The marshmallow test and the crisis in social policy

Two very significant studies were published in May. On the face of it they had little relation to each other, but together they have shattered powerful myths that have brought our social policy to the point of crisis. In this post, Atif Shafique explores what all this might mean, and how we can use this moment to redesign social policy for the better. This post originally appeared on the RSA blog.


One study reported on a repeat of the famed ‘marshmallow test’ experiment. The original study from the 1960s and 70s appeared to show a link between children’s ability to delay gratification - which demonstrated a capacity for emotional self-regulation - and long term social outcomes, including higher intelligence and more successful employment. The latest study, which used a much more robust design, found that once socioeconomic characteristics were accounted for, the ability to delay gratification had a negligible impact on outcomes in later life. Environmental factors largely determined success.

The second study from May presented the findings of an extensive five-year research project examining the use of conditionality (making state support contingent on recipients behaving ‘responsibly’) in the UK’s social security system. The study period followed major welfare reforms including a significant expansion in the use of sanctions and the extension of conditionality to in-work recipients of benefits, lone parents and those with disabilities.

The final report of the project deployed gentle academic language, but its conclusions were damning. The arguments (or more appropriately, myths) that are at the very heart of our welfare system were shown to be false. Welfare conditionality was not effective in helping people enter or progress through employment. Most of those that did find some work were cycling in and out of insecure jobs and unemployment. For a significant minority of people, conditionality led to negative ‘behaviours’ and outcomes, including disengaging from the welfare system completely, experiencing sharper poverty, destitution, and ill-health, and even moving into survival crime to make ends meet.

Benefit sanctions were especially bad, routinely triggering profoundly negative outcomes. At the launch of the report, a GP in the audience recalled how the environment of policing, stigmatising and sanctioning people had become so awful that even JobCentre staff were coming to him with mental health issues linked to what they had seen or were asked to do.

Social policy’s myths about the poor

Both of these studies are connected because they tell us something profound about how the state and wider society have thought about the poor.

The marshmallow test captured the imagination of the public, policy makers and educationalists. Some understood it to imply that if poorer people could mimic middle-class behaviours, they too could become successful. The original research was conducted in the same decade that American anthropologist Oscar Lewis first widely used the term “the underclass” to describe people trapped in poverty. The central characteristic of the underclass according to Lewis is a “strong present-time orientation, with little ability to delay gratification and plan for the future.” As the solidarism of the post-war period faded, so did the idea that addressing poverty was a collective responsibility. If individuals wanted to escape poverty, they would need to change their behaviour.

One of the milder (though misguided) consequences of this was in education. Educationalists - some of them excited by the marshmallow test - thought that they could help tackle poverty and inequality of opportunity by teaching "character,” even as neo-liberal economic reforms were tearing apart the communities they taught in. Character education went on to become a huge movement (and industry) - some of it useful, some of it completely misdiagnosing the root causes of the social mobility crisis.

On the more serious end of the scale, the toxic concept of conditionality was transforming social policy and the social security system in particular. The state’s role was becoming less and less about providing security, stability and opportunity, and more and more about policing the behaviour of the poor. The idea that Reagan-Thatcherite reforms weakened state power is wrong. Instead what we saw was the state intervening in the lives of particular groups within society more than it ever had.

Lawrence Mead’s philosophy on the new paternalism was a major catalyst. Mead argued that creating a self-reliant society in the future (a major right-wing aspiration) required government to tell ‘dependent’ people how to live their lives today, in order to snap them out of the entrenched cultures and ingrained behaviours that drove their poverty. The welfare system needed to change the signals it was sending to people. Foreshadowing the toxic narratives and social attitudes about benefits recipients today, Mead wrote that “government must persuade [poor people] to blame themselves.” Income support should no longer be an entitlement, but rather a privilege that is earned by exhibiting good behaviour. Government should have the ability to deny support to those that don’t comply, with sanctions acting as a major lever to change the behaviour of the poor. Mead’s philosophy influenced the welfare reforms introduced in the US and the UK, initially by the Clinton administration and New Labour.

If Mead established a link between state support and good behaviour, it was sociologist Charles Murray who convinced policymakers that the underclass was gaming the system; that they were reaping economic benefits from their welfare dependency, and that their welfare dependency was a lifestyle choice. As a result, “fairness” became a central concept in the lexicon of welfare reformers, allowing them to distinguish between the ‘deserving’ and the ‘undeserving’ poor; between ‘strivers’ and ‘scroungers’.

Conditionality: A terrible idea that has taken hold of social policy

The writings of Lawrence Mead and Charles Murray were highly influential. They shaped both the substance of policy change and the public narratives which gave new policies political and societal cover. From the 1990s onwards, conditionality became a mainstay of social policy. Technocrats and politicians - who for the most part had no lived experience of hardship or poverty - could introduce these dramatic changes while being comforted by the notion that they were ultimately doing it for the benefit of those affected, in particular the ‘deserving’ poor that were behaving responsibly and respecting the rules of the system.

‘Welfare to work’ has been the most visible area in which conditionality has been applied, but its tentacles have reached deeper than that. Since 2010, the British government has extended it to support for lone mothers, disabled people and people in-work but reliant on benefits (deemed not to be ‘working enough’). Tenants of social housing have also had long-term tenancies removed, with tenancy renewal dependent on good behaviour. All manner of support - from services to the homeless to support for those in hardship - has either some form of behavioural requirement baked in or is designed to create incentives for behaviour change. Some of the reforms aren’t even about changing behaviour, but simply removing people from the benefits system altogether by creating an unfriendly, deliberately complicated system of means-tested support. This is poverty-by-design, for an underclass that is apparently undeserving of state support.

The conclusions of the Welfare Conditionality project could not be clearer on how negative these radical reforms have been. It comes as no surprise that the report found policies and support that were non-stigmatising and holistic were most effective in achieving good outcomes.

Change policy, not behaviour

The conditionality project and the marshmallow test have smashed one simple, powerful myth: that if the poor were to change their behaviour (or ‘build character’), they too could become successful. The structures of exclusion, inequality and marginalisation they face become either immaterial or secondary.

Recent research supports arguments that it is not poor people’s inherent behaviours or cultures that impacts how they live and the decisions they make, but rather the material scarcity and precarity that they face. A groundbreaking study by Eldar Shafir and Sendhil Mullainathan found that the experience of scarcity imposes a heavy cognitive burden on those affected. This can mean they make decisions for the short rather than long term, and lack the cognitive resources to take on mentally demanding tasks because so much of their energy is exerted on the day-to-day challenge of making ends meet. Experimental studies show that everyone (not just ‘the poor’) can be impaired in the same way if they experience conditions of scarcity.

Policies based on conditionality and compliance make this worse. If so much of people’s cognitive capacity is taken up by navigating the absurd complexities and requirements of the welfare system - on top of the everyday pressures of hardship - how can we realistically expect people to simultaneously plan for the long term, invest in their skills, and progress through work? It is the stability of affluence that creates an environment for ‘building character’ and achieving success, not the other way around.

It follows that the most desirable approach for tackling the challenges facing the poor is to confront material scarcity and precarity, and to create the conditions in which all people have the cognitive resources to make positive decisions about their life - whether it’s about education, training, work or community life. In Finland, a remarkable decline in homelessness was achieved by ditching conditionality all together in housing support. While accommodation is provided in England as a “reward” for good behaviour (for example engaging with treatment services), the Finnish government gave homes to homeless people without any preconditions. Homelessness went down, engagement with support services increased, and rates of recovery went up.

This is also why proposals for universal basic income (of which the RSA is a big advocate) are so important, and why criticisms of it are often wide of the mark. It is not that on its own a UBI will eliminate poverty, but rather it may create the basic security and freedom (and release the cognitive resources) for people to make positive decisions about their life.

A few years ago, the idea of a “zero-based” spending review became popular, as a way of examining every pound of public money spent. I think the energy was misplaced. If there’s one area that’s ripe for a root-and-branch examination, it’s our welfare system and the wider social policies connected to it. Conditionality has failed. It’s more important than ever to ask what can take its place.

To find out more about the RSA's wo