone responsibly conclude the poll numbers in Brikama and Kanifing are simply correct? Honestly – not yet, and not with full confidence. That doesn't mean they are wrong or right. It means this is precisely the kind of regional finding that needs a larger regional sample on a relatable benchmark, a second round to see if it holds, and the level zero data point this analysis has already asked Gambia Participates to publish – before anyone, on any side, treats it as more than an open question.

Going further and declaring him the clear leader of both regions may be asking the survey to tell us more than it is statistically capable of telling. Because when sample sizes are small or fall short of quality representations and margins of error are large per LGA, the wisest analysts know the difference between a signal and a certainty. And in politics, confusing the two has led to many surprising election nights.

6. 𝐓𝐡𝐞 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧 𝐁𝐮𝐫𝐢𝐞𝐝 𝐢𝐧 𝐏𝐥𝐚𝐢𝐧 𝐒𝐢𝐠𝐡𝐭: 𝐖𝐡𝐚𝐭 𝐃𝐨 𝐆𝐚𝐦𝐛𝐢𝐚'𝐬 𝐖𝐨𝐫𝐤𝐞𝐫𝐬 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐓𝐡𝐢𝐧𝐤?
Here's something worth sitting with. Buried in the socio-demographic tables at the front of this survey, Gambia Participates tells us exactly what everyone in the sample does for a living: 19% work in the public sector, 17% in the private sector, 2% for NGOs, 23% in the informal sector, and 39% are unemployed. That's a genuinely useful thing to measure.

And then the report never or barely mentions it again.
Every major question in this survey – should Barrow contest a third term, would you vote for Darboe, who's leading where – gets broken down by gender. It gets broken down by age group. It gets broken down region by region, LGA by LGA. But not once, anywhere in either round of this poll, is a single political question broken down by employment sector. The data was collected. It was simply never read back to us.

Think about what that means in practice. Imagine a doctor who takes your blood pressure, your temperature, and your weight during a check-up, then hands you a report that only discusses your temperature. The other two readings sit in the file, measured and paid for, but never actually examined. You walk away with half a diagnosis and no idea whether the parts nobody mentioned were the parts that mattered most.

𝐖𝐡𝐲 𝐭𝐡𝐢𝐬 𝐩𝐚𝐫𝐭𝐢𝐜𝐮𝐥𝐚𝐫 𝐠𝐚𝐩 𝐬𝐡𝐨𝐮𝐥𝐝 𝐛𝐨𝐭𝐡𝐞𝐫 𝐲𝐨𝐮: Add unemployed and informally employed respondents together, and you get 62% of this entire sample – nearly two out of every three Gambians who answered this survey. That's not a minor subgroup. That's the majority of the country the poll actually spoke to. And this isn't a side issue either: the same survey found, in its own numbers, that "the economy and cost of living" is the single biggest factor shaping how Gambians say they'll vote, named by 76% of respondents, comfortably ahead of every other concern.

If economic anxiety is genuinely driving this election more than any candidate's name or any party's colours, then knowing whether a government salary, a market stall, or no fixed income at all changes how someone views the incumbent isn't a nice-to-have. It is arguably the single most relevant cross-tab this survey could have run – and it's the one nobody ran.

Here is what we are left not knowing, that we could have known:
Do public sector workers – whose salaries and job security flow directly from the government of the day view President Barrow's record differently from everyone else?

Does the unemployed 39% of this country, living most directly with the consequences of a struggling economy, break toward the incumbent, toward the opposition, or toward the undecided column in numbers different from the rest of the sample?

Do informal traders – the market stalls, the bike riders, the roadside vendors who make up 23% of respondents respond to "cost of living" messaging in a way that's measurably different from salaried private-sector workers?

Does the small NGO-employed slice of the sample, often the group most exposed to civil-society and governance-reform conversations, lean in a direction the rest of the country doesn't?

𝐓𝐨 𝐛𝐞 𝐟𝐚𝐢𝐫 – 𝐭𝐡𝐢𝐬 𝐢𝐬 𝐚 𝐦𝐢𝐬𝐬𝐞𝐝 𝐨𝐩𝐩𝐨𝐫𝐭𝐮𝐧𝐢𝐭𝐲 – 𝐚𝐧𝐝 𝐰𝐞 𝐡𝐨𝐩𝐞 𝐢𝐭 𝐰𝐚𝐬 𝐧𝐨𝐭 𝐝𝐞𝐥𝐢𝐛𝐞𝐫𝐚𝐭𝐞!
None of this suggest that Gambia Participates buried something deliberately, I have no evidence to suggest that. It is far more likely this is simply what happens when a demographic question gets asked out of good survey practice, and then nobody circles back to actually use it. But the effect is the same either way: a genuinely important lens on this election – how Gambians in different economic positions relative to the state see their choices – was collected and then left on the shelf.

It wouldn't take a new question or a bigger sample to fix. The raw material already exists in GP's own dataset. It would simply take running the cross-tab that is already sitting there: candidate support, third-term approval, and vote intention, broken down by public sector, private sector, NGO, informal, and unemployed – exactly the way gender, age, and region already are. Given that the economy is what this survey itself says will decide the election, that's not a detail. It's arguably the missing centerpiece.

7. 𝐖𝐡𝐚𝐭 𝐏𝐞𝐨𝐩𝐥𝐞 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐓𝐨𝐥𝐝 𝐭𝐡𝐞 𝐏𝐨𝐥𝐥𝐬𝐭𝐞𝐫𝐬 𝐓𝐡𝐞𝐲 𝐂𝐚𝐫𝐞 𝐀𝐛𝐨𝐮𝐭:
Buried under all the horse-race numbers is, in my view, the most human and least talked about part of this entire survey: what Gambians say will actually determine their vote.
(i) Economy and cost of living – 76% of respondents named it as a factor shaping their vote.
(ii) Security and national stability – 55% of respondents named it as a factor shaping their vote.
(iii) Healthcare – 50% of respondents named it as a factor shaping their vote.
(iv) Electricity and water supply – 35% of respondents named it as a factor shaping their vote.
(v) Employment – 25% of respondents named it as a factor shaping their vote.
(vi) Infrastructure development – 18% of respondents named it as a factor shaping their vote.
(vii) Corruption – 29% of respondents named it as a factor shaping their vote.

Notice what tops that list. Not a candidate's name. Not a party color. The price of rice, oil, and transport. That is worth remembering every time a headline reduces this election to a two- or three-man horse race – the electorate itself is telling us, in its own words, that this contest will be decided on kitchen-table issues at least as much as on personality.

8. 𝐄𝐱𝐩𝐞𝐜𝐭𝐢𝐧𝐠 𝐒𝐨𝐦𝐞𝐨𝐧𝐞 𝐭𝐨 𝐖𝐢𝐧 𝐈𝐬 𝐍𝐨𝐭 𝐭𝐡𝐞 𝐒𝐚𝐦𝐞 𝐚𝐬 𝐖𝐚𝐧𝐭𝐢𝐧𝐠 𝐓𝐡𝐞𝐦 𝐓𝐨:
Think about the weather forecast for a moment. When the forecaster says there's a 70% chance of rain tomorrow, that is not a guarantee – and it is definitely not the forecaster telling you they want it to rain. It's a statement about likelihood, built from patterns, not a wish or a verdict.
GP's Round 2 poll asked two separate questions, and the gap between the answers is, in my professional opinion, the single most under-discussed number in this debacle.
(i) Adama Barrow: would vote for him today 32%, expected to win regardless of preference 51% (Gap: +19)
(ii) Ousainou Darboe: would vote for him today 25%, expected to win regardless of preference 23% (Gap: -2)
(iii) Talib Bensouda: would vote for him today 19%, expected to win regardless of preference 12% (Gap: -7)
A 19-point gap for Barrow between "I expect him to win" and "I would vote for him" tells us something specific and, frankly, encouraging for anyone who feels discouraged this week: roughly one respondent in five believes Barrow is likely to win without saying they want him to. That is a description of incumbency, visibility, and name recognition doing their normal work – not a description of consolidated, committed support. Weather forecasters call this the difference between a forecast and an outcome. Election watchers should call it the same thing.

9. 𝐇𝐨𝐰 𝐘𝐨𝐮 𝐂𝐚𝐧 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐔𝐬𝐞 𝐓𝐡𝐢𝐬 𝐏𝐨𝐥𝐥 – 𝐖𝐢𝐭𝐡𝐨𝐮𝐭 𝐋𝐨𝐬𝐢𝐧𝐠 𝐒𝐥𝐞𝐞𝐩 𝐎𝐯𝐞𝐫 𝐈𝐭
None of the above means you should throw this survey in the bin. A well-read poll, warts and all, is genuinely useful – you just have to know what it's good for and what it isn't.

𝐔𝐬𝐞 𝐢𝐭 𝐟𝐨𝐫 𝐭𝐡𝐢𝐬:
(i) To know which issues to press every candidate on when they come canvassing – the economy, security, and healthcare, in that order, are what your neighbours actually said matters most.
(ii) To identify the genuinely competitive areas – Kanifing, Brikama, and Janjanbureh in particular – where a small extra push in turnout could plausibly change the outcome, and where your individual vote carries unusually real weight.
(iii) To see, in black and white, that no candidate is close to a majority of declared support – which means this election is still entirely open, whatever any single headline claims.
(iv) To track a candidate's or party's own promises over time, by comparing this round against the next – opinion polling, done responsibly across repeated rounds, is one of the few tools ordinary citizens have to hold public figures accountable to the public mood.
(v) To start a conversation with your family, your workplace, your compin – not to end one. A poll is a prompt for debate, not a substitute for it.

𝐃𝐨 𝐧𝐨𝐭 𝐮𝐬𝐞 𝐢𝐭 𝐟𝐨𝐫 𝐭𝐡𝐢𝐬:
(i) As a reason to stay home on election day because "it's already decided." It isn't – not even close, on this poll's own numbers.
(ii) As proof that a specific regional result is settled, especially in Brikama, Kanifing, or any LGA where the sample size is small enough that the margin of error swallows the reported gap.
(iii) As a verdict on any coalition or alliance – several of the exact names tested in this survey had already changed political camps before the report was even published.
(iv) As a reason to feel your preferred candidate has "lost" or "won" anything. Nobody has won or lost anything yet. That happens once, on 5th December 2026, at the ballot box – not in a survey in August.

10. 𝐘𝐨𝐮 𝐀𝐫𝐞 𝐍𝐨𝐭 𝐀𝐥𝐨𝐧𝐞 𝐢𝐧 𝐃𝐨𝐮𝐛𝐭𝐢𝐧𝐠 𝐚 𝐏𝐨𝐥𝐥:
If you feel like polls have burned you before, you have good company, and good reason – well beyond our own borders.
(i) In the 2016 US presidential election, national polls had Hillary Clinton ahead by roughly 3 points. Donald Trump won the Electoral College.
(ii) In the 2016 UK Brexit referendum, polls pointed to a narrow "Remain" victory. "Leave" won 52 to 48.
(iii) In the 2020 US election, national polls overstated Joe Biden's eventual margin by several points once every vote was counted.
(iv) In the 2024 US election, polls showing a dead heat broke more decisively than the averages suggested once the count was final.

None of those countries – with far bigger polling budgets and far longer polling traditions than ours – stopped voting because their polls were imperfect. None of their citizens should have, and neither should you. The lesson, repeated on every continent, in poor countries and rich ones alike, is the same: a poll informs guesswork about a moving target. It is not a prophecy, and it can never designed to be one.

11. 𝐓𝐡𝐞 𝐄𝐬𝐬𝐚 𝐅𝐚𝐚𝐥 𝐌𝐲𝐬𝐭𝐞𝐫𝐲: 𝐇𝐨𝐰 𝐃𝐨𝐞𝐬 𝐓𝐡𝐞 𝐋𝐚𝐬𝐭 𝐏𝐨𝐥𝐥’𝐬 𝐏𝐨𝐥𝐢𝐭𝐢𝐜𝐚𝐥 𝐇𝐞𝐚𝐯𝐲𝐰𝐞𝐢𝐠𝐡𝐭 𝐃𝐢𝐬𝐚𝐩𝐩𝐞𝐚𝐫 𝐖𝐢𝐭𝐡𝐨𝐮𝐭 𝐋𝐨𝐬𝐢𝐧𝐠 𝐚 𝐅𝐢𝐠𝐡𝐭?

Imagine this. One year ago, a candidate was recording some of his strongest numbers for perceived win in the two most important electoral battlegrounds in the country. In Brikama, he posted 18%. In Kanifing, he reached 21%. These are not obscure villages hidden deep in the countryside. These are the very places where elections are won and lost. Fast forward twelve months. The same candidate by the same pollster now appears almost invisible on the political radar.

In Brikama, his number reportedly falls from 18% to 2%. In Kanifing, it falls from 21% to 0%. Read that again. Not 21% to 11%. Not 18% to 8%. But from a serious contender territory to almost a statistical extinction. For anyone who loves numbers, that is not merely surprising. It is extraordinary. The first duty of a data scientist is not to shout "conspiracy." The first duty is to ask: "Does this result make logical sense?"

And this is where things become fascinating. Political support is not like water spilled on the ground. It does not simply evaporate overnight.
In the politics of science, voters usually move for reasons.
A major scandal.
A crushing electoral defeat.
A public fallout.
A catastrophic policy failure.
A complete withdrawal from public life.
Yet in the case of Essa Mbye Faal, one can reasonably argue that the opposite occurred.

His Sobeya movement merged into APP-Sobeya, creating a larger organisational structure with a wider national footprint. Normally, political mergers are intended to increase visibility, strengthen networks, and expand support, not erase them. That alone raises an important question: "Where did all those voters go?" Even if every single one of his supporters became disappointed, it would still be difficult to explain such a dramatic collapse mathematically.
What is more likely? That support weakened? That support fragmented? That support migrated toward coalition politics? But that an individual with significant national visibility suddenly became politically irrelevant in the polls across the country's largest voting regions? That requires a much larger explanation than a single headline number can provide.

Then comes the political twist. Only days ago, Essa Faal appeared at the UDP's 30th Anniversary celebration and received one of the warmest receptions of the day. Formerly unimaginable political combinations suddenly began entering public discussion. No formal alliance exists. Nobody should invent one. But politics is the art of possibilities. And if Gambian politics has taught us anything, it is this: "Today's opponent often becomes tomorrow's ally."

𝐖𝐡𝐢𝐜𝐡 𝐥𝐞𝐚𝐯𝐞𝐬 𝐮𝐬 𝐰𝐢𝐭𝐡 𝐨𝐧𝐞 𝐮𝐧𝐚𝐯𝐨𝐢𝐝𝐚𝐛𝐥𝐞 𝐜𝐨𝐧𝐜𝐥𝐮𝐬𝐢𝐨𝐧: Writing Essa Faal out of the political story today may prove to be one of the most expensive analytical mistakes anyone can make. Politics has a strange habit of resurrecting people who newspapers, pundits, and polls have already buried.

12. 𝐓𝐡𝐞 𝐃𝐨𝐮𝐛𝐭𝐬, 𝐋𝐚𝐢𝐝 𝐁𝐚𝐫𝐞 – 𝐑𝐞𝐚𝐝 𝐓𝐡𝐢𝐬 𝐏𝐚𝐫𝐭 𝐓𝐰𝐢𝐜𝐞:
Let me put every reasonable doubt about this survey in one place, plainly, so nobody has to go digging for reasons to keep hope alive:
(i) The two largest, most consequential LGAs in the country – holding 56% of registered voters – were under-represented in the sample by 15 percentage points, which alone is enough to move the national topline by two to three points once corrected.
(ii) The sample remains more male (59%) than the country actually is (roughly 49% male, 51% female, per the 2024 national census) – with no disclosed correction for that imbalance.
(iii) The coalition question tested a grouping that had already begun falling apart, in the real world, during the very days the interviews were being conducted.
(iv) At the regional level, several of the closest "leads" reported – including in Brikama and Kanifing – sit well inside the survey's own statistical margin of error, meaning the poll itself cannot confidently say who is ahead there.
(v) Seventeen percent of respondents remain undecided on current vote intention – more people than voted for third place in the last presidential election, and more than enough, on their own, to decide this one.
(vi) No breakdown of party affiliation was collected, making several of the coalition and succession questions difficult to interpret with confidence.
Every single one of these doubts points the same direction: toward an election that is genuinely, mathematically open – not toward a result you should treat as settled, whichever side of the contest you are on.

13. 𝐖𝐡𝐚𝐭 𝐏𝐨𝐥𝐢𝐭𝐢𝐜𝐚𝐥 𝐏𝐚𝐫𝐭𝐢𝐞𝐬 𝐒𝐡𝐨𝐮𝐥𝐝 𝐑𝐞𝐚𝐥𝐥𝐲 𝐋𝐞𝐚𝐫𝐧 𝐅𝐫𝐨𝐦 𝐓𝐡𝐢𝐬 𝐏𝐨𝐥𝐥:
"THE MOST EXPENSIVE SOUND IN POLITICS IS THE SOUND OF GUESSWORK"

Every election season produces the same spectacle. Candidates hold rallies. Parties print banners. Supporters flood social media. Everyone claims victory. Everyone insists the people are behind them. Everyone says the ground is strong. But then election day arrives. And reality introduces itself.

The uncomfortable truth is that many political campaigns spend months talking to voters without ever truly listening to them. That is like trying to sail a boat across the Atlantic using confidence instead of a compass.

The most successful political organisations in the world do not rely on rumours. They rely on data.
They do not ask: "How do we feel?" They ask: "What do the numbers actually say?"

Real political intelligence answers questions that rallies cannot: Where are your strongest voters? Where are you losing ground? What issue matters most to undecided citizens? Which communities are shifting away from you? Which message works? Which message fails? Where is your campaign wasting money? Where should you focus the next thirty days?

These are not emotional questions. They are analytical questions. And analytical questions require analytical answers. That is why serious political organisations around the world invest heavily in internal polling, demographic mapping, sentiment analysis, behavioural research, and voter forecasting. They understand something that too many campaigns learn too late: The purpose of data is not to make you feel good. It is to prevent unpleasant surprises.

A good internal poll does not tell a candidate what they want to hear. A good internal poll tells a candidate what they need to hear. Sometimes that truth is encouraging. Sometimes it is painful. But it is always useful.

At DataEd and with Touba, that is exactly the service we provide. Not propaganda. Not wishful thinking. Not partisan cheerleading. Evidence, and evidence only!

𝐖𝐡𝐚𝐭 𝐭𝐡𝐚𝐭 𝐥𝐨𝐨𝐤𝐬 𝐥𝐢𝐤𝐞 𝐢𝐧 𝐩𝐫𝐚𝐜𝐭𝐢𝐜𝐞:
Independent political intelligence
Scientific analysis
Strategic voter insights
Regional support assessments
Undecided voter behaviour
Turnout modelling
Campaign diagnostics

And perhaps most importantly: an honest mirror. The greatest mistake in politics is not losing an election. The greatest mistake is losing an election when the warning signs were visible months earlier and nobody bothered to measure them.

As The Gambia moves closer to December 2026, every serious political party should ask itself one question: are we running our campaign on evidence, or are we running it on assumptions?
Because assumptions make noise. Data wins strategy.

14. 𝐖𝐡𝐚𝐭 𝐈 𝐀𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐖𝐚𝐧𝐭 𝐘𝐨𝐮 𝐭𝐨 𝐓𝐚𝐤𝐞 𝐟𝐫𝐨𝐦 𝐓𝐡𝐢𝐬 𝐏𝐨𝐥𝐥:

If you support Barrow and you are feeling good this week – good. Stay engaged, and go vote anyway, because 32% is not 51%, and half the country expecting him to win is not the same as half the country choosing him.

If you support Darboe or another candidate and you are feeling discouraged – please don't be. A properly weighted read of this same data narrows the gap, not widens it. 17% of the country told this survey they haven't decided anything yet. Undecided voters are not a footnote in this story, in a race this close, they are the story.
And if you belong to neither camp and you are simply tired of all of it – I understand that too. But I will leave you with the same words I will close every one of these pieces with, because they are the truest sentence I know how to write about any election poll, anywhere in the world: Percentages don't vote. Snapshots don't vote. Headlines don't vote. People do.

The 962,157 plus Gambians who can walk into a polling station on 5th December hold more power in their collective hand than any 1,556-person clipboard exercise ever could. A poll can describe a mood. It cannot cast a ballot. Only you can do that.

History is full of polls that missed the mark and elections that defied expectations. Let us therefore treat surveys with interest, not worship; with curiosity, not panic. The final chapter of this story has not yet been written, and the author will be the Gambian voter.
Fellow Gambians, let's not promote a survey to the position of the IEC Chairman. Polls can inform us, entertain us, and sometimes even surprise us, but they will not vote on December 5th. So, relax your shoulders, keep your blood pressure where your doctor wants it, keep asking sharp questions of every number anyone hands you, mine included, and when December comes, show up in your numbers. That is the only survey whose margin of error is exactly zero.

The future of our country will not be written by a pollster's pen, but by the collective voice of the Gambian people. May the best ideas win, and may The Gambia come out stronger than ever.

𝐈𝐟 𝐲𝐨𝐮 𝐰𝐨𝐮𝐥𝐝 𝐥𝐢𝐤𝐞 𝐭𝐨 𝐭𝐚𝐥𝐤, 𝐩𝐥𝐞𝐚𝐬𝐞 𝐄𝐦𝐚𝐢𝐥: 𝐓𝐨𝐮𝐛𝐚𝐌𝐚𝐫𝐫𝐢𝐞@𝐃𝐚𝐭𝐚𝐄𝐝.𝐨𝐫𝐠

𝐓𝐡𝐚𝐧𝐤 𝐲𝐨𝐮.
𝐓𝐨𝐮𝐛𝐚 𝐌𝐚𝐫𝐫𝐢𝐞,
𝑫𝒂𝒕𝒂 𝑺𝒄𝒊𝒆𝒏𝒕𝒊𝒔𝒕 & 𝑴𝒂𝒕𝒉𝒆𝒎𝒂𝒕𝒊𝒄𝒊𝒂𝒏, 𝑶𝒓𝒊𝒈𝒊𝒏𝒂𝒍𝒍𝒚 𝒇𝒓𝒐𝒎 𝑭𝒐𝒏𝒊 𝑲𝒂𝒏𝒅𝒐𝒏𝒌𝒖.

And before I forget, as promised in the beginning. If you really want to see the full technical A fuller, more technical version of this same dissection exists too – the one with every calculation shown, every margin of error worked out in full, and enough tables to make a spreadsheet blush. Think of it as the director's cut, for readers who are technical with interest and ability in STEM.

𝐁𝐮𝐭 𝐟𝐚𝐢𝐫 𝐰𝐚𝐫𝐧𝐢𝐧𝐠: entry is not free. The bouncer at that door has exactly one question for you, and he will not accept "I forgot" as an answer.

𝐓𝐚𝐤𝐞 𝐭𝐡𝐞 𝐛𝐚𝐬𝐢𝐜 𝐪𝐮𝐚𝐝𝐫𝐚𝐭𝐢𝐜 𝐞𝐪𝐮𝐚𝐭𝐢𝐨𝐧, 𝐚𝐱² + 𝐛𝐱 + 𝐜 = 0. Now derive, from scratch, the almighty quadratic formula – yes, that one, x equals negative b plus-or-minus the square root of b-squared minus 4ac, all over 2a. Show your working. Complete the square if you must. No calculators, no Google, no calling a cousin who "did well in maths." Yap. You heard me loud and clear.

If you can do that, the technical report is waiting, and it will treat you like the fellow scientist you clearly are. If you can't (and honestly, most of humanity can't, on a random day, without warning), don't panic and don't feel bad. That's exactly why this plainer version exists. Stay right here with the rest of your friends.