Opinion polls and 2023 presidential contest: Prediction versus reality


By Gbenga Salau

In the build up to the 2023 general elections, several opinion polls were conducted to give insight into the likely outcome of the elections, especially the February 25, 2023 presidential elections.

A list of the organisations that conducted the opinion polls includes the Nigerian Human Rights Community (NHRC), Political Africa Initiative (POLAF), NOI Polls Limited, Stears, SBM Intelligence, Bloomberg, Fitsch Solutions, Country Risk Research, Economist Intelligence Unit (EIU) and June Group Research and Council for African Security Affairs (CASA).

Although the Independent National Electoral Commission (INEC) had declared Bola Ahmed Tinubu of the All Progressives Congress (APC) winner of the election, a number of the opinion polls conducted by local and international organisations predicted that Peter Obi and Waziri Atiku Abubakar could win the election.

For instance, while Bloomberg, Stears, and New Nextier predicted through its survey that Labour Party’s candidate, Mr. Peter Obi might win, Fitsch Solutions and Country Risk Research, Nigerian Human Rights Community (NHRC) as well as the Economist Intelligence Unit (EIU) projected Tinubu could win the election. This is also just as opinion polls by Political Africa Initiative (POLAF), June Group Research and Council for African Security Affairs (CASA) gave the forecasts about the Peoples Democratic Party (PDP) candidate, Atiku coming out victorious.

Of note, however, was that each time the opinion polls and survey outcomes were made public, they were contentious, especially as the candidates whom the result did not favour often rejected the findings, claiming it was not a true reflection of the will of the people.

Each time the opinion poll’s results were released, conductors of the surveys are often quick to note that the predictions are not absolute, as there are still dynamics and indices that could alter the forecast before and on the Election Day. POLAF in its report clearly stated that things might change before and on Election Day.

A case in point is the poll commissioned by Anap Foundation and conducted by NOI polls Limited (NOIPolls). It was noted in the report that given the large pool of undecided voters and/or those who refused to disclose their preferred choice, Peter Obi’s 8 per cent-point lead at this stage is significant, but not sufficient to separate him from a leading pack of candidates scoring 13 per cent, 10 per cent and 3 per cent.

“Undecided voters and those who prefer not to reveal their preferred candidate add up to a whopping 23 per cent and 30 per cent respectively. The gender split of undecided voters shows that 27 per cent of women are undecided versus 18 per cent of male registered voters,” the report said.

In a statement signed by Atedo Peterside, the president and founder of ANAP Foundation, undecided voters and those who prefer not to reveal their preferred candidate add up to a whopping 23 per cent and 30 per cent respectively. The gender split of undecided voters shows that 27 per cent of women are undecided versus 18 per cent of male registered voters.

Similarly, the interim report from the nine-month study led by Dr Oludare Ogunlana of June Group Research and Council for African Security Affairs (CASA) with other Intelligence practitioners in the United States, Europe, and Nigeria pointed to the emergence of Atiku.

It was gathered that Atiku rates higher than other candidates from All Progressive Congress (APC) and Labour Party (LP), and New Nigeria Peoples’ Party (NNPP), using four key variables: political geography, religion, resources and class. The study’s scope is limited to observing political actors and interviews with stakeholders in organizations from across six zones of Nigeria and the Nigerians in the diaspora.

These include the likes of political leaders, women groups, religious organizations, civil societies, trade unions, and youth organizations.
According to the poll conducted by POLAF, a not-for-profit organisation, the election will be keenly contested among the top contenders. The results of the poll, which was conducted between July 2022 and February 2023, ruled out the possibility of a run-off, insisting that the Adamawa-born politician will win in most of the states.

On the methodology used for the poll, the organisation said three states were selected in each geopolitical zone, with an extra state selected from the South-South and South-West to make the number of states from which samples were drawn 20 out of 36 plus the Federal Capital Territory.

It said the sample was drawn from the voter register of the Independent National Electoral Commission (INEC), adding that 3,123,660 respondents across 165 local government areas were contacted via telephone and were spoken to in English and any other native language they were most comfortable speaking in.

The poll report stated clearly that things may change before and on election day. In giving credibility to its survey, NOIPolls said that the methodology used is almost the exact same methodology that was used in previous presidential polls in 2011, 2015 and 2019.

“In all those past presidential polls, the front-runner that was identified by our polls ended up winning the elections, irrespective of a rather large percentage of voters who were undecided and/or refused to indicate who their preferred candidate was.

“We are of course aware of some significant differences between the 2023 Presidential race and those of prior years e.g. 1) A longer Campaign season: 2) the presence of 4 major candidates instead of 2; and 3) greater voter enthusiasm,” the poll report signed by Peterside.

“In December 2022, we took the extra step to expand the methodology in our 2nd Presidential Poll in this series by carrying out 3 different polls using sample sizes of 1,000, 2,000 and 3,000 and confirmed that (as expected) the results did not change significantly. This notwithstanding, we chose a sample size of 2,000 for our third and final poll in February 2023 as opposed to the 1,000 sample size we used for the published September and December 2022 polls.”

Speaking on the different opinion polls and the election outcome, a researcher cum psychologist, Dr Ayotunde Elegbeleye, observed that those who gathered the data from the opinion polls might not have subjected the data to any rigorous statistical process. She, however, noted that there is hardly an empirical research that can reliably predict the outcome of an election.

Elegbeleye noted that there is no guarantee that everyone who responded to the poll exercised his or her civic responsibility at the end of the day, which in this particular context is more important to the outcome of the election than the mere prediction.

“If the result of opinion polls would be important to the eventual outcome of the election, research institutes should take up that responsibility of conducting the opinion polls, because as researchers they know what to do and assumptions to fulfill, research expected guidelines to follow to arrive at results that can accurately or almost accurately forecast.”

She noted that though there is always an error margin to the predicted outcome, most of the people conducting the opinion polls are not necessarily researchers, so they do not know the rudiments.

She added that sometimes the opinion polls are conducted by random interviews and questionnaires, asking citizens for their opinions, which might not fulfill the statistical assumption as well as research assumptions that needed to be fulfilled for an empirical result.

“So, with unfilled research assumptions, there is no way the outcome will be accurate or will reliably predict the outcome of the election. If it is important for the forecast of the opinion polls to meet the outcome of the election, then research institutes should pick up the task.

“Again, with elections in Nigeria, there are other factors that may come to play. So, there is always the human factor that will play out. Thus, there could be a particular projection and at the end of the day the outcome of the election will not align with the opinion polls because of the extraneous factors that may come to play during the election,” Elegbeleye said. She noted that sample size was definitely part of the issue because data must be gathered from a representative sample size in order to generalise.

Also commenting, Professor Lai Olurode, said that the presidential election is one election the country has never had. “What do I mean, if you look at 2015 and 2019 presidential elections, the margins of scores between the leading candidate who eventually won and the runner up was huge compared to the 2023 presidential election.

Between 2015 and 2023, there have been a lot of political permutations and movements. To a reasonably extent, the outcome of the opinion polls may not be faulted because some of the issues the election threw up may not have been seen by the different polls research.

“They are not God; they are just trying to do their best. Of course, you cannot remove the element of bias, but I think all of them were right because if Atiku Abubakar did not have any problem with the five governors, probably Atiku would have won.

“If you look at the results, we have never had this three horse race before. Also, may be the margin with which APC won would have been higher, if it did not have internal crisis.

“For me, preparation for 2027 should start now and all the stakeholders should take into cognizance observations from the fields to see how the process can be improved. It is a good development that our democracy is being consolidated. The electorate is the winner of last week’s election. They gave them a good fight that they cannot go to sleep again, but they should provide good governance,” Olurode stated.

The Guardian

Leave a Reply

Your email address will not be published. Required fields are marked *