Dr Alieu SK Manjang
Following the publication of my initial critical analysis of the survey, I promised to present a second part focusing on the sampling method and the issue of external validity. Before doing so, however, it is necessary to address certain concerns regarding the survey’s reliability, particularly in its attempt to project the electoral viability of candidates and voters’ preferences.
For instance, respondents who identified a particular individual as the “best suited to replace Ousainu Darboe as the UDP presidential candidate for 2026” may not necessarily vote for that individual in the actual election, especially when other candidates, who may be their genuine preference, are also in the race. This question would have carried greater weight if it had been limited to self-identified UDP supporters, who are the most likely to cast their votes for the party’s eventual nominee.
To illustrate, many of the favorable responses for Talib or Yankuba as potential successors to Darboe might well have come from NPP supporters or members of other political parties. These individuals, though willing to express a preference in the survey, may not vote for Talib or Yankuba if they become the UDP candidate. By contrast, favorable responses for Rohey (even though low for example) as Darboe’s successor may have come predominantly from genuine UDP supporters, whose votes would translate into actual electoral support. This discrepancy highlights a significant limitation: favorable survey responses do not necessarily predict electoral outcomes, thereby putting the reliability of the survey into question.
Turning to the sampling method, a more troubling issue emerges. In a 31-page report, the authors allocated only a single page to explaining the methodology, which is arguably the most critical component of any scientific study. While results are important, understanding how those results were obtained is even more essential.
The report states that “a stratified random sampling technique was adopted to capture variations in political attitudes across regions. Stratification was based on geographic location (urban/rural) and LGA boundaries to ensure proportional representation.” This raises several methodological questions: Why did the researchers opt for stratified sampling instead of cluster sampling, given the heterogeneous nature of the Gambian population? On what specific traits were the strata defined? Within each stratum, what makes the units comparable? More specifically, which villages and districts were included in the sample, and what similarities were assumed among them?
The survey only presented respondents’ gender, education level, employment status, age, and LGA as variables. Yet, the extent to which these characteristics determine voting preferences in The Gambia is questionable.
In most cases in The Gambia, the voting preferences of a compound, village, or district can be highly predictable in favor of, or against, a candidate, regardless of the gender, level of education, age, or employment status of individual voters. Yet, a neighboring compound in the same village, a nearby village in the same district, or an adjacent district in the same Local Government Area may favour different candidates, also irrespective of those demographic factors.
This indicates two key points that the survey missed. First, there is another characteristic influencing voter behaviour: ethnicity. This explains the variations in candidate popularity across regions, constituencies, and villages, not because of geography itself, but because of the relative dominance of particular ethnic groups. Omitting ethnicity from the survey weakens the explanation of the results.
Second, this highlights the heterogeneous composition of regions, constituencies, and villages. It suggests that cluster sampling may offer a statistical advantage over stratified sampling for capturing Gambians’ voting attitudes. For example, a random selection of ten respondents in Basse, Kombo South, or Old Yundum constituencies may not guarantee equal representation of all resident ethnic groups. The same concern applies to the random selection of wards and villages. Ignoring this issue calls into question the external validity of the survey results.
Furthermore, since the survey was not conducted anonymously at the level of geographical locations, it would have been better to present the names and numbers of constituencies, villages selected , as there was no concern of confidentiality. This would also help explain the implications of the results for each party and candidate, given that each has identifiable strongholds. Limiting the presentation to the number of respondents in each Local Government Area is insufficient, especially since only a small portion of the population in these areas was sampled.
In conclusion, while such surveys are valuable in The Gambia, their credibility depends on methodological rigor and transparency. Researchers must consider the diverse interests of their audience: while some readers focus on the results, others prioritize the methodology. This requires careful choice of methods, thorough explanation of procedures, and openness to self-criticism to preempt questions of validity. Finally, highlighting these concerns does not invalidate earlier surveys conducted by the same institution. My critique is specific to the present survey and should be understood in that context.
