by Adam Perkins
Next we will examine the size of genetic effects on personality because, although the genetic influence on personality is modest compared to other traits (for example, intelligence), it is real. Therefore a full picture of the scale of welfare-induced changes in personality can only be obtained if genetic factors are also taken into account.
Finally, we will end this chapter with the presentation of a model that provides a quantitative estimate of the scale of welfare-induced personality mis-development through environmental and genetic channels, as well as an estimate of its likely cost to the public purse.
The Perry Preschool Project revisited
As we saw in Chapter 5, the Perry Preschool Project shows that two years of intensive preschool tutoring from the age of three to five years old can significantly improve the life outcomes of disadvantaged children decades later. For example, at the age of 40, the children in the Perry Preschool Project who did not receive intensive preschool tutoring earned approximately 25 per cent less, had twice the rate of unemployment and were twice as likely to have been incarcerated as the tutored children (Schweinhart et al., 2005).
These facts provide a compelling argument for government investment in intensive preschool tutoring as a means of boosting the life chances of disadvantaged children. But not all children are disadvantaged, and so the Perry Preschool Project only gives us a partial picture of the scale of the damage to adult life outcomes that can be done by childhood disadvantage. To fill this gap, we need to compare the life outcomes of the two groups of Perry Preschool participants to the life outcomes of less disadvantaged children with a similar demographic profile. A useful start point for this exercise is to note that the intensive preschool tutoring only lasted for two years and so the tutored children remained disadvantaged compared to individuals who were born into families who provided adequate care from birth onwards. This insight allows us to conceptualise, as shown in Figures 8.1 and 8.2, that the untutored children in the Perry Preschool Project represent a high disadvantage group, the tutored children in the Perry Preschool Project a medium disadvantage group and the population average a low disadvantage group (average African Americans who were around 40 years old in 2004).
By comparing life outcomes across three levels of disadvantage, we can put the effects of both childhood disadvantage and intensive preschool tutoring into perspective and begin to get a sense of how much damage could be done to society by a welfare state that increases the number of children born into disadvantaged households. It should be noted that this exercise will only provide an estimate of the relative scale of effects of childhood disadvantage on life outcomes (threefold, fourfold, fivefold and so on) and not the absolute level, because the age 40 follow-up of Perry Preschool participants was published in 2005 and so absolute figures for annual income, unemployment rates and other such historical data are not especially meaningful now.
Figure 8.1 Percentage of Perry Preschool participants who were unemployed at the age of 40 versus the average rate of unemployment for African Americans in 2004
Source: US Department of Labor/US Bureau of Labor Statistics in column one and Schweinhart et al. (2005) in columns two and three.
What we find is that childhood disadvantage has an approximately linear, dose-dependent effect on life outcomes. For example, Figure 8.1 shows that significantly fewer of the tutored children in the Perry Preschool Project were unemployed at age 40 (24 per cent) compared to the untutored children (38 per cent), yet their rate of unemployment was still almost three times higher than the average rate of unemployment for African Americans in 2004 (8 per cent; US Department of Labor/US Bureau of Labor Statistics).
Figure 8.2 Percentage of Perry Preschool participants who had served a prison sentence by the age of 40 versus the average rate of incarceration for African Americans in 2001
Source: Bonczar (2003) in column one and Schweinhart et al. (2005) in columns two and three.
A similar pattern occurs with criminality. For example, Figure 8.2 shows that significantly fewer of the tutored children in the Perry Preschool Project (28 per cent) had served a prison sentence by the age of 40 compared to the untutored children (52 per cent), yet their rate of incarceration was still more than twice as high as the average ‘ever-incarcerated’ rate for African Americans in 2001 (12 per cent; Bonczar, 2003).
Viewed as whole, the data presented in Figures 8.1 and 8.2 show that childhood disadvantage damages life outcomes in both employment and criminal domains but, more importantly in the context of this book, they also allow us to obtain an estimate of the scale of the burden that is imposed on society by childhood disadvantage. It should be noted that this will be a conservative estimate because the population average that represents the group with low levels of childhood disadvantage includes life outcomes for disadvantaged individuals: if they were stripped out of the data, the low childhood disadvantage group would have even better life outcomes.
The estimation process is a simple one: divide the number for the high childhood disadvantage category by the number for the low childhood disadvantage category. For example, if we look at Figure 8.1, we can see that 38 per cent of the high childhood disadvantage group were unemployed whereas only 8 per cent of the low childhood disadvantage group were unemployed. If we divide 38 by 8, we can see that the high childhood disadvantage group on average had a rate of unemployment that is 4.75 times greater than individuals in the low childhood disadvantage category. Similarly, Figure 8.2 shows that 52 per cent of the high childhood disadvantage group had served a prison sentence whereas only 12 per cent of the low childhood disadvantage group were former prisoners. If we divide 52 by 12, we can see that the high childhood disadvantage group on average had a rate of imprisonment that is 4.3 times greater than individuals in the low childhood disadvantage category.
A critic might dismiss these numbers as an artefact of the sampling differences between the low and high childhood disadvantage groups – after all, we must remember that the life outcomes of the low childhood disadvantage group are merely the population averages for African Americans who were aged about 40 in 2004, not those of a carefully selected experimental group as was the case of the Perry Preschool Project. But this is where the medium childhood disadvantage group becomes useful, because it acts as a sanity check for our estimate: as can be seen in the bar charts, the life outcomes of the putative medium childhood disadvantage group fall in between those of the low and high childhood disadvantage groups, suggesting that the differences between the low and high childhood disadvantage groups are not merely a fluke but instead are part of an approximately linear association between the severity of childhood disadvantage and the frequency of problematic life outcomes.
Based on these estimates, which are conservative for the reasons previously stated, we can now see that childhood disadvantage has a hugely damaging effect on life outcomes – in two key areas of life, namely employment and criminality, it causes a fourfold worsening of outcomes compared to average individuals. Moreover, by inspecting Figures 8.1 and 8.2, we can see that intensive preschool tutoring can only cancel out about half of that damage, because in every chart we can see that the life outcomes of the tutored children from the Perry Preschool Project are positioned approximately midway between those of the untutored children and the average for the population as a whole.
This insight leads us to a key finding of this book, which is that a welfare state which increases the number of children born into disadvantaged households risks imposing a significant per capita headwind on society, because each of those children will, on average, go on to create a burden for the welfare state and the criminal justice system that is four times larger than the burden imposed by average individuals. Bearing in mind that the average includes disadvantaged individuals, we can be confident that the real headwind is even stronger. Moreover, even if we lived in some utopia in which we could afford to give every disadvantaged child a place in a scheme for intensive preschool tutoring, we can see that this hea
dwind cannot be fully cancelled out by remedial action: the tutored children still create a burden for the welfare state and the criminal justice system that is approximately two times larger than of the population baseline.
This quantification of a per capita headwind is key for later in the book because if we can next obtain an estimate of the numbers of extra children born due to welfare incentives, we can combine the two estimates to make an educated guess about the scale of the damage caused to a nation by a welfare state that increases the number of children born into disadvantage.
But this book is not just about disadvantage – it is also about personality. We now need to estimate the proportion of that headwind that can be attributed to personality. For example, it may be true that childhood disadvantage increases the rate of unemployment in adulthood, but if none of that effect is caused by personality deficits resulting from exposure to disadvantage, then welfare-induced personality mis-development is not worth worrying about.
Of particular importance is the already-cited study by James Heckman, Rodrigo Pinto and Peter Savelyev (2013), who used data from the Perry Preschool Project to show that childhood disadvantage blights life outcomes at age 40 by altering personality. More specifically, disadvantaged children who received two years of intensive preschool tutoring developed personality profiles that were significantly less aggressive, antisocial and rule-breaking than the untutored children (Heckman et al., 2013). This study is key to my model because it not only showed that experimentally induced changes in personality influence life outcomes in adulthood at age 40, but also estimated the effect size of personality compared to other contributory factors.
This study shows us that personality matters: Heckman and colleagues (2013) found that approximately 50 per cent of the crime-related treatment effect and 20 per cent of the employment-related treatment effect can be attributed to experimentally induced changes in personality development and the remainder to other causes (treatment effect means the difference in life outcomes between the children who received intensive preschool tutoring and those who did not receive it). We should note that there were sex differences. For example, personality improvements amongst tutored male participants are responsible for approximately 70 per cent of the treatment effect in the case of felony arrests at the age of 40 and approximately 40 per cent of the treatment effect in lifetime arrests at age 40. Likewise, personality improvements were responsible for approximately 20 per cent of the treatment effect in the case of employment record at the age of 40.
Personality improvements amongst female participants are responsible for approximately 65 per cent of the treatment effect in the case of felony arrests at the age of 40 and approximately 70 per cent of the treatment effect in lifetime violent crimes at age 40. Likewise, personality improvements were responsible for approximately 10 per cent of the treatment effect in the case of months in marriage at the age of 40.
Viewed in tandem with Figures 8.1 and 8.2, these data show that disadvantaged children create a burden for the criminal justice system and for the welfare state that is four times larger than the burden imposed by average individuals and that approximately half of this burden can be attributed to personality problems caused by childhood disadvantage. From this appraisal we can conclude that disadvantage-related personality problems are a key driver of blighted life outcomes and the scale of such outcomes is non-trivial.
Socialisation as a model of welfare-induced personality damage
The data discussed up to this point in the chapter suggest that a welfare state which causes many thousands of children to be born into disadvantage can cause personality damage that imposes a significant economic and social headwind on the nation. But these data only capture personality damage due to environmental factors, because the children in the Perry Preschool Project were randomly assigned to receive tutoring and thus genetic factors played no part in this set of results. Therefore we cannot obtain a full estimate of the scale of welfare-induced personality damage, both environmental and genetic. To move the argument forward, we need a more sophisticated model that integrates environmental and genetic factors in a single framework. We can do this by conceptualising the employment-resistant personality profile as a tendency to break social norms, which allows us to utilise existing knowledge concerning the process by which the norms of a society are instilled in its members. This process is known as socialisation (Clausen, 1968) and has been studied for decades. The leading theory is that an individual’s level of socialisation is determined by an interaction between the size of their dose of genes for being difficult to socialise and the quality of their upbringing.
This idea was chiefly pioneered by David Lykken, whose model of socialisation goes like this:
With the best parents and home environments, the only antisocial offspring will be those who are the most fearless, aggressive, impulsive, and so on – psychopaths with truly hard to-socialize temperaments. In the worst home environments, a large fraction of all offspring will remain unsocialized. Over the broad middle range of parental competence and environmental risk factors, the incidence of antisocial offspring will be a product-function of parental incompetence (or indifference or parental sociopathy) and the child’s innate proclivities. A complicating factor is that the worst parents are likely to contribute hard-to-socialize genetic tendencies as well.
(Lykken, 1995, p. 563)
This model might seem confusing, but it has at its heart the simple principle that children at both the low and the high extremes of the genetic spectrum are relatively insensitive to their environment. So a child with a high dose of the genes for being difficult to socialise is likely to develop the employment-resistant personality profile despite growing up in a nurturing environment that is conducive to the development of solid citizenship. Conversely, a child with a low dose of genes for being difficult to socialise is likely to develop into a solid citizen despite minimal parental interventions. In the majority of children, who carry an average-sized dose of genes for being difficult to socialise, the environment is more important. Competent parenting will convert most of these children into solid citizens whereas incompetent parenting will sway most of them towards employment-resistance.
A general point must be absorbed here, namely that there are many more genetically average children than there are genetic outliers, meaning that the childhood environment is crucial in determining the overall level of socialisation in the population. This explains why developed countries have converged on investing in schools as an essential measure for the improvement of human capital by boosting socialisation.
Unfortunately, as we saw in the experiments on the effects of intensive preschool tutoring described in Chapter 5, most of the work in socialising a child occurs before school age. This discovery has prompted leading economists to urge governments to think about more than just schooling when attempting to boost human capital and, in particular, to pay more attention to the role of the family: ‘An effective skill formation policy must account for the role of the family in producing skills and motivation. Dysfunctional families produce impaired children’ (Heckman & Masterov, 2005, pp. 15–16).
Returning to the theme of personality, when scores on a trait influenced by thousands of genes are plotted along a scale, 68.2 per cent of the population score near to the middle of the scale. These are the average scorers. The remaining 31.8 per cent of the population score significantly lower or higher than average. This bell-shaped pattern is known as the normal distribution and occurs in almost all frequently measured human variables, such as height or penis length. Personality data are no exception, as we see in Figure 8.3, which illustrates the normally distributed nature of questionnaire scores on conscientiousness in 2,532 participants in one of my own studies.
Figure 8.3 The distribution of questionnaire scores on conscientiousness in 2,532 participants from one of my own studies (dashed lines indicate one standard deviation)
If we combine the statistical concept of the normal
distribution with Lykken’s model of socialisation, we can see that approximately 16 per cent of children are born with a significantly higher than average dose of the genes for being difficult to socialise. These children are genetic hard cases, who are relatively insensitive to their upbringing and thus have a significantly elevated risk of turning out to be employment-resistant adults regardless of whether they are neglected as children. This explains the existence of individuals who grew up in privileged, nurturing households but nevertheless turned out to be employment-resistant adults. Since employment-resistant individuals are over-represented amongst welfare claimants, we can predict that the genes for being difficult to socialise are over-represented amongst welfare claimants. This means that a welfare state which boosts the number of children born to claimants will inflict direct genetic harm on the social and economic prospects of the nation (that is, harm its human capital) by swelling the number of babies who fit into the left-hand side of Figure 8.3 due to being born with a significantly higher than average dose of the genes for being difficult to socialise.
At the other end of the scale, we can predict that approximately 16 per cent of individuals are born with a dose of the genes for being difficult to socialise that is significantly smaller than average. Although it may seem incongruous to think of them as such, these people are also genetic hard cases because they too are relatively insensitive to their upbringing. In these children, their genetic profile means they have a high probability of turning out to be solid citizens, whether or not they are neglected. This explains the existence of individuals who suffered neglect during childhood yet nevertheless turned out to be solid citizens. Thus, a welfare state that boosts the number of children born to claimants will inflict genetic harm to the human capital of the nation via a second route, namely by shrinking the number of babies who fit into the right-hand side of Figure 8.3.