Lies, damned lies and statistics
- Deryn Graham
Should we have more faith today in the statistics that shape policy and lives?
Attributed to two-time British Prime Minister Benjamin Disraeli, the quote, “There are three kinds of lies: lies, damned lies and statistics”, suggests that statistics are the most deceptive form of untruth.
In South Africa today, with a national statistics agency that is severely under-capacitated, can we trust that Statistics South Africa is being transparent? Can we have faith that there is no hidden agenda in the information that it publishes? Certainly, governments elsewhere have been accused of manipulating or deliberately misreporting their findings to paint a different picture from the country’s political reality. In the USA, Trump fired the head of the Bureau of Labor Statistics because he, Trump, didn’t like the numbers. In India, the government was accused of doctoring data to skew inequality levels.
“South Africa’s statistical agency is independent and credible,” asserts Dr David Francis, a Senior Researcher at the Southern Centre for Inequality Studies. “In South Africa, we live with the unburnished facts. We generally have good data, which people may or may not choose to believe or trust.”
David Everatt, µÚÒ»³Ô¹ÏÍø Professor of Urban Governance and former Chair of the South African Statistics Council, believes that part of the problem is that people want their own view of the world confirmed by data. If it does not meet their expectations, they assume that the numbers are wrong. “We want statistics to confirm our biases, not challenge our beliefs, and so we use data to support a weak argument or particular agenda,” he says.
Actions speak louder than numbers
If Statistics South Africa is reliable and credible, it is public perception around government performance that has largely led to public mistrust in the data it publishes.
“We have conflated our mistrust of politicians’ words and actions with the statistics. If data is used as intended – to shape government policy and to allocate resources, to build schools and hospitals where they are most needed – and these are not forthcoming, it is political will that is missing, not credible data,” says Everatt.
Yet we blame the data.
Associate Professor Mark Collinson, Reader in Population Health in the µÚÒ»³Ô¹ÏÍø School of Public Health, South African Medical Research Council/µÚÒ»³Ô¹ÏÍø Agincourt Unit, goes back to looking at the purpose of data.
“If people believe that the information gathered in the census, for example, is going to result in better service delivery but their lives do not materially improve, then they are justifiably disgruntled. When the process is perceived to have failed and not the implementation, this leads to an erosion of trust in the data.”
The data from the 2022 Census – collected post-Covid and following the 2021 political unrest and public riots – has been a contentious issue since its publication in October 2023. A deep-seated anxiety and mistrust around personal data usage and questions around anonymity, led to many people choosing not to complete the online survey or to answer the door to moderators. This resulted in a reported undercount of 31%, much higher than in most other countries.
Everatt believes that using big data – information held by banks, cell phone providers and other corporate entities - could supplement census data, which relies on voluntary submission.
“Accessing this kind of data may require legislation and other guardrails to assure our individual rights, and while it can’t substitute for a census, it could augment the census very nicely,” he says.
Data disconnect
The quarterly unemployment numbers are another important gauge of the health of the country, but they, too, are not without contestation. Our understanding of the concept of work versus employment goes to the heart of what the data reflects.
“The Labour Force Survey does not account for people who work for no compensation, such as caregivers. It relies on people’s interpretation of a job. For example, informal market vendors may not believe that they have a job, but they are working for money,” says Francis.
The current slowdown in the rate of unemployment reflected in official numbers is cold comfort for those who remain out of work. How can those people trust the data if it does not reflect their reality of no jobs?
While there is no silver bullet to the unemployment crisis that would reignite public trust, Imraan Valodia, Professor of Economics, Pro Vice-Chancellor: Climate, Sustainability and Inequality, and Director of the Southern Centre for Inequality Studies, believes that the data can influence behaviour. “When unemployment figures go down, hope is buoyed and more people may be encouraged actively to seek work again,” he says.
The right thing to do
Even if hope is restored, restoring trust is another matter. When citizens do not trust the data, they may abdicate from the democratic process. We see this in low voter turnout during elections and part of the census underreporting may be attributed to an unwillingness to participate.
Much like the slogan ‘It’s the right thing to do’, used by the SABC to encourage TV owners to buy a licence, Everatt believes that good citizenship must be the motivation for active participation in data gathering processes whose results impact us all. There is no other reward.
It would seem that the government’s track record over the last 30 years constitutes a massive own goal when it comes to trust. The trust of citizens has been squandered through politicians’ actions – or lack of action – and the value and credibility of national data has been compromised as a result.
We can use data in a different way, says Valodia. “Data is not just about the policies which come out of the numbers but a way of empowering citizens to hold politicians accountable.”
- Deryn Graham is a freelance writer.
- This article first appeared in CURIOS.TY, a research magazine produced by  and the .
- Read more in the 21st issue, themed #Trust — the invisible glue binding society, science and technology.