Therapists - who needs them?

Feeling depressed? You might as well see an astrologer as a psychothera pist, according to a devastating critique of the 'talking cure'.

New research shows that many children who have problems at school and then become delinquents and/or addicts, have Saturn rising in the house of Neptune in their astrological chart. It doesn't of course, and even if it did no one would take any notice, because although we may think horoscopes are fun, we don't base social policy on them.

The reason astrology is not considered scientific is that it is not backed up by empirical research. Instead, we use the findings of psychology to explain why people behave in a deviant or undesirable way, and psychotherapy to help them change. But according to a new and devastating critique, psychotherapy has so ignored scientific research that we might as well pick out likely offenders by their star signs.

Take, for example, the current emphasis on self-esteem. It is just one target among many in House of Cards by Robyn Dawes, but it's a good place to start because almost all educators and psychotherapists believe that a host of social problems, such as learning difficulties and delinquency, are linked with low self-esteem. The theory sounds very plausible. If you don't think very much of yourself, you are not going to try hard at school and later you may get into drugs or promiscuous sex in a doomed attempt to make yourself feel better. But the result is a vicious spiral, and the only way out is therapy to help you uncover the root cause.

The trouble is, there is no evidence that humans actually function like that. "Several years ago the Californian State Assembly set up a commission to investigate the role of low self-esteem in social problems," says Professor Dawes, who is head of psychology at the University of Oregon. "They spent three years and several million dollars looking at over 30,000 research papers dealing with self-esteem, but were forced to conclude that the evidence wasn't there - even the apparently obvious link between low esteem and sexual abuse." He is particularly scathing about the way the authors of the report, all of whom believed the theory, tried to cling on to it despite the lack of supporting research.

It is not that Professor Dawes is against psychotherapy, in the sense of it being good to talk. In fact he did one of the major studies 20 years ago showing that patients in a group that got psychotherapy had twice the chance of feeling better off than those who didn't. However, his findings came with a nasty sting in the tale for the profession. Not only was there no correlation between the sort of therapy used and the cure rate, but who you talked to didn't make any difference either. Patients who saw a highly-trained senior psychotherapist did no better than those who visited a pleasant professor of something like chemistry or mathematics, who they thought was fully qualified.

More recently, Professor Dawes's department conducted another study to compare the performance of high school students and psychiatrists in predicting which patients on a ward in a mental hospital were likely to become violent. "Both groups were given a check list of symptoms - the sort of thing psychiatrists use to make decisions all the time - and the students made exactly the same prediction as the psychiatrists," he says. "But what's really interesting is that both groups got it wrong. They both simply predicted that the ones with the most severe symptoms were the most likely to be violent, which is inaccurate."

Many other studies have shown that psychotherapists are not particularly good at deciding how individuals are going to behave - which is highly damaging, because if they claim expertise in anything, it has to be that they have a special insight into the individual. They regularly make judgements about releasing patients and prisoners, decide who would respond to one type of treatment rather than another, or in court, whether a defendant is likely to have committed such a crime.

The other way of making such decisions is on a statistical basis - what do most people like this one do? It turns out that this approach is more accurate than the judgement of a psychotherapist based on personal experience of the person. For instance, if you want to know if someone is going to benefit from electro-shock treatment, you do better by simply knowing how long they have been disturbed, whether or not they are married and what they think about their problem, than by asking a psychotherapist to predict it. Similar findings apply to letting people out on parole.

In one study, four senior child abuse experts were shown case notes and various videotaped interviews with members of a family in which the father was accused of abuse. When asked to rate the probability that the child had been abused, their answers ranged from 5 per cent to 75 per cent. So what, you might say, experts in all fields disagree. But there's an important difference here. Psychotherapy, in theory, is not like dance or literary criticism, it's not ultimately a matter of personal judgement, but based on a body of knowledge about how humans behave and change.

In most professions, generally, the more experienced someone is, the more expert they are, but for psychotherapists there is a problem. We know quite a lot about how people learn, and one of the most important elements is quick and accurate feedback. "If a doctor makes a mistaken diagnosis it can often be checked against a blood test or a biopsy," says Professor Dawes. "But psychotherapists rarely get clear messages when they are wrong.

"Often they see a patient for a while and then have no idea what happened in the long term. What's more, they have no way of knowing if they did the right thing. Was staying with a poor mother better than being fostered? Can't tell. Then again, some of their predictions are self-fulfilling. Someone who is disturbed may become worse when put into hospital. Would he have been better out? Who knows."

But there's a deeper difficulty still. "Psychotherapists become experts by seeing a lot of one type of patient, but that only tells them about a highly selected group, from which they then generalise wildly" says Professor Dawes. "Take experts in sex abuse. We've seen they can't even agree on whether an individual is an abuser, then they make totally unsupported claims - like one in three or one in five women is abused. How can they possibly know? Or take the widespread claim that abusers never stop without therapy. They can't know that, because if abusers do stop without therapy, they never see them."

The experience problem is made worse by what we know about memory. "The few cases from which a therapist does actually learn that he was right will loom large in his memory and he'll become more confident of his judgement. But it's a totally biased perception," says Professor Dawes. In fact the issue of memory is another hugely problematic area for psychotherapy.

The whole field has recently been riven by the agonised battle between recovered versus false memory of childhood sexual abuse. Did clients really remember it, or did they create it under the therapist's prompting? Once again, Professor Dawes's point is that psychotherapists arrogate to themselves a knowledge and a certainty that they have no right to, and which is not based on research. Memory, especially of childhood, lies at the heart of the whole psychotherapeutic enterprise.

Ever since Freud, there has been a basic belief that our childhood years are crucial in defining the dynamics of the adult personality. But the only way we know about those years is through memory, and increasingly it is clear that memory is a creative act rather than a matter of pulling something out of a file. "We know that people when they are depressed remember their parents as being much more aloof than when they are feeling more cheerful," says Professor Dawes.

And it's not just individual moods that affect memory - cultural expectations do as well. In a recent study at Oregon, Professor Dawes explored the popular idea that women are more likely to be worried and anxious about relationships than men are. "We first of all asked people to remember what they had been stressed about in the past, and sure enough this male- female difference showed up. But then we did a study that asked another group three times a week what had been stressing them, and the difference disappeared.

"Men were just as likely to be stressed by relationships - they just didn't remember it so well because it wasn't expected."

House of Cards is an impressive indictment, and if science worked the way it is supposed to - modify your theories according to the evidence - psychotherapy would have to undergo an enormous shake-up. But it won't for at least two reasons, both uncovered in psychology laboratories. Firstly, because humans will do almost anything if told to in a sufficiently authoritative way, and psychotherapy gives an illusion of authority in a vague and difficult area, and secondly, because of something called "attributional bias". This is the technical term for the powerful tendency we all have to overemphasise the role of personality when explaining why other people do things - "He's angry/uptight/jealous" - and stress the importance of the situation when explaining ourselves - "I'd had a bad day, it was late, the traffic was terrible." Psychotherapy has turned this tendency into a whole system, which is why it feels so right even when it's wrongn