Mr. Dankwah, as the lead pollster at GlobalInfo Analytics, your July 2025 National Tracking Poll has a significant influence over public dialogue and decision-making in Ghana. With that power comes a duty to adhere to the highest standards of methodological rigour, transparency, and honesty. Upon deep and objective scrutiny, it is clear your report suffers from a series of systemic flaws that undermine its credibility and utility.
Your sampling design is seriously flawed. The respondent pool is heavily biased, with 58% male and 42% female, and a notable 68% of respondents under 44 years old, and just 1% over 65 years old. This does not accurately represent the actual Ghanaian adult population. You also omit any mention of post-stratification weighting, which is used to adjust the sample to match national census data —a significant oversight in professional survey research. Furthermore, the large proportion of your in-person sample (over 75% of all respondents) dominates the overall results, which could overshadow the more varied opinions from online and telephone respondents.
Your data collection periods are uneven: online (May 10–20), telephone (June 9–25), and in-person (June 19–27). Public opinion can change quickly due to events or news cycles, yet you treat these results as if they are from the same time. The overwhelming number of in-person responses is also worrying, given the well-known “social desirability bias,” where respondents tend to overstate their support for government in face-to-face settings. Your report does not address or adjust for these mode effects.
Nowhere do you publish your complete survey instrument, exact question wordings, or sampling protocols? You do not provide response rates, refusal rates, or details on how randomisation was ensured in online and telephone modes. The absence of these details prevents independent verification and leaves your findings susceptible to suspicion of selection bias and leading questions.
Your reporting is filled with positive framing and the grouping of response categories in ways that inflate approval, such as merging “very good” and “good” into a single metric. This obscures nuance and diminishes dissatisfaction. You routinely ignore or diminish large “no opinion” groups, which could indicate disengagement or scepticism, and downplay outliers like the Ashanti Region, where the public mood is markedly different from the national trend.
You claim a 99% confidence level with a sub-2% margin of error despite uneven sample sizes, non-simultaneous data collection, and clustering. Such statistical certainty is unlikely. Frequent rounding errors, with percentages totalling 99% or 101%, confuse and suggest a lack of rigour in quality control. You fail to declare your funding sources or sponsors, a basic requirement for impartiality. In a politically sensitive context, transparency here is essential.
Mr. Dankwah, your report’s optimistic numbers and confident headlines are not underpinned by methodological discipline. To preserve public trust and the value of your research, I urge you to disclose your full methodology immediately, apply proper statistical corrections, and invite an independent review. Anything less undermines both your work and the public good.
Isaac Ofori
Executive Director,
Research Express Consult
[email protected]
