There are several reasons why a diagnosis may be invalid.  One reason is when psychiatrists allow clinical biases to negatively influence their judgment. Clinical biases are cognitive strategies that are used to make a decision but limit the objectivity of the psychiatrist.  Clinical biases may be the result of basing decisions on previous experience with patients, time pressure, overconfidence or suggestions from colleagues or the patients themselves. Three examples of clinical biases are confirmation bias, anchoring bias, and illusory correlation.

The Dual Processing Model argues that there are two key ways that we make decisions. System 1 thinking is based on past experience.  It works with what it sees as evidence and makes decisions based on patterns.  It is quick and requires less effort.  System 2 thinking is based on conscious reasoning.  It is slow and more effortful.  Although we would like to think that psychiatrists would always practice system 2 thinking, time pressure and overwork may lead to the use of system 1 thinking.

One clinical bias is confirmation bias. Confirmation bias is the tendency to interpret information in a way that affirms one's beliefs or hypotheses.  This is when we “see what we want to see.”  Temerlin carried out a study with clinical psychologists.  They were asked to watch a video of an interview.  The person being interviewed did not have a mental health problem.  One group was told by a respected psychologist that although this person looked fine, he was actually psychotic. The other group served as a control.  As a result of hearing this comment before watching the video, 60% of the participants diagnosed the man as psychotic; in the control group, no one did. Having heard this information before watching the video, the psychologists lost objectivity and paid attention to behaviours that confirmed the diagnosis of psychosis.

The study is a highly controlled experiment, so cause and effect can be established. However, the situation is highly artificial.  Clinical psychologists do not make diagnoses by watching a single video.  In addition, the experiment used deception, which is ethically problematic.  The respected psychologist shared a diagnosis that was not true. In spite of this, the experiment is important.  Since we don’t like to employ System 2 thinking because it requires effort, the participants relied on the “expert’s opinion.” Once that initial diagnosis is made, it shows that it is difficult for psychologists to be objective and open to a different diagnosis.

Another clinical bias is anchoring bias. This is when we rely too much on an initial piece of information when making decisions. Friedlander and Stockman tested the role of anchoring bias in diagnosis with a group of US clinicians. They wanted to find out if it made a difference if a patient described significant symptoms of mental illness in an initial interview had a greater effect on diagnosis than when that information was shared in a later interview. Participants were given a case study of a patient with anorexia which was made up of five interviews. In one group, the first interview contained information about her most serious symptoms.  In the other group, this information was revealed in the fourth interview. The researchers found that when the information was revealed in the first interview, the diagnosis was more severe and the prognosis poorer than when it was revealed in the later interview. It appears that the first interview served as an anchor for making decisions about the patient.  Once they had heard this information, they judged further interviews with this perspective in mind.  Anchoring bias is often used when we feel that we do not have enough information to make a decision. We then use the initial information we receive to help make a decision.

This study was an experiment, so a cause and effect relationship can be inferred. The sample was made up of experienced professionals, so the research can be generalized. However, this should be done cautiously since the sample was small and was limited to US clinicians – which means the study may have a cultural bias. In addition, the situation is to some extent artificial.  Although psychologists rely on interviews for information, they would carry out the interview face to face, rather than reading someone else’s interview.  They would also be able to ask clarifying questions and make note of non-verbal behavior.  Finally, the case study was focused on only one disorder – anorexia. It could be that a different disorder may have led to different results.

A final clinical bias is illusory correlation – this is when stereotyping can influence a doctor’s judgment.  A good example is the belief that women are more emotional and therefore suffer more from anxiety and depression than men do. Swami carried out a study of how this belief would influence diagnosis. They used a sample of British adults who were given a summary of an interview.  The description met the DSM criteria for depression. The participants were told either that the interviewee was a man or a woman.  When asked if to discuss the interviewee’s mental health, participants were less like to label the male as having mental health problems than the female.

This study could help to explain differences in prevalence rates of disorders.  It could be that illusory correlation with regard to gender, ethnicity or socioeconomic status would have a significant fact on diagnosis.  However, the study was done on people that were not clinicians. This makes it difficult to generalize to professionals in mental health.  

Much of the research on clinical biases does not reflect the actual conditions in which diagnoses are made.  However, they do show that professionals, just like average people, often rely on system 1 thinking when they feel the need to use less effort or they are lacking enough information to make a decision. It is important that clinicians are aware of their biases with the goal of increasing the validity of the diagnosis of their patients.