Cognition & Reality

Wednesday, 4 August 2010

When Assessing Twin Concordance, Be Stupid

Filed under: Behavioral Genetics — drtone @ 7:51 am

A person wishing to understand what a truly idiotic research paper looks like need search no further than an article entitled “When Assessing Twin Concordance, Use the Probandwise Not the Pairwise Rate” (McGue, 1992). In accordance with a general tendency in biological psychiatry, it pretends to be logical and scientific, but is actually nonsense through and through.

First, it justifies using the so-called “probandwise” rate of concordance in twin studies because it is a better indicator of “risk at the individual level” rather than of the population rate of concordance. That would be OK if the question asked in twin studies of mental illness were about individual risk, but the question these studies ask is explicitly about the rate of concordance in the population. This mistake is so basic and so obvious, and is made so explicitly, that it is difficult to understand.

McGue says, “…geneticists are primarily interested in forecasting risk at the level of the individual rather than at the level of the pair. For example, in a genetic counseling situation, what is needed is the risk that the individual will become affected given knowledgethat his/her relative is affected.”

In other words, McGue claims that the famous twin studies of mental illness are designed to address a clinical question. That is completely and utterly incorrect. The prior question, the one the public expects reported figures to reflect, is a scientific question regarding the likelihood that one twin is, say, schizophrenic given that his or her other twin is schizophrenic. This is not a question about individuals, as McGue insists, but about the population. Confusing the two can only be ascribed to out and out stupidity or out and out dishonesty.

McGue follows up his first error with a second of perhaps even greater magnitude. It can be shown arithmetically that the probandwise rate inflates the concordance between twins, systematically yielding exaggerated rates of concordance, depending on the the “ascertainment rate,” the likelihood that an individual will be diagnosed as ill. The lower the ascertainment rate, the larger the error. The reason for this is that a pair of twins in which both members are affected is more likely to be included in any study. That this is so is well-known and not a matter of controversy: The “pairwise” rate only counts a concordant pair of twins once, but the probandwise rate counts such pairs twice.

The pairwise method  yields different estimates of concordance depending on the ascertainment rate, but the probandwise method yields the same concordance regardless of the ascertainment rate, as McGue shows. He concludes, based on this, that the probandwise method is superior because it is insensitive to the ascertainment rate. This is not unlike saying that the feature of my car that I like best is that the speedometer does not work, because I know I drive too fast.

These huge errors have compounded over the nearly two decades since McGue published his paper, because his “logic” is often cited to justify the use of the probandwise rate to obtain estimates of the genetic factor  in many areas of medicine.


Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at

%d bloggers like this: