A former Statistician General of the National Bureau of Statistics of Nigeria from 2011 to 2021, Dr. Yemi Kale, has offered reasons why results from surveys and polls fail in Nigeria.
Kale, who has been given an appointment by KPMG, offered the reasons in a series of tweets on his Twitter handle.
The tweets come amid the debate over why the presidential candidate of the Labour Party in the February 25, 2023 poll in Nigeria, Peter Obi, came this when several polls declared him the favourite.
The Eagle Online recalls that Obi came third in the results of the election announced by the Independent National Electoral Commission in the early hours of March 1, 2023.
While Asiwaju Bola Ahmed Tinubu of the All Progressives Congress was declared winner, Atiku Abubakar of the Peoples Democratic Party came second.
In the series of tweets on why results from polls and surveys are not accurate, Kale said: “I was asked this recently. Why will results of a survey/poll be wrong? Was it intentional manipulated to mislead & if not how can results say one thing & outcome another? Polls saying one thing & outcome being different occurs everywhere in the world. But why does it happen?
“This is an increasing trend with several polls/surveys in 🇳🇬 . One major reason I have observed is non representation of the sample. Eg number of undecided/non response/refused to say, is more than the total respondents which creates a bias & might be flawed or unreliable.
“For eg, you survey 10k people on say whether they will buy a new product & 7k refuse to response or are undecided etc. Of the remaining 3k, 2k say they will buy & 1k say no so you conclude product will be a hit as 66% say they will buy. It has a 66% approval rate is announced.
“But upon enthusiastically releasing say 1m units of the product & expecting about 660k in sales since survey said 66% will buy, only 20% is sold. How can this be when you had a 66% survey result? Well key is in the 7k no response/undecided. Your representative sample was 10k.
“Only 3k actually responded to the actual info you wanted. This is an important part of any survey/poll. It has to be representative.If it is not then it’s results are not reliable & no useful insights can/should be made from it. But many polls/surveys in 🇳🇬 make these errors.
“So if you want to check level of reliability of any poll/survey on whatever topic go into the methodology & check the degree of representation & particularly the ratio of non responses. If it’s too large then it might not be reliable & it’s results should be treated with caution.
“A way to deal with such a dilemma is called “replacement” If your representative sample requires 10k spread across but get only 3k who actually respond, then you replace the 7k that gave no answers using same spread. You just don’t go with the 3k & make conclusions from it.”