A new study has found that tests producing “polygenic” genetic scores cannot be used to accurately predict pupils’ later academic achievements and “shouldn’t be used to design a genetically personalised curriculum or tailor teaching” for them.
The study, carried out by the University of Bristol, which compared the genetic scores of 3,500 pupils born in the 1990s with their exam results at age 7, 11, 14 and 16, found that “genetic data did not predict how the children would perform throughout their time at school as accurately as more general information about their family background and other socioeconomic factors”.
Related: Base education around pupils’ genetics, schools advised
Opinion: ‘Education isn’t about genetics: it’s about hope’
Long read: ‘Genetics might predict intelligence, but every child still benefits from an academic education’
Noting that some researchers and politicians now suggest individuals’ genetic data should be collected by organisations to personalise their education, the researchers added that, currently, “it appears that DNA would be less useful for personalising education than easily measured information like test results taken in primary school, education of the child’s parents and other social data”.
The study found that genetic tests which produce “polygenic scores” could be used to modestly predict educational achievement, but that when prior attainment was accounted for, polygenic scores were of “little utility to providing accurate predictions of a child’s future achievement”, while parental levels of education and socioeconomic factors were “more strongly predictive of achievement than a pupil’s genome”.
The report noted that some experts advocate for the use of polygenic tests to personalise learning for students but said the scores should not be used in this way.
In October 2019, Professor Robert Plomin, a geneticist from King’s College London, told an audience of private school heads that genetic testing could be used to “predict or prevent” educational problems through providing pupils with personalised learning based on their test scores.
Today, Professor Plomin told Tes that he disputed the interpretation of data in the Bristol study and that the variance in individuals’ attainment determined by DNA was 15 per cent, which he said was very significant.
“It’s weird they’d take one of the exciting findings we have and say ‘Oh well’. Here’s a shocker: no predictions are perfect in behavioural sciences,” he said.
“Now, it is true that if you take kids’ achievement in school that predicts kids’ achievement pretty well but DNA predictors are unique, you can predict from birth and from infancy who might have trouble at school.”
“Schools are going to have to face this one way or another - if we can predict reading disability, we can move away from treating disorders and preventing them. DNA is the only predictor you have for a very young child and it doesn’t change. It’s unique and it’s not biased - you can’t get an expensive tutor for this, you can’t change your DNA score.”
“Like it or not, this will impact on education,” he said.
The Bristol study showed that those children with higher polygenic scores had higher exam scores than average than children with lower scores, but that while they could be used to identify group-level differences, they had “no practical use” for predicting the attainment of individuals or for personalised education.
Dr Tim Morris, the study’s lead author, said: “Our analysis shows that some pupils with a low polygenic score are very high performers at age 16.
“Some of those who would be predicted from their genes to be in the bottom 5 per cent are actually in the top 5 per cent of performers. This contradicts the notion that it is possible to accurately predict how well any one child will perform in education from their DNA.”
He added: “Educational achievement is incredibly complex and influenced by a large range of factors including parental characteristics, family environment, personality, intelligence, genetics, teachers, peers and schools, and - often overlooked - chance or random events.
“This complexity will make it perhaps irresolvably difficult to accurately predict how well any one pupil will do from their DNA.”
“The best piece of information we currently have for predicting how well a pupil will perform is how well they did in school earlier in childhood.
“Where we don’t know this, such as at the start of schooling, we can make better predictions about a pupil’s future educational performance by knowing how educated their parents are than by knowing their DNA.”
Can education be personalised using pupils’ genetic data? is by Tim Morris, Neil Davies and George Davey Smith from the University of Bristol
