Welfare critic under fire after correcting ‘shockingly bad’ errors

An academic who argues that the welfare state will stimulate the growth of an “employment-resistant personality profile” by increasing the number of children born into unemployed households has published corrections to one of his papers, prompting one critic to label his work “junk science”.

However, Adam Perkins, author of the 2016 book The Welfare Trait: How State Benefits Affect Personality, said that despite the errors in the paper – which meant that some results were overstated by a factor of 10 at one point – the findings “remain statistically significant, hence the substance of the paper is unaffected”.

The 2013 paper, “Personality and occupational markers of ‘solid citizenship’ are associated with having fewer children”, was published in Elsevier’s Personality and Individual Differences. Dr Perkins, lecturer in the neurobiology of personality at King’s College London, and five co-authors looked at two existing studies, including one on the honourable/non-honourable discharge status of about 15,000 men who served in the US military during the late 1960s.

The authors found that the non-honourably discharged men “fathered significantly more children and also experienced significantly greater unemployment” than those honourably discharged.

But a corrigendum, published last month, makes clear that, in the summary of results in the main text, the interpretation of adjusted R-squared, a measure of goodness of fit, was overstated by a factor of 10 on two occasions.

For example, adjusted R-squared in the model predicting risk of unemployment in relation to discharge status was stated as “.031 or 31 per cent of the variance in risk of unemployment” in the original paper. But this should be changed to “.031 or 3.1 per cent”, the corrigendum says.

The corrigendum also corrects a number of references in tables to “p < 0.000” to “p < 0.001”.

Dr Perkins and his co-authors concluded in the paper that their results “provide support for previous findings…that the human personality profile in developed countries such as the USA may be gradually evolving by natural selection towards lower levels of personality traits that predispose an individual to be a solid citizen”.

In 2016, a lecture by Dr Perkins at the London School of Economics was postponed after fears that it might be disrupted by disability activists, before later going ahead.

Toby Young wrote in The Spectator that year that Dr Perkins had become a victim of “liberal McCarthy-ism”. Jenni Russell, writing in The Times, praised Dr Perkins for “at least daring to think the unthinkable”.

Andy Fugard, senior lecturer in social science research methods at Birkbeck, University of London, was the first to highlight the errors. He said that it was “shockingly bad” that they had made it into a peer-reviewed journal.

“The relationships between the variables that he [Dr Perkins] looks at are just tiny,” Dr Fugard added. “Because the data that he used had such a big sample size, then basically pretty much everything will be statistically significant.”

Dr Fugard said that Dr Perkins “leans a lot” on the findings of this paper for the arguments in The Welfare Trait.

Jonathan Portes, professor of economics and public policy at King’s, said that the paper’s conclusions were drawn “on the basis of the most astonishingly flimsy, indirect and, it turns out, flatly erroneous evidence”.

He added that some journalists “appear to be quite happy to validate and promote obviously junk science when it supports their ideological position”.

Dr Perkins said that “the key point to note is that the reported output” from the analysis “is correct, it was the summing-up that required correction”.

On the suggestion that the errors undermine The Welfare Trait, Dr Perkins said that “the results of the paper remain statistically significant” and that, aside from those results, “we still have a range of evidence showing that there are significant links between personality, employability and reproduction”.

john.morgan@timeshighereducation.com