One of the factors that can influence the validity of a diagnosis is clinical biases.  Bias is the errors or inaccuracies that stem from individuals’ perceptual and decision-making strategies.  Tversky and Kahneman argued that there were three major categories of cognitive biases that influence our decision-making: availability, representativeness, and anchoring.

Anchoring is when we make decisions based on limited information - using the first information that we have received to help us make a judgment. Anchoring can be based on numbers.  Studies like the one carried out by Englich and Mussweiler (2001) find that presenting suggested lengths of sentencing can actually influence a judge's decision.  Anchoring, however, does not have to be numerical.  If you have an "average" meal at a restaurant, your perception of it could change based on the starter.  If your starter is really amazing - then an average main course will be seen as "not great."  However, a "not great" starter could make an average main course taste better than average.

Friedlander and Stockman recognize that diagnosing someone with a mental illness takes time.  Often there are several sessions with a client before a diagnosis is given.  They wanted to see if the information provided in earlier sessions would serve as an anchor for diagnosis.

The researchers carried out an experiment using a sample of 46 US clinicians. The clinicians were a combination of psychiatrists, psychologists, and social workers from private practice, university, and community clinics.

The researchers used a repeated measures design.   The research question was whether giving significant evidence of mental illness in an early interview would have more influence on a final diagnosis and prognosis than giving that same evidence in a later interview. 

Clinicians were given two different case studies.  Joanne was a severe case of depression with a rather dramatic suicide attempt; Gina was less severe with some symptoms of anorexic behaviours.  Each participant read five consecutive interviews per case. Participants were asked to evaluate the level of functioning and the prognosis after each interview.  The score after the final interview is what was compared. The study was counter-balanced.  Some read Joanne first, some read Gina first. 

When reading the case studies, the "time" at which the important information was revealed was different for each condition. In the "early" condition, information about Joanne's suicide attempt or Gina's disordered eating was discussed in the first interview that was read.  In the "late" condition, this information was revealed in the fourth (the next to last) interview.

The findings were that in the less severe condition, the participants that read the evidence of mental illness in the first interview indicated a lower level of functionality and a poorer prognosis than the clinicians that received the information later.  It appears that the early exposure to the symptoms and the subsequent rating of the client influenced the final diagnosis.  However, in the group that had more mild symptoms in the first interview, the rating that they gave also served as an anchor which resulted in a less severe diagnosis and a better prognosis.

In the more severe case, although the initial diagnosis ratings were also different, they did not show a significant anchoring effect. In other words, regardless of when they read about the suicide attempt, there was no effect on their final diagnosis and prognosis.