The Pseudoscience of Denying Childhood Bipolar Illness
Science is about looking at all the evidence, not rejecting what we don't like.
Posted Feb 05, 2013
There is something odd about science, which makes it (contrary to common belief) the opposite of common sense: Science is centrally characterized by looking for evidence that opposes one's own viewpoint, not just what supports it.
In his blog posts, my colleague makes many scientific-sounding critiques of studies (retrospective methods, sample size, measures used) but these detailed critiques are usually made about studies which produce results opposed to his viewpoint. When he cannot find specific methods to criticize, he argues that the researchers themselves are not to be "believed", not realizing, it seems, that science is not about believing others, it's about disbelieving yourself.
For instance, there is an excellent study by Geller and colleagues which shows that about one-half of all 10 year old children who have just depression (or major depressive disorder) go on to demonstrate that they really had bipolar disorder (showing manic or hypomanic episodes). This is based on ten year prospective follow-up into adulthood, i.e., age 20. There are very few prospective studies that follow children into adulthood to see what their psychiatric diagnoses are, when seen in the light of adulthood. Such research is very difficult to do, and rare.
Yet my colleague rejects the results. He does so without any specific empirical problem in the methodology of the study provided. Rather, his main criticism is that the researchers had supported the idea that there is such a thing as childhood bipolar disorder. That in itself is enough reason to discount what those researchers found.
I don't like your results, so I won't believe your data.
Also the research is "old", meaning from the mid 1990s. it takes a few years to do any study, and in the case of the Geller study, ten years of prospective follow-up means that it takes at least a decade. If a study takes a decade or longer to conduct and then publish, it doesn't seem reasonable to view it as "old" merely a decade or so afterwards.
Another common critique, as shown in this post, is the view that we cannot believe studies on adults which involve retrospective analysis of childhood diagnoses, such as this huge National Comorbidity Survey (NCS) epidemiological study which found that about 10% of adult bipolar disorder began before age 10. Well, of course, there is a problem of recall bias in all research. But presumably my colleague also rejects the results of the NCS looking at the actual prevalence rates of bipolar disorder in children and adolescents, which found that about 0.5% of children and adolescents are diagnosable with bipolar disorder at the present time. This is not retrospective, but real-time diagnosis.
So, one rejects the adult research because it is retrospective, and one rejects the child research which is prospective because one doesn't trust the researchers. And one rejects the child research which is real-time because...
Science is not about finding reasons to disbelieve what we don't like.