Keith E Rice's Integrated SocioPsychology Blog & Pages

Updated: 19 January 2018

David Rosenhan’s classic 1973 study remains one of the most important in the field of mental health. For all its faults – and there are a number of serious ones – it almost encapsulates the difficulties in trying to determine if someone is mentally ill. The last half of the 20th Century and the first decades of the 21st have witnessed virtual ‘epidemics’ of ‘mental illness’ inflicting themselves upon the Western world. These ‘epidemics’ have, in turn, spawned huge industries in mental health care and pharmacology – and there is increasing concern about the role of the pharmaceutical industry in influencing what is classified as mental illness and how such illnesses are treated. Lisa Cosgrove &  Sheldon Krimsky’s 2012 expose is just one of many focusing on just how many of the authors of the Diagnostical & Statistical Manual of Mental Disorders – latest version DSM-5 (2015) – more and more have financial interests in the pharmaceutical industry. Thus, to some, it looks like the ORANGE vMEME’s desire for profit is driving changes in DSM – the medicalisation of mental illness that can be cured, or at least managed, by chemotherapy. Opponents to this direction tend to favour GREEN’s motif of holistic care, with emphasis on ‘talking therapies’ and care-in-the-community.

In these increasingly-embittered contentions, Rosenhan’s study is more relevant than ever and, thus, is given its own in-depth description and analysis here.

It should be noted that David Rosenhan was a leading expert on the relationship between Psychology and the law and an advocate for better treatment of those with mental health problems. He was also concerned about psychological categorization of mental illness and the way such categorisations were being used. He wrote that they were “useless at best, and downright harmful, misleading, and pejorative at worst. Psychiatric diagnoses, in this view are not valid summaries of characteristics displayed by the observed.”

So, it can be argued there could have been some bias  in his research….

For his 1973 study Rosenhan aimed to investigate whether psychiatrists could distinguish the difference between people who are genuinely mentally ill (‘insane’) and those who aren’t (‘sane’), using the then-current DSM-II (1968).  He aimed to find out if “the salient characteristics that lead to diagnoses reside in the patients themselves or in the environments and contexts in which the observers find them?” He also wanted to learn more about what it was like on a psychiatric ward from the patient’s viewpoint. (Researchers had, on previous occasions, spent time on psychiatric wards carrying out non-participant overt observations which made it unlikely they would see anything staff didn’t want them to see – a form of demand characteristic.driven by social desirability.

The first study – diagnosis and admission
A group of 8 ‘normal’  people attempted to gain admission to 12 different hospitals in 5 different states on the American east and west coasts. The hospitals varied from old to new; some had good staff-patient ratios while others were understaffed; some were research based and one was a private hospital.  Rosenhan’s ‘pseudopatients’ consisted of 3 women and 5 men – a 20-year-old graduate, 2 psychologists, a paediatrician, a psychiatrist, a painter and a housewife, with the eighth being Rosenhan himself. (Rosenhan reputedly recruited the pseudopatients by ringing around his friends who included the noted psychologist Martin E P Seligman.) Obviously, from the numbers, some of the pseudopatients presented themselves more than once.

The pseudopatients called the hospital for an appointment, stating that they were hearing voices which were of the same sex and, whilst slightly unclear, seemed to be saying “empty”, “‘hollow” and “thud”. These symptoms were purposefully chosen for their similarity to existential symptoms (the alleged meaninglessness of life) and their absence in the psychiatric literature. Apart from falsifying symptoms, name and employment, no further pretences were made. They all employed pseudonyms and those in the mental health profession gave another occupation to avoid embarrassment to colleagues. They described their childhoods, life events, relationships and, for those who weren’t involved in mental health, occupations accurately.  None of them had any prior history of pathological behaviour.

All the pseudopatients were admitted, except in one case, with a diagnosis of Schizophrenia. The 12th diagnosis was Manic Depression (aka Bipolar Disorder). Rosenhan was the first pseudopatient to be admitted, having confided only to the hospital administrator and the chief psychologist there. In all other 11 cases the hospital authorities were not informed.

The diagnostic element of the study could be classified as a field experiment, with the symptoms described being the independent variable and the diagnosis the dependent variable. As such, the 92% reliability of the findings over the 12 trials can be considered truly remarkable. However, the findings were completely lacking in validity as none of the pseudopatients was mentally ill.

The lengths Rosenhan went to control the study is illustrated by him discarding the data from a 9th pseudopatient because, amongst other things, he did not stick closely enough to his brief.

The first study – life as a psychiatric patient
The  pseudopatients, like a true psychiatric patient, entered a hospital with no knowledge of a discharge date. (Each had been told that they would have to get out by convincing the staff that they were sane. A  lawyer was retained to be on call for emergencies if it became clear that a pseudopatient would not ever be released voluntarily. ) The length of hospitalisation ranged from 7 to 52 days, with an average of 19 days.

Following admission to hospital no further symptoms were ‘acted out’ and, other than some understandable nervousness and tension about being found out, the pseudopatients behaved perfectly normal. As all but one found the experience stressful and wanted to be released, they co-operated with staff and obeyed instructions. If asked how they felt, they said they felt fine and no longer experienced any symptoms. In some cases the pseudopatients approached staff and said: “Pardon me, can you tell me when I am likely to be discharged?” (Care was taken never to approach a particular member of staff more than once a day to avoid irritating them or arousing their suspicions.)

The pseudo-patients spent their time talking to the other patients and making notes of observations of patients and staff on the ward. Initially the notes were done in secret but it soon became apparent that the staff did not really care. The closest to being questioned about the note-taking was when one doctor told the pseudopatient making a note of his medication: “You needn’t write it. If you have trouble remembering, just ask me again.” Some notice clearly was taken, though, as nursing records for 3 pseudopatients indicate that their writing was seen as an aspect of their pathological behaviour (‘patient engages in writing behaviour”).

During the research the pseudopatients were given a total of 2,100 tablets – including Elavil, Stelazine and Compazine – which they did not swallow but flushed down the toilet. None of the psychiatrists or nurses noticed that the medication was not being taken. Average daily contact with the psychiatrists was an average of 6.8 minutes per day (based on data from 6 patients over 129 days of hospitalisation).

Whilst at the hospital a  number of ‘real’ patients regularly voiced their suspicions. 35 (30%) out of 118 patients made statements to the first 3 pseudopatients such as “you’re not crazy – you’re a journalist’” or “you’re a professor checking up on the hospital”. However, in no case did doctors or nurses notice there was nothing wrong with the pseudopatients. This lack of insight may have been influenced by the fact that doctors (71%) and nurses (88%) mostly ignored the pseudopatients when they they tried to speak to them. The staff at the hospitals tended to interact only minimally with patients – for caretaking purposes, to issue medication, conduct therapy or group meetings, instruct and reprimand patients. According to Rosenhan, staff on average only spent 11.3% of their time outside their glassed quarters patients referred to as ‘the cage’ – the nurses interacted  with the patients 6.7 times on average per day. The table below shows the full results of attempts by pseudopatients to initiate interaction with staff. (The table also includes data from Stanford University (‘Faculty’) where Rosenhan, for comparison purposes, had a young woman approach individual busy-looking faculty members and medical centre staff to ask them 6 questions, including how to get to various parts of the campus. All the questions were answered fully, without being rushed and with good eye contact.)

11 of the peeudopatients were discharged with a diagnosis of Schizophrenia-in-Remission – the twelfth discharge diagnosis being Manic Depression-in-Remission.

It was clear from comments made by staff and from records viewed retrospectively that perceptions of behaviour were coloured by the diagnosis of the pseudopatient’s condition Having once been labelled schizophrenic, there was nothing the pseudopatients could do to overcome the label.  Eg: one pseudopatient was described by Rosenhan as: “…had had a close relationship with his mother, but was rather remote from his father during early childhood. But during adolescence and beyond, his father became a close friend while his relationship with his mother cooled. His present relationship with his wife was close and warm; apart from occasional angry exchanges, friction was minimal. The children had rarely been spanked.” Yet the hospital record described him as: “This white 39-year-old male…manifests a long history of considerable ambivalence in close relationships, which begins in early childhood. A warm relationship with his mother cools  during adolescence. A distant relationship with his father is described as becoming very intense. Affective stability is absent. His attempts to control emotionality with his wife and children are punctuated by angry outbursts and, in the case of the children, spankings. And, while he says that he has several good friends, one senses considerable ambivalence embedded in these relationships also….”

Many normal behaviours were overlooked completely or profoundly misinterpreted. Eg: waiting outside the canteen before lunch was interpreted by one psychiatrist as ‘Oral-Acquisitive Syndrome’. Pacing the corridors out of boredom was seen as ‘nervousness’.

Rosenhan’s pseudopatients also recorded a number of instances of maltreatment, ranging from lack of privacy – eg: initial examination in a semi-public room and records available to anyone who wanted to look at them – to verbal abuse to physical beatings (in front of other patients but not other staff). The pseudopatients felt this treatment led to feelings of powerlessness and a loss of identity which could again be interpreted as symptoms of their ‘illness’ rather than a response to the social role they had been given. In other words, their behaviour was dispositional – to do with them – and not a situational product of environmental expectations (memes).

This part of the study could be considered a covert participant observation.

The follow-up study
Senior staff in a research & teaching hospital that had not received any pseudopatients had somehow got wind of Rosenhan’s study. Offended by what they perceived as a slur on their profession, they claimed that it could not happen in their hospital. Rosenhan informed them that in the next 3 months, one or more pseudopatients would present themselves. The staff were asked to rate, on a 10-point scale, their confidence level that the participant was genuine. Over the years many commentators have expressed real surprise that the hospital agreed to this. As Stephen Ginn (2010) writes: “As for the non-existent impostor experiment it is surprising that it was agreed to by the teaching hospital in question.  The poor reliability of psychiatric diagnoses means that the design of the experiment could only produce an outcome where actual patients were incorrectly identified as pseudopatients.” One can only assume some real no-consequences RED was driving the hospital management decisions!

Judgements were obtained on 193 patients admitted for psychiatric treatment. Of 193 admitted, 23 were thought to be a pseudopatient by at least one psychiatrist. 41 patients were thought to be pseudopatients by at least one member of staff and 19 by one psychiatrist and one other member of staff. In fact, Rosenhan did not send any pseudopatients to the hospital.

This study could be classed as a structured observation.

Rosenhan’s conclusions
Rosenhan wote: “It is clear that we cannot distinguish the sane from the insane in psychiatric hospitals. The hospital itself imposes a special environment in which the meaning of behaviour can easily be misunderstood. The consequences to patients hospitalised in such an environment – the powerlessness, depersonalisation, segregation, mortification and self-labelling – seem undoubtedly counter-therapeutic…. Once labelled schizophrenic, there is nothing the pseudopatient can do to overcome the tag.”

Rosenhan concluded that the failure to detect sanity during the course of hospitalisation may be due to the fact that the doctors were showing a strong Type 2 error – that is the physicians were more inclined to call a healthy person sick (a false positive, Type 2) than a sick person healthy (a false negative, Type 1). Rosenhan considered the consequences of turning away a sick person may have influenced the psychiatrists’ readiness to diagnose mental illness.

With regard to the 23 patients thought to be sane in the follow-up study, Rosenhan speculated that, in the course of avoiding a Type 2 error, the staff may have made a Type 1 error and called “crazy people sane”. In other words, the potential publicity from Rosenhan’s study and his threat to send pseudopatients influenced the psychiatrists towards non-diagnosis of mental illness.

From these results, Rosenhan drew the conclusion that it is indeed difficult to determine who can be categorised as ‘sane’ or ‘insane’, based on the diagnostic criteria of DSM-II.

Evaluation and criticisms
There were a number of positives to the study. Firstly, the investigation was carried out in real hospitals, using real staff who were unaware of the study, thus giving it high ecological validity. As a range of hospitals were used from around the United States, it was reasonable to generalise the results to other psychiatric hospitals at the time. Finally, the study yielded both quantitative data – eg: the number of days the pseudopatients were kept in hospital – and qualitative data – eg: the notes the pseudopatients made about life on the wards – meaning it was both generalisable and detailed.

However, Seymour Ketty (1974) criticised Rosenhan’s deception from a validity perspective, saying: “If I were to drink a quart of blood and, concealing what I had done, had come to the emergency room of any hospital vomiting blood, the behaviour of the staff would be quite predictable. If they labelled and treated me as having a bleeding peptic ulcer, I doubt that I could argue convincingly that medical science did not know how to diagnose that condition.”  Ketty’s point was that psychiatrists will hardly expect someone to carry out a deception in order to be admitted to a psychiatric hospital. In other words, the study lacked mundane realism.
The fact that 11 of the 12 diagnoses were consistent – Schizophrenia – may actually indicate reliability – ie:  present the same symptoms, get the same diagnosis! Theodore Sarbin & James Mancuso’s (1980) riposte to Ketty, using his example,  is to ask whether, when there was no sign of bleeding the next day and tests were all negative, would it be appropriate to discharge the patient with a diagnosis of ‘Bleeding Peptic Ulcer in Remission’?

Robert L Spitzer (1976) notes that the diagnosis ‘Schizophrenia-in Remission’ is extremely rare. He examined the records of discharged schizophrenic patients in both his own hospital and 12 other American hospitals and found that in 11 cases ‘Schizophrenia-in-Remission’ was either never used or used only for a handful of patients each year. Therefore, Spitzer claims the psychiatrists’ discharge diagnosis was, in fact, due to how the pseudopatients behaved and not to the fact the psychiatrists couldn’t tell they were normal.

Although the pseudopatients attempted to record their observations objectively, almost inevitably there would be some subjectivity influenced by their emotions, especially given the treatment some of them and real patients were given on the wards.

There are significant ethical issues with the study as the hospital staff were deceived about the patients’ symptoms. Consequently they could neither give their consent nor exercise their right to withdraw. However, Rosenhan did protect confidentiality – no staff or hospitals were named, thus minimising the risk of identification.

Another ethical issue with Rosenhan’s study is the crisis of public confidence in the American mental health system it aroused – which may have prevented people who genuinely needed help from seeking it.

Psychiatrists point out that the DSM has been revised several times since Rosenhan’s study and that diagnostic criteria have been tightened considerably. It can be argued accordingly that it is much less likely his pseudopatients could have fooled psychiatrists steeped in DSM-III (1980) as a characteristic hallucination was now required to be repeated several times. In DSM-IV (1994) hearing voices must be experienced for over a month before a diagnosis of Schizophrenia can be made.

Lauren Slater (2004) reported that she had attempted to replicate Rosenhan’s study by presenting herself at 9 psychiatric emergency rooms with the lone complaint of an isolated auditory hallucination (hearing the word ‘thud’). In almost all cases Slater was given a diagnosis of ‘Psychotic Depression’ and prescribed anti-psychotics or anti-depressants. (Ironically Slater had once been diagnosed with Depression!) Robert L Spitzer, Scott Lilienfeld & Michael Miller (2005) sought to challenge Slater’s findings by giving 74 emergency room psychiatrists her detailed case description and asking questions about diagnosis and treatment recommendations. Only 3 psychiatrists diagnosed ‘Psychotic Depression’ and only a third recommended medication.

G Langweiler & Michael Linden (1993) sent a trained pseudopatient to 4 physicians, each with a different professional background. Although the pseudopatient presented the exact same symptoms to the physicians, 4 different diagnoses were offered along with 4 different treatments, indicating real problems with reliability. This study illustrates the ongoing difficulties in developing accurate diagnostic criteria in mental health. The junking of sub-types of Schizophrenia and Schneiderian ‘first rank symptoms’ in DSM-5 – key indicators in diagnosing Schizophrenia for over 30 years – shows just how unreliable psychiatric diagnosis can be.

In which case, even if Rosenhans’s pseudopatients might not fool modern psychiatrists, the points made about difficulties in diagnosis are extremely well made and as relevant as ever. Ironically, though, Spitzer makes a key point which partially undermines Rosenhan: that ‘sane’ and ‘insane’ are legal concepts, not psychiatric ones, and that no psychiatrist would make a diagnosis of sanity or insanity.

As to the second of Rosenhan’s concerns – conditions on psychiatric wards – there have been many efforts to improve conditions on the wards over the years and, from the 1990s on, GREEN-driven campaigns to locate psychiatric patients in the community wherever possible. However, as Ginn concludes: “Rosenhan’s observational study of conditions on psychiatric wards…still has relevance today and remains a note of caution for anyone who works in mental health.”