Opinion / Columnist
The myth of crowd sizes
13 Jun 2018 at 18:02hrs | Views
As human beings, we are always trying to predict the future. Who will win the World Cup? What will the weather be like next weekend? And of course, who will win the upcoming election?
In the absence of perfect information, we look to trends and different indicators to help us to most accurately work out what will happen.
The question of course is what indicators to choose. In football, form, quality of squad and past record are usually better predictors of the likely winners, then for example, the quality of a kit. In weather, while not perfect, basing your weekend plans on scientific forecasting is probably a better bet than on a friend's advice.
And what of politics? Throughout this election cycle, we have been hearing people make predictions based on crowd sizes. Each time Emmerson Mnangagwa or Nelson Chamisa pack out a venue, social media is full of their supporters somehow claiming that a big attendance at one of their rallies means that "victory is certain!"
The problem is that historically, crowd sizes are an awful predictor of an election outcome for three main reasons. First, they are highly subjective. Short of counting the number of attendees as they arrive at an event, how can we tell exactly how many people turned up? At almost every rally (of both candidates), we are seeing two types of pictures – one showing a sea of supporters, and the other from a different angle, highlighting the empty spaces. Which to believe?
Second, the number of people attending campaign events is insignificant in comparison to the overall electorate. With around 5.5 million registered voters, the winning candidate will likely need over 2 million votes. Once we understand this, we realise it really isn't predictive if 5000 or 8000 people attend an event.
Finally, crowd sizes are self-selecting. Generally, only partisan party supporters attend rallies. So while crowd sizes may be reasonable measures of the enthusiasm of a party base and the fervour of hard-core supporters, they do little to measure the voting intentions of the swing or undecided voters who determine elections, and are unlikely to ever attend either party's events.
Instead of crowd sizes, the only accurate way to predict an election result is by credible opinion polls. Unlike crowd sizes, opinion polls are scientific, using the key principle that each person is equally likely to be selected in the poll. This randomness is key. 100,000 people attending a rally is less significant that a random sample of only 1000 people that accurately represent the population as a whole. Like weather forecasting, while not perfect, polling is the best option we have.
Last week, Afrobarometer, an extremely well respected, independent research body, released the first scientific and credible poll of this election season. The nationally representative sample of 2400 Zimbabweans, gave a margin of error of +/-2% at a 95% confidence level. Crucially, it was based on the principles that all respondents are randomly selected, the sample is distributed across provinces and urban/rural areas in proportion to their share in the national population, and every adult citizen had an equal chance of being selected.
The results of the poll - that Emmerson Mnangagwa currently leads Nelson Chamisa by around 12% - are not the focus of this article. I don't want to become a part of that debate. Polls are a ‘snapshot in time' that reflect the public mood at the time the poll was taken. A lot can change over the next couple of months.
Instead the point I am trying to make is that as Zimbabweans, we must be less emotional, and more open to science. Over the past few days, I have seen many people that I otherwise respect trying and failing to discredit Afrobarometer and their findings, based on their own biases. It is absurd to claim an independent, thorough and scientific study is rubbish, while claiming your candidate is certain to win as he attracted a few thousand more people to a march or rally. In weather terms, it is the equivalent of ridiculing the forecast that it is going to rain, because your mother told you it would be sunny. And so you go outside in shorts and a t-shirt, only to get soaked! If our political culture is ever going to evolve and become peaceful and issues-based, we have to be smarter than this.
I am not claiming polling is perfect and I never will. But it does give us our best chance of understanding what is likely to happen. To adapt Winston Churchill's well-worn quote, "polling is the worst form of predicting election results, except for everything else that has been tried!"
In the absence of perfect information, we look to trends and different indicators to help us to most accurately work out what will happen.
The question of course is what indicators to choose. In football, form, quality of squad and past record are usually better predictors of the likely winners, then for example, the quality of a kit. In weather, while not perfect, basing your weekend plans on scientific forecasting is probably a better bet than on a friend's advice.
And what of politics? Throughout this election cycle, we have been hearing people make predictions based on crowd sizes. Each time Emmerson Mnangagwa or Nelson Chamisa pack out a venue, social media is full of their supporters somehow claiming that a big attendance at one of their rallies means that "victory is certain!"
The problem is that historically, crowd sizes are an awful predictor of an election outcome for three main reasons. First, they are highly subjective. Short of counting the number of attendees as they arrive at an event, how can we tell exactly how many people turned up? At almost every rally (of both candidates), we are seeing two types of pictures – one showing a sea of supporters, and the other from a different angle, highlighting the empty spaces. Which to believe?
Second, the number of people attending campaign events is insignificant in comparison to the overall electorate. With around 5.5 million registered voters, the winning candidate will likely need over 2 million votes. Once we understand this, we realise it really isn't predictive if 5000 or 8000 people attend an event.
Finally, crowd sizes are self-selecting. Generally, only partisan party supporters attend rallies. So while crowd sizes may be reasonable measures of the enthusiasm of a party base and the fervour of hard-core supporters, they do little to measure the voting intentions of the swing or undecided voters who determine elections, and are unlikely to ever attend either party's events.
Instead of crowd sizes, the only accurate way to predict an election result is by credible opinion polls. Unlike crowd sizes, opinion polls are scientific, using the key principle that each person is equally likely to be selected in the poll. This randomness is key. 100,000 people attending a rally is less significant that a random sample of only 1000 people that accurately represent the population as a whole. Like weather forecasting, while not perfect, polling is the best option we have.
Last week, Afrobarometer, an extremely well respected, independent research body, released the first scientific and credible poll of this election season. The nationally representative sample of 2400 Zimbabweans, gave a margin of error of +/-2% at a 95% confidence level. Crucially, it was based on the principles that all respondents are randomly selected, the sample is distributed across provinces and urban/rural areas in proportion to their share in the national population, and every adult citizen had an equal chance of being selected.
The results of the poll - that Emmerson Mnangagwa currently leads Nelson Chamisa by around 12% - are not the focus of this article. I don't want to become a part of that debate. Polls are a ‘snapshot in time' that reflect the public mood at the time the poll was taken. A lot can change over the next couple of months.
Instead the point I am trying to make is that as Zimbabweans, we must be less emotional, and more open to science. Over the past few days, I have seen many people that I otherwise respect trying and failing to discredit Afrobarometer and their findings, based on their own biases. It is absurd to claim an independent, thorough and scientific study is rubbish, while claiming your candidate is certain to win as he attracted a few thousand more people to a march or rally. In weather terms, it is the equivalent of ridiculing the forecast that it is going to rain, because your mother told you it would be sunny. And so you go outside in shorts and a t-shirt, only to get soaked! If our political culture is ever going to evolve and become peaceful and issues-based, we have to be smarter than this.
I am not claiming polling is perfect and I never will. But it does give us our best chance of understanding what is likely to happen. To adapt Winston Churchill's well-worn quote, "polling is the worst form of predicting election results, except for everything else that has been tried!"
Source - Anthony Mkondo
All articles and letters published on Bulawayo24 have been independently written by members of Bulawayo24's community. The views of users published on Bulawayo24 are therefore their own and do not necessarily represent the views of Bulawayo24. Bulawayo24 editors also reserve the right to edit or delete any and all comments received.