David Rosenhan is known for the classic, yet controversial study “On Being Sane in Insane Places” of progress within the mental health field. Rosenhan’s study (1973) of eight people with no previous history of mental illness were admitted at various mental hospitals in America and complained of individual symptoms (auditory illusions, e. g. , ‘thud’). He investigated whether psychiatrists could distinguish between those genuinely mentally ill and not. Each pseudopatient behaved normally, and symptoms were not re-reported. However, the average length of hospitalisation was 19 days.
This shows context has a powerful role in determining how behaviour is labelled. This led to question the truth in psychiatric diagnoses. The predominant issue was unauthorised diagnoses and needless treatments for a fictional mental illness tolerably accepted. Today, it is the difficulty in gaining treatment for real symptoms of mental disorders. Rosenhan’s hypothesis was clearly and precisely operationalised (defined variables) to easily test whether patients behaviour (independent variable, IV) influenced diagnoses (dependent variable, DV). The sample was staff (nurses and doctors) in the hospitals.
It allows us to understand unique views of professionals who offer psychological knowledge and diagnoses to patients. Study 1, a naturalistic observation found psychiatrists failed to detect the pseudopatients sanity due to Type 2 errors. It provided unique insights as behaviour was directly observed in its context. It is appropriate in the real world due to mundane realism therefore, highly ecologically valid. However, this method has difficulty controlling extraneous variables (EV’s), as it is likely that other significant phenomena may have affected objectivity.
A confounding variable (e. g. , within the hospital) may not have been recognised, thus effected the interpretations of the labels. Also, the participants were not randomly selected or allocated to conditions which may reduce validity, nevertheless, it is the only way to study certain behaviours. In study 2, staff made more type 1 errors, presumably in attempt to avoid type 2 errors. Rosenhan failed to explain as to why we make type 1 errors or type 2 errors more than the other, neither the effects of giving false diagnoses for the pseudopatients.
Participant observations enabled psychologists to study relevant phenomena that have detrimental effects. This is beneficial to mental health, thus increasing external validity (EV). The patients experience on the ward provided significant insights of what people do rather than claims, making it potentially more internally valid (1. V). However, observer bias may have occurred due to expectations of observers which effects the IV and reliability. Therefore, if replicated different findings may be found. Study 3, a field experiment, required patients to ask staff questions (e. g. ‘Pardon me, what is the likelihood of being discharged? ‘).
Only 0. 5% of the nurses and 4% of the psychiatrists stopped, and 2% in each group stopped to converse. This was conducted in the natural environment therefore, had control over the IV, the EV’s and artificiality. Ecologically validity is high as it can be generalised in real life settings (e. g. , other hospitals). Experimenter effects and demand characteristics were minimised as the staff and real patients participated unknowingly. However, the I. V may have been affected by the pseudopatients nervousness, as none had previously visited a psychiatric setting.
This is potentially a confounding variable, as this may have been viewed as abnormal, thus caused further observation for pseudopatients. The findings are limited, as it is not based on large sectors of the population or hospitals, thus decreases reliability. Participants varied in socio-economical and socio-cultural factors which may have influenced the study as real patients may differ (Rosenhan noted this). The hospital staff and real patients were deceived, this is undoubtedly unethical as honesty is an important ethical principle.
This could have been dealt with by gaining permission of an ethical committee or debriefing, but does not make ethically sound research. However, Rosenhan did not expose the names of hospitals or staff to eliminate any clues which might lead to identification, abiding the Data Protection Act 1998. This was to protect privacy and confidentiality. Rosenhan argued to gain the most empiricist data the study needed to be covert.
The study was culturally biased and ethnocentric as hospitals in America may operate differently to other countries. It cannot be generalised beyond the actual setting (e. . , British psychiatrists use ICD). Nevertheless, hospitals were in different states, old, new, research-orientated and not, well staffed, poorly staffed, and one privately funded. Results showed minor differences between institutions therefore, it is widely applicable and suggests the same findings would be found. However, the hospitals were ridiculed for misdiagnosis and offering treatment with ease. To note, perhaps Rosenhan was being harsh because if patients seek for psychiatric care it is the duty of hospitals to provide it.
Quantitative data is not time-consuming, simple to analyse (e. . e. g. , 6. 7 times per day, 11% staff spent outside quarters), and easily presented in different forms (e. g. , graph). However, this oversimplifies the complexities existing in human experience due to statistics (significant) but humanly insignificant. Qualitative data show the realistic complexities of human behaviour (e. g. , staff comments). Although, it is more difficult to detect patterns and draw conclusions due to large amounts of data. Rosenhan argued pseudopatients behaviour was highly influenced by being in an abnormal environment (e. g. ignored by most staff).
This supports nurture (nurture versus nature debate), as it shows that the environment shapes behaviour. However, Rosenhan did not consider that not all people will behave to the label accordingly, as some people can overcome it. The pseudopatients should have been treated individually (idiographic) but it was ignored as a nomothetic approach was used. The study is deterministic as Rosenhan argued people are controlled by the events and the people around. The fact that people have free will and choose how to behave was ignored.
Experimenter bias may have affected Rosenhan’s findings as it is likely that the pseudopatients views were biased on the treatment of the insane or real patients. This would cause the study to be subjective, thus decreasing the I. V. Findings may have been affected by the Hawthorne effect, as the real patients may have changed their behaviour due to awareness of being observed by researchers. Slater (2004) previously diagnosed with clinical depression, presented herself at nine different psychiatric rooms with complaints (auditory hallucinations).
Slater was mostly rediagnosed and prescribed with antipsychotics or antidepressants. This supports Rosenhan as it shows that hospitals are able to distinguish the sane from the insane. Spitzer et al. (1975) gave 74 psychiatrists a detailed description from Slater’s clinical depression book and asked questions regarding diagnosis and treatment. Only three psychiatrists offered a diagnosis and only one-third recommended medication. This challenges Rosenhan and Slater findings as it does not match, thus effecting the validity and reliability. Sarbin and Mancuso (1980) argued psychiatrists using DSM-III. ould not diagnose the pseudopatients with schizophrenia as hallucinations must be reported several times.
Andre (1999) found DSM IV and ICD manuals reliable and shared a consensus of 68%. Amongst 1,500 people, only 35% agreed on what was post-traumatic stress disorder. This indicates a problem still with diagnoses and challenges Rosenhan. However, reliability of diagnostic procedures have improved since which mean errors may be eliminated. Scheff (1966) showed that psychiatric labels remain stigmatic for patients in a way medical labels do not, resulting in self-fulfilling prophecies (SFP) (e. . , the normal writing behaviour viewed as abnormal).
Rosenthal and Jacobsen (1968) found children made greater progress based on treatment by teachers (e. g. , high expectations, labelling of intelligence). This verifies rosenhan’s argument of how crucial labelling is for SFP. Neisser (1973) studied the irreversibility of labels, noting psychiatrists can never be incorrect because diagnoses cannot be proven wrong, supporting Rosenhan. Cooper et al. (1972) found based on clinical interviews, New York psychiatrists were twice as likely to diagnose schizophrenia than London psychiatrists.
This confirms a cultural bias exists, thus effecting generalisability from the context to another. Lindsay (1982) found schizophrenic patients were rated more abnormal than controls. This shows schizophrenic is a justifiable label, and in fact realistic. MacLeod (1998) argued if the effects of labels are powerful, why were the real patients in the study not convinced? Therefore, the pseudopatients must have overcome the effects of the labelling. Rosenhan, alongside other anti-psychiatrists, argues that mental illnesses are a social phenomenon that is simply a consequence of labelling. It is clear to distinguish between the sane and nsane is not possible.
Rosenhan concluded ‘any diagnostic process that lends itself so readily to massive errors cannot be a very reliable one’ and is ‘useless at best, harmful, and misleading’. However, it is clearly more dangerous to misdiagnose illness than health (false negative, type 1). The study established the flaws of diagnoses and the conditions within many psychiatric hospitals, thus influenced the improvement of many institutions. It has stimulated further research and has gained a long-term position in seminal research that remains to be used by contemporary professors today.