A diagnosis is considered valid when it is based on appropriate data and the diagnosis addresses the needs of the client, leading to improvement. An invalid diagnosis is made when other factors overly influence the psychiatrist, including his/her own biases. In order for a diagnosis to be valid, it must also be made objectively.

A diagnosis is reliable when two or more clinicians come to the same decisions regarding diagnosis independently of one another. Diagnosis is often unreliable because two doctors often give different diagnoses to the same patient.

Self-reported data is really the only way that we can determine the symptoms of a client. To date, there are no simple blood or urine tests that can be carried out to determine what may be affecting a patient. Self-reported data as several problems relevant to diagnosis. Patients are often unable to remember when the symptoms started. They also may not be able to really explain the severity of their symptoms. The "peak-end rule" predicts that patients will best remember the most significant symptoms or moments (the peak) and the most recent events (end). Patients are also open to memory distortion since they are recalling past behaviour and symptoms. Self-reported data also assumes that the patient is aware of his or her own behaviour. Finally, even though the patient may be there to seek out help, s/he may also succumb to social desirable effect, not wanting to disclose information that they may feel makes them look bad.

Personal biases regarding gender may lead to the overpathologization or underpathologization of an individual. A woman may be diagnosed as clinically depressed whereas a man may be diagnosed simply as stressed. Cultural biases may also affect one's diagnosis - as seen in the study by Li Repac.

Reactivity is when one changes his or her behaviour because they are being observed. This makes it difficult to diagnose because a clinician has to consider that anxious behaviour - both behaviours like fidgeting or stumbling over one's words and increased heart rate - may be the result of the nervous situation created by coming to a doctor and asking for help. This means that diagnosis, ideally, should take place in an environment where patients feel relaxed and unthreatened.

There are several questions that one should ask about this study. First, what were the actual tests that were carried out on the patients? To what extent did the researchers actually interact with the patient prior to making a diagnosis? Was there consent from the hospital to carry out this research? And most significantly, how might treatment over time have changed the behaviours of the patients and thus affect any future diagnosis? It is also interesting that a diagnosis was given for all patients, perhaps because of the assumption that if they are in a hospital, they must be ill.

There are clearly ethical considerations about the way that the data was collected - that is, covertly without the consent of the participants. In addition, there were assumptions made by the team with regard to what the nurses were writing. The researchers did not confirm their data with the nurses themselves and thus the conclusions lack credibility. Finally, there is significant bias in the collection of data. The "patients" were taking data on how they were being treated. This is not objective and also not reliable. There is no evidence to confirm how they were actually treated by the staff.

In the original study, Rosenhan was looking at the validity of diagnosis - in other words, does the diagnosis given actually address the symptomology of the patient and begin a process of effective treatment? When Rosenhan later did a study to see if doctors would be able to identify patients who were diagnosed schizophrenic as being schizophrenic, this was a study of the reliability of diagnosis.

Triangulation in diagnosis is the whole basis of the case study approach. Some of the ways that triangulation of data could take place are:

• Carrying out blood testing to determine toxins or drugs that may be influencing behaviour.
• Carrying out MRI or fMRI scanning to detect brain abnormalities
• Observation of the individual's behaviour while carrying out a routine task or interacting with family members.
• Administering psychometric testing - for example, IQ tests, depression rating scales
• Carrying out interviews with close family