Are Children Less Like Their Parents Than We Thought?

On April 23, the Times ran an article with the title ‘Children are less like their parents than we thought, study finds’. According to that article, “the study suggests that people inherit surprisingly little of their personality –– so much so that parents and their children are only slightly more likely to resemble each other in temperament than pairs of strangers are”.

This is a very odd summary. Why? Well, the study in question was titled ‘Familial Transmission of Personality Traits and Life Satisfaction Is Much Higher Than Shown in Typical Single-Method Studies’. In other words, its main finding could be characterised as ‘children are more like their parents than we thought’.

So what’s going on?

René Mõttus and colleagues begin by noting that parent-offspring and sibling-sibling correlations for personality traits are usually based on self-report data. This means that researchers usually estimate the degree to which relatives resemble one another by asking each relative to fill out a questionnaire about their own personality.

However, the use of self-report data may lead to underestimation of parent-offspring and sibling-sibling correlations due to relatives using different reference groups for comparison, or having different interpretations of survey items. A father and his son may be similarly extraverted (by some objective standard) but the father may, for whatever reason, rate himself as less extraverted. For example, he may have spent more time with highly extraverted people during the course of his life.

In order to get round this problem, Mõttus and colleagues analysed data on the personality traits of relatives from two different sources: self-reports and ratings by third parties (mostly spouses/partners). So every relative’s personality was rated twice, once by themself and once by somebody else. This gave the researchers more accurate measurements of personality that are less affected by the idiosyncrasies of individual raters.

When they ran their statistical analyses using these more accurate measurements, they observed higher correlations than have been reported by previous studies. In other words, they found that relatives are more similar with respect to personality than those studies have suggested.

Specifically, they obtained estimates around r = .20, as compared to typical estimates of around r = .15 – which means that theirs are about a third higher. They also obtained higher estimates of heritability (which quantifies the genetic contribution to individual differences in personality). Interestingly, the latter finding is noted in the Times article, despite the fact that it basically contradicts the title.

So where did the bizarre title come from? Well, Mõttus and colleagues state in their conclusion that it is still “impossible to accurately predict a child’s personality traits from those of their mother or father”. Note: they are not saying it is impossible to predict a child’s personality traits, only that it is impossible to do so “accurately” (a somewhat subjective judgement).

To sum up: a new study was published (as a pre-print), and rather than reporting its main finding, the Times decided to emphasise a brief section in the conclusion, thereby giving the impression the study found the opposite of what it actually did.