"If there were some way, however, to measure accuracy without a 'gold standard,' mental health experts might be more credible," Mossman says. "Over the last two decades, statisticians have developed mathematical techniques that - in some cases - make it possible to estimate diagnostic accuracy without gold standards."

These techniques - which have been successfully used in areas as diverse as imaging liver cancer and detecting infections in dairy cattle - form the backbone of the study. Using statistical methods known as latent class modeling, the study looked at the performance of psychiatrists who made evaluations based on the 156 case files presented to them.

"The techniques are applicable to lots of questions in law and mental health," Mossman says. "There are many, many other kinds of cases where courts depend on mental health experts' opinions. If you have the right kind of data, these methods would allow you to evaluate the accuracy of court evaluations."

Mossman, himself an experienced psychiatric expert from dozens of court cases, says that by using this method to establish the accuracy of experts, the value of their opinions can be demonstrated and even assigned a mathematical quantity. But experts are still going to reach different conclusions.

"The legal system asks experts to give 'yes' or 'no' answers, but that's not how things usually are in medicine," he says. "Very often, a physician's diagnostic judgment really is a probability, an in-between answer. In courtroom testimony, experts are supposed to provide a clear opinion, not an ambiguous answer, even when the problem involves a shade-of-gray kind of question. That's where the real opportunity for difference of opinion comes into play."

Source: University of Cincinnati