The problem with prison population predictions

The problem with prison population predictions

Prison population forecasts: using the future to predict the past*

Statistics are a way of telling stories through numbers and graphs. Like any story, sometimes a statistical narrative can feel so compelling or authoritative that we lose the capacity to question it, and therefore to decide how it ends. In this post I look at how this happens with prison population projections , and I make two claims. First, I challenge the idea that past prison populations are a good predictor of future ones. And second, I suggest that the future of prison populations is not inevitable but squarely within our control.

When I was advising the Scottish Prisons Commission on their work in 2008, the prison population was hovering just below 8,000 prisoners, more than twice as many as neighbor Norway (3,600). The latest Scottish forecast shows a prison population in 2021 of 9,500 (figure 1). What seemed like a catastrophic threshold in 2008 has now become a reality which is quickly being overtaken by even bigger numbers. How did we get to 8,000, and what will lead us to 9,500?

Figure 1 OUR FUTURE? Scottish prison population projected to the year 2021

Source: Scottish Government (2012)

Statistical forecasts answer these questions by looking to the past. There are many and highly sophisticated methodologies for analysing how the past matters, but in relying on historical data in the first place an important assumption is being made about prison populations: that past growth has a predictable cause and pattern of change that can be discovered and thus extended into the future. Figure 2 shows historical growth in the Scottish prison population, and it doesn’t take an expert to see how figure 2 is connected to figure 1.

Figure 2 THE PAST? Scottish prison population growth 2002 - 2012

Source: Scottish Government (2012)

The story – literally a linear narrative – told by figures 1 and 2 is that the future looks a lot like the past, a gradual but persistent march upwards. It’s hard to argue with a straight line. It is both easy to understand but uses high science to produce, exerting a powerful legitimacy. A straight line also makes for a boring story, but boring stories are dangerous. We tend to overlook them as topics of critical inquiry; we assume they are correct. This story tells us what we might already have expected, that the Scottish prison population has been expanding and will keep on doing so. It makes a future of rising prison populations feel inevitable.

Policy makers want straight line stories because predictability, their central virtue, is essential to planning. Accurately predicting the future allows for sensible investment strategy. Getting it wrong has significant costs: building too many cells wastes millions of pounds (that could have been used on hospitals, schools and pensions); building too few risks public safety and outrage.

Statisticians have done their best to assist policy makers, trying to find the factors that explain past levels of growth.  They have been given an impossible task, however, because prison growth often is a function of volatile and unpredictable events. The murder of Jamie Bulger, 9/11, the 2008 financial crash, the election of the New Labour Government in 1997, and the election of a nationalist Government in Scotland all have had major consequences for law and policy, and thus for prison numbers. Despite this, statistical regression models continue looking at the past to generate a vision of the future.

This is why I suggest that prison population forecasting is like walking up a hill backwards: looking down the slope you have come up is a strange way of figuring out when you will hit the peak. It is no wonder that prison projections regularly get it wrong, failing to anticipate the unprecedented explosion of the penal system beginning in the late 1990s in England and Wales (when projections based on 1980s growth suggested populations would stabilise or fall), or the surprising decline of America’s world beating prison numbers in 2007/2008.

The reason for the title is that predicting the future of prison populations also has been a way of imposing coherence on the past, turning jagged spikes and dips into an even rate of growth. In this sense forecasting models absorb isolated and extreme events spreading their impact evenly across the past. It is like averaging deaths during a few peak years of the Great Plague over a century. Instead of an extraordinary moment in human history standing out, we see an unremarkable annual loss of life over a long period.

A real danger is that we build prisons to suit the projections (and this looks to be happening with the revived Titan Prisons programme down south) rather than our values about the role and quantity of punishment in society. There is growing evidence that prisons themselves ‘cause’ prisoners, that as prison spaces are created, the larger the total population gets. Prison forecasts then are at risk of triggering a self-fulfilling prophecy by encouraging capital expansion that is then filled in a dystopian version of Field of Dreams – if you build it, they will come.

We know that Scotland uses prison much more actively than its small country neighbours and we also know that high prison populations are associated with reduced social investment and welfare. Rather than letting prison forecasts become a self-fulfilling prophecy, perhaps we should read them as an indictment of our current practices. And if we dare to question the right of a straight line to tell us our destiny, we might reclaim the future as something we select rather than predict.

This blog was written by Dr Sarah Armstrong, a Senior Research Fellow in Glasgow University and the SCCJR. Her research interests revolve around prisons and punishment: policy processes that shape and sustain them


*The full version of this essay appears in Issue 3 of Scottish Justice Matters ( )

Scottish Government (2012), Prison statistics and population projections Scotland, 2011-12 (29 June 2012). Online at:

Category Penal Policy