Is psychology a voodoo science?

We are in the middle of a remarkable explosion of psychology as news, yet many of the results and studies reported are less reliable than they seem.

Sign Up

Get the New Statesman's Morning Call email.

It will, no doubt, be a cause for wry amusement that, according to recent research, men’s brains and women’s brains do not differ after all. Any thought of giving someone for Christmas Men are from Mars, Women are from Venus, now goes out of the window. But hold on. What has been much less well reported is that, only twenty-four hours after this "latest research" was published, other "research" showed that women’s brains do not deteriorate in old age as fast as men’s brains do. So there seems to be some difference there that muddies the picture quite a bit.

Not long before these researches were released, another psychological survey identified what it called "frequent flyer" patients, who know the ropes of the health services all too well and, as a result, demand unnecessary antibiotics, a form of assertive behaviour which has fuelled a dangerous rise in drug-resistant bugs. Research suggests that nine out of ten GPs feel pressured to prescribe antibiotics and that "pushy patients" are a new phenomenon which needs to be challenged by doctors.

On the very same date that this report appeared, however, another survey – carried out by the charity, Ovarian Cancer Action – showed that, rather than being "pushy", more than half of young women in Britain are too shy to discuss sex with their GPs. These young women were much more hesitant than older women "about saying orgasm, labia, vagina or discharge". Women under 25 were four times less likely to visit a doctor to discuss a sexual health problem than were 55-to-64-year-olds, and were generally speaking scared of gynaecological examinations. "A quarter said they simply did not know the words they would use." As one official was moved to remark: "Saying 'vagina' won’t kill you, but avoiding saying it could."

These confusions, paradoxes and ironies might not matter so much but for the fact that we are in the middle of a remarkable explosion of psychology as news. On just one day recently, the Times carried no fewer than five psychological stories, what we might call "news from inside". These were: how the physical appearance of your first boyfriend or girlfriend determines who you later fall for; a new psychological condition called "bore-out", especially prevalent in France (where there are so many state jobs), where people become clinically depressed through having boring occupations;  how depressed children are increasingly seeking help on the web because they have lost confidence in their GPs or school nurses, to such an extent that mental illness among the young is beginning to be accepted as normal; that it’s lonelier at work than in retirement; and that cheating by parents to obtain a place in a good school for their children (by lying about where they live) tripled between 2012 and 2015. On the following weekend the Times’s sister paper, the Sunday Times, had six articles about our current psychological lives and how they are changing.

There is nothing inherently wrong with this new category of news, which has been gradually making itself known over the past months. But it helps our sanity to understand that many of its "findings" need, quite frankly, to be swallowed (if at all) with a large helping of salt. One of the reasons for this is that the research – such as it is – is very varied, not just in subject matter but in quality. Some of it is carried out by professional psychologists, some by polling organisations, some by charities (who realise that provocative "news" about this or that aspect of our behaviour is more likely to get them media coverage than anything else), and some is carried out by government or education authorities who also have an interest in attracting attention to their initiatives.

As a result, the research – copious as it is – is patchy, random and piecemeal in its range, and frequently suspect methodologically (very small samples of twenty or forty people, over-simple statistical analyses, ill-thought-out concepts). From which it follows that conclusions are often inconsistent and in several cases, as we have seen, paradoxical and contradictory.

But it doesn’t stop there. The trustworthiness – or rather the untrustworthiness – of psychological research has come under much more heavyweight attack than this. Doubt was first aired in 2013, when John Ionnidis, an epidemiologist at Stanford Medical School in California, released a celebrated paper arguing that neuroscience, what we might think of as a "hard" form of psychology – in particular since the invention of fMRI brain-scanning techniques in the early 1990s – has produced a vast number of what are termed "voodoo correlations ... highly implausible results strongly linking behaviours or traits to one or just a few specific areas of the cortex." This phrase, "voodoo correlations", was in fact coined by his colleagues, Edward Vul and Harold Pashler. Inspecting 53 high-profile neuroscience studies, they concluded that half of them were untrustworthy, "too good to be true" and "seriously defective" in their methodology.

In another study, of 134 published psychological surveys, as many as 42 per cent were found to have made crucial methodological errors. One of the errors found to be most widespread was the fact that scientists tend not to report – or even write up – negative findings or ‘null’ results, results which do not point either way. In one survey, it was found that fully 62 per cent of null results were not even written up. This, of course, can have a marked effect on statistics.

More heavyweight still, a report in Science, published in August this year, was entitled, baldly, "Many psychology papers fail replication test". It reported a remarkably orchestrated experiment in which no fewer than 270 researchers from all over the world (Germany, the United States, Great Britain, Australia, Canada, Hong Kong, Paraguay, Holland, Sweden etc, etc) had repeated exactly 100 psychological experiments published in three journals, Psychological Science, the Journal of Personality and Social Psychology and the Journal of Experimental Psychology: Learning, Memory and Cognition. In other words, all these studies had been originally published in well-respected peer-reviewed journals.

The main results were sobering. First, it was found that the "average mean effect size" in the replications was half the magnitude of the average effect size in the original experiments. In other words, for example, if a teaching technique produced an improvement in learning of, say, 12 per cent in the original experiment, the improvement in the replication was, on average, around 6 per cent. Second, 97 per cent of the original experiments were judged as statistically significant in the original experiments. In the replications, however, only 36 per cent were judged to be significant statistically. And third, while 47 per cent of the original effect sizes were judged to be significant at the 95 per cent confidence level (a statistical device, generally regarded as the probability level required for some effect to be reliably established), only 39 per cent of results in the replications were judged to be established at the 95 per cent confidence level. These were average figures – many psychological experiments failed to be replicated at all.

There is a massive appetite for psychological news – of our emotional, social, moral and intellectual lives. We are more interested in ourselves than in anything else. But not to the point where we should believe everything "research" tells us. Male brains and female brains should be equally sceptical.

Peter Watson’s latest book is: "The Age of Nothing: How We Have Sought to Live Since the Death of God"