The Welfare Trait
Page 15
MZ twin siblings turn out to be more similar on almost all types of behaviour than DZ twin siblings and, consequently, show that almost all types of behaviour are genetically influenced (Turkheimer, 2000). However, it is important to note at this stage that behaviour genetics research also shows that the degree of heritability is not the same for all behavioural traits. For example, intelligence is one of the most heritable behavioural traits, with approximately 60 per cent to 80 per cent of the variance in intelligence in adults being explained by genetic factors (Deary, Johnson & Houlihan, 2009). Personality traits are also heritable but much less so than intelligence: for example, twin studies show that the personality traits which are the focus of this book, namely conscientiousness and agreeableness, are somewhere between 30 and 40 per cent heritable (Bouchard, 1994).
This point about personality being significantly less heritable than intelligence might seem tangential to the topic of this book – personality and welfare – but it is important to make in order to avoid this book being mischaracterised as just another version of the long-running argument that higher birth rates amongst individuals with low intelligence can cause genetically-based decreases in the intelligence level of the population. Since empirical tests of this latter (dysgenic) hypothesis usually show the opposite, namely that the average level of intelligence of the population is rising (the so-called Flynn effect; Flynn, 1994), critics might seek to tar the welfare trait theory with the same brush, on the basis that the heritability of personality is less than that for intelligence and so if intelligence is not suffering from dysgenic effects then personality is even less likely to do so.
But this is lazy thinking: apart from the obvious rebuttal that this book is about personality and welfare, not intelligence and welfare, the lower heritability of personality makes more room for environmental effects. It therefore makes more room for the personality profile of the population to be harmed by a welfare state which increases the number of children born into disadvantaged families. This fact means that my argument is not the tired old Social-Darwinist dysgenic one that has been offered many times before in intelligence-related debates, but rather hinges less on genetic factors and more on the crucial role of childhood disadvantage in forming the employment-resistant personality, as demonstrated by James Heckman and colleagues using evidence from randomised controlled trials (Heckman et al., 2013). No critics have been able to offer cogent evidence to contradict Heckman’s finding – because there is none.
It should however be noted that twin studies have a variety of limitations. For example, they usually make the assumptions that mating is random, that identical twins are not treated more similarly than fraternal twins and that only one form of genetic mechanism is acting on a particular trait. There is evidence to cast doubt on these assumptions and thus on twin studies: for example, some studies suggest that mating is not random with regard to personality and that there is a slight trend in humans to mate with those whose personalities resemble our own (Le Bon et al., 2013).
To reject the results of twin studies of personality because of these issues would however be to throw the baby out with the bathwater. We know this because the results of twin studies are backed up by two other types of behaviour genetic experiment. First, in studies of personality in twins reared apart, we find heritability estimates for personality that are similar to those generated by studies of personality in twins that were reared together (for example, Tellegen et al., 1988). Second, twin study results are backed up by a new way of estimating genetic influences on traits, known as genome-wide complex trait analysis (GCTA). Pioneered by Peter Visscher and colleagues, GCTA allows an estimate of the variance in a trait that is explained by genetic variants that are in common between the people surveyed, even if those people do not come from the same family. Briefly stated, GCTA compares DNA profiles to estimate the degree of genetic overlap or relatedness amongst the participants. The degree of overlap is then related to the amount of overlap on some measured characteristic, such as height, weight or personality. The significance of GCTA is that it allows genetic effects on traits to be studied in any random sample of people, meaning that GCTA analyses don’t suffer from the limitations of twin studies.
Importantly for the argument about genetic effects on personality, GCTA studies confirm twin study findings, showing that the more related two people are, the more similar they tend to be in their personality characteristics (for example, Verweij et al., 2012; Vinkhuyzen et al., 2012). As a caveat, it should be noted that GCTA studies only capture additive genetic effects on traits but are insensitive to non-additive effects. This contrasts to twin studies, which are a metaphorical dragnet and are able to detect all forms of genetic influence on a trait. Therefore, whereas twin studies indicate personality traits are approximately 30 per cent to 40 per cent heritable, GCTA studies provide lower estimates of personality heritability, in the range of four per cent to 12 per cent. For any readers unfamiliar with the meaning of the term ‘additive genetic effects’, it refers to when alternative forms of a gene (known as alleles) add up to influence a trait. For example, if the genes in question are additive, an organism with one gene for blue colouration and one for yellow would turn out to be green. When genes interact with each other or the environment to influence a trait, we see what is known as a non-additive genetic effect: GCTA studies are blind to this form of effect.
The advent of GCTA means that, even though there are limitations to twin studies, we can accept their basic message, namely that genes influence personality and that identical twins have personalities that are more similar than non-identical twins because of their greater genetic similarity. Furthermore, we can accept that this effect is real and not merely an artefact of some confounding factor such as non-random mating or the tendency for parents to treat identical twin pairs more similarly than non-identical twin pairs. If the picture still seems confusing, think about it like this: if personality was 100 per cent controlled by genetic factors, identical twins would have identical personality profiles (assuming zero measurement error) whereas non-identical twins would not. Conversely, if personality was 100 per cent controlled by environmental factors, then the personalities of identical twins would on average be no more similar than those of non-identical twins. What we find is something in between: identical twins do not have identical personalities, but their personalities are on average more similar than the personalities of non-identical twins, showing that there is a clear influence of genes on personality.
However, in order to pull together the research summarised in this chapter into a coherent whole, we must finish off by thinking more about what is meant by environmental effects on personality. As the studies summarised in this chapter showed, differences between individuals in personality attributes are significantly influenced by genetic variants, but they also showed that on average environmental variation shapes personality even more strongly, in a ratio of roughly 40 per cent to 60 per cent. However, the partitioning of environmental influence on personality is odd: despite being raised together, identical twins have personality profiles that are far from identical, suggesting that the shared/family environment seemingly has little or no effect on how our personality characteristics turn out. As far as twin studies are concerned, non-genetic influences on personality appear to be almost all from non-shared environmental factors (plus measurement error; Turkheimer & Waldron, 2000). However, the preschool experiments also described in Chapter 5 contradict this conclusion by showing that if children from disadvantaged, welfare-claiming families receive intensive tutoring in pro-employment behaviours before the age of five (especially planning, executing and reviewing tasks and resolving interpersonal conflicts), they turn out to be significantly less employment-resistant in their personality profiles and also more successful as adults than children from similarly disadvantaged families that did not receive the tutoring (Heckman et al., 2013).
Moreover, the preschool studies showed that it is the shared/family environm
ent influence that was doing the damage to the personalities of the deprived children, as it turns out that home-delivered special educational tutoring was not effective, in contrast to tutoring delivered in an education centre. Therefore, in order to have a good chance of developing the personality of a solid citizen, the child needs to be removed from their dysfunctional family environment and placed in the ordered, constructive environment of a special education centre. In other words, special educational tutoring, as long as it is delivered in a dedicated centre, is effectively stepping in and partially making up for the personality damage caused by growing up in a dysfunctional, impoverished welfare-claiming family with parents that themselves tend to possess mis-developed, employment-resistant personalities.
The twin and preschool studies summarised in this book therefore present a seemingly confused and contradictory picture of environmental effects on personality: twin studies indicate that the shared/family environment has almost no effect on personality, but preschool studies suggest that a disadvantaged family environment warps personality towards employment-resistance. How can they be reconciled?
One plausible explanation is provided by the observation that disadvantaged families rarely participate in twin studies (Turkheimer et al., 2003). Conversely, preschool studies do not recruit participants from middle-class families because the focus of this type of research is to understand disadvantage (Heckman, 2006). This difference in participant type between the two types of research is important to the theme of this book because parental effectiveness differs between families at different levels of the socio-economic ladder (Lykken, 1998). Briefly stated, in affluent middle-class families there is relatively little variance in parental effectiveness and so the vast majority of affluent middle-class children will have the opportunity to express whatever genetic tendencies they possess. For example, almost all children in affluent middle-class households will be taught to read, will learn mathematics, study science, take music lessons, play sport, debate the day’s news over dinner and so on. Overall, this will tend to exaggerate the influence upon subsequent behaviour of genetic factors and unique (non-shared) experiences, but downplay the effects of the shared (family) environment.
In contrast, children who grow up in disadvantaged households tend to have a much more uneven exposure to these important developmental stimuli: some will be taught to read, some won’t. Some will have access to a musical instrument, some won’t. Some will get coached in sport, some won’t. Some will be given a computer, some won’t. This greater variation in parental effectiveness in disadvantaged households (relative to middle-class households) hypothetically means that individual differences in psychological characteristics are more strongly influenced by the specific family environment in children raised in disadvantaged households than in children raised in affluent, middle-class households.
This hypothesis is backed up by studies that show shared environmental experiences play a larger role in the variations in IQ in lower SES children than in high SES children (for example, Turkheimer et al., 2003; Hanscombe et al., 2012). The issue of SES modification has not been as widely studied with regard to personality, but results thus far are supportive. For example, Tuvblad, Grann and Lichtenstein (2006) investigated the moderating effect of SES on the family environment’s capacity to develop antisocial traits in 1,133 Swedish twin pairs, aged 16–17 years. This study found that the more disadvantaged the family, the stronger the influence of the family environment on antisocial behaviour in the offspring. We would thus expect that family/shared environmental effects on personality would be minimal in twin studies but highly pronounced in the preschool studies, which is what we find.
This insight is important to the welfare trait theory because it counters the argument that welfare policy may indeed damage personality, but the size of the effect will be miniscule. More specifically, a critic might admit that there are detrimental upbringing effects on conscientiousness and agreeableness amongst the offspring of welfare recipients, but might then claim that such effects will be too small to worry about because twin studies show that family environment effects on personality development are small. Furthermore, a critic could argue that the effects of disadvantaged family circumstances on personality cannot be large because, if they were, the population-level link between personality and fertility would be much larger than the low levels shown in the studies cited in Chapter 4.
Because we now know that the more disadvantaged the family, the larger the effect of the family environment on antisocial traits in the offspring, we can see that these criticisms do not stand up to close scrutiny. Both the small size of family environment effects on personality development and the weak links between personality and fertility can be explained as artefacts of the tendency for middle-class families to participate in research projects (Lykken, 1998). As we have already seen, when researchers include disadvantaged families in their studies, they tend to find not only that there is a large effect of family environment on personality (Heckman et al., 2013), but also that there is a strong link between low conscientiousness/agreeableness and high fertility (Tonge et al., 1975).
Conclusion
Human parents transmit their personality attributes to their children by genetic and environmental means, with children in affluent families tending to display greater genetic influence and children in disadvantaged families displaying greater influence from the family environment.
8
A Model of How the Welfare State Leads to Personality Mis-Development
Up until this point in the book, we have primarily been preoccupied with laying out the evidential building blocks of the welfare trait theory: now it is time to assemble those blocks of evidence into a coherent model of welfare-induced personality mis-development, beginning by reminding ourselves of what this book is about. This book is about personality and the welfare state. It’s about our attitude to work and the factors that shape that attitude. But most of all, it is about asking how the welfare state might alter the personality profile of the population to any significant degree.
As we know from our own experiences in the workplace, as well as decades of scientific research, our employment prospects don’t just depend on how intelligent we are, but also on how keen we are to turn up on time at work, to do what we are told, to be polite to customers, to cooperate with colleagues, to obey workplace regulations and so on. In short, our employment prospects are influenced by our scores on two dimensions of personality that, in the jargon of personality research, are usually labelled as conscientiousness and agreeableness. Briefly stated, people with personality profiles that are relatively lacking in conscientiousness and agreeableness tend to do worse than average in the workplace (Barrick et al., 2001; Hogan, 2011). In line with this notion, people with this particular combination of personality characteristics – what I call the employment-resistant personality profile – are over-represented amongst welfare claimants (for example, Caspi et al., 1998). Studies suggest that part of this personality difference between welfare claimants and working citizens is likely to be a result of the demoralising effects of unemployment (for example, Boyce et al., 2015), but we know that part of it is caused by these personality traits themselves, since longitudinal research shows that low levels of conscientiousness and agreeableness as measured in childhood are predictive of unemployment in adulthood (for example, Moffitt et al., 2011).
In the previous three chapters, we saw evidence that personality characteristics are transmitted from parent to child via both environmental and genetic channels. The over-representation of employment-resistant individuals in the welfare-claiming sector of the population therefore suggests that a welfare state which increases the number of children born to claimants risks proliferating the employment-resistant personality profile, because these children will tend to take after their parents in personality terms for both environmental and genetic reasons. This seems like an unfortunate situation, but scarcely a reason for hitting the panic button, b
ecause thus far we have not estimated the size of the problem caused by welfare-induced personality mis-development. The primary purpose of this chapter is to estimate whether welfare-induced personality mis-development is likely to be a large enough problem to deserve the attention of policymakers.
We will begin by revisiting the Perry Preschool Project which we first examined in detail in Chapter 5. This time, instead of comparing the life outcomes of the two groups of Perry Preschool participants (tutored versus untutored), we will compare their life outcomes to those of average individuals from approximately the same sector of society (African Americans who were approximately 40 years old in 2004). Such a comparison is very much a blunt instrument, but it will allow us to obtain an initial, tentative estimate of the magnitude of the effects on life outcomes of childhood disadvantage in the population as a whole.