I am very happy to see Mussa Dankwah get the attention, recognition, and respect that he deserves as a pollster, researcher and analyst. As one who heads an organisation that generates and relies on evidence and data to inform its advocacy policy and civic engagement,
I started to take a keen interest in Mussa’s work in connection with the 2024 Ghana elections after I heard him deliver a presentation sometime in April or May this year on the findings of the polling he had conducted at the time of the 2024 elections.
His data showed Mahama and the NDC beating Bawumia and the NPP handily even at that early stage of the season. When, during the Q&A, I was invited to comment on Mussa’s findings, I noted that, though the projected margin of victory appeared to me unusually high, I had no data or evidence to contradict his findings, as CDD had not yet conducted its Afrobarometer (“AB”) Ghana Round 10 survey.
Nonetheless, I added that, based on our AB Round 9 data from 2022 (where among respondents indicating their voting intentions most, for the first time, picked NDC over NPP) and the regions where Mussa’s data showed NDC significantly outperforming NPP, notably Greater Accra, Central and Western Regions, I found Mussa’s findings credible.
In August, CDD went to the field to collect data for our AB Round 10. The survey’s findings were released in October. Once I saw our own data, my mind went back to Mussa’s presentation earlier in the year.
Our AB data, though not a pre-election survey as such, contained more than enough information for one to draw firm conclusions as to where the elections were headed. And they generally and strongly confirmed what Mussa had found much earlier. I began to follow with keener interest, in his regular poll updates on social media.
Sometime in early November, I reached out to him privately so we could chat more directly about his polling. Since his polling was more granular and focused specifically on elections, they provided additional valuable ‘apɔ’ for curious minds like me. For example, I was interested to know how the parties and candidates stood in certain specific regions and constituencies, the kind of detail I could get only from Mussa’s work.
As D-Day drew to a close, some of the details he was sharing from his weekly and, in some cases, daily polling were just astonishing. If I was a betting man and had bet on any of his data regarding which constituencies NDC would flip, I would have lost badly. He got everything right that I was unsure of, including that, the NDC and Mahama would have a near-perfect sweep from coast to coast, that my own Tema West constituency would flip, and that the NDC would be exceptionally competitive in Ashanti and make nonsense of NPP’s voter turnout and vote margin or percentage targets.
My conversations with him were very enriching for me. And he was funny too, especially when I brought in other people’s “research” or commentary that painted a contrary picture. When I asked that his projected margin of victory for Mahama seemed exceptionally wide, his reply was, “Prof, these are not normal times.”
How could I disagree? (I have attached excerpts of some of our chats from a week or two or a few days before the elections). In all, I got a richer understanding of his methodology and came to appreciate both his technical competence and his integrity. Now that he’s been vindicated, I can only say to him, “Ayekoo”!
There is a second person whose analytical model predicting the outcome of the December 7 elections has received far less publicity but who also got it dead right! He’s Kwaku Antwi-Boasiako, one of the thoughtful and analytical sharp minds on Facebook.
Unlike Mussa, KAB is not a pollster. Kwaku’s methodology could be described, roughly, as a multi-variable analysis of the key factors or variables around which an election would turn, with each variable assigned a weighting based on its perceived importance to voters in the election. I’m sure Kwaku can do a much better job than I of describing his #GhPVModel – Presidential Election Prediction Model.
Kwaku first launched his tool in 2020, as a presidential election prediction model to predict Ghana’s 2020 presidential elections–and all future presidential elections. For the 2020 elections, his model predicted 47.07% for the NDC and 51.14% for the NPP in the presidential contest. The actual presidential results were: NDC 47.36%; NPP 51.30%.
Here, click on this link to see KAB’s predictions
On 17th November 2024, KAB’s model, which he shared privately with me and several other selected individuals, predicted 51.25% for the NDC and 46.67% for the NPP. He had shared an earlier version of the model with me in March 2024, to roughly the same effect. You can still find his March analysis on #GhPVModel. (He didn’t share his November 17 results publicly because of the sensitive nature of the work he was doing for a client at the time). In his earlier analysis, he also disaggregated his results by region, and it showed NDC winning in all regions but Ashanti and Eastern Region.
An important difference between Mussa’s work and KAB’s is that, because Mussa does granular level polling about voters’ voting intentions, using statistical sampling tools, he is able to predict not just who will win but also by how wide or small a margin. KAB’s is an analytical tool based on desk research, that focuses on the issues driving the elections and how the two leading contestants are likely to fare on each of those weighted issues, using both historical and current secondary data.
KAB’s work also helps us predict who will win; the margin of victory or defeat is not the key thing in his prediction, though he does supply an educated estimate. I recall that, sometime in November, when I shared KAB’s Model with Mussa, his comment, based on the fact that KAB’s Model showed a win for Mahama but only by less than 52%, was that “that would be the worst case scenario for JDM.”
KAB also fielded many questions from me about his methodology. As D-Day got close, and the disinformation campaigns heated up, I queried whether and to what extent disinformation and misinformation could affect his projections. He was sure they wouldn’t have much of an impact. He thought they would cancel each other out.
It would be interesting to see, now that the elections are over, what Mussa’s exit polls tell us about what voters said influenced their voting decisions and how these match up against KAB’s weighted variables.
In a country where political types would rather consult occultists, prophets and mallams about their electoral prospects or fortunes, Mussa’s and KAB’s work, as well as the success of their respective approaches, are an important victory for the use and superiority of scientific methods and analysis in understanding voter intentions, behaviour and dynamics. I hope that we use these avenues and resources more, not just to improve the quality of our electioneering, but to enhance both the quantity and quality of data and evidence that inform our policymaking and public decision-making in general.
#PowerIsNotKnowledge. Pay attention!