Per Bech: Clinical Psychometrics (Collated):  

Jose de Leon’s comment on Martin Katz’s comment

 

          I was trained in the European psychiatric system, which includes performing the mental status examination according to German descriptive psychopathology from the Heidelberg school of psychiatry (Ban 2013). The textbook that initiated me into the rite of descriptive psychopathology (Scharfetter 1980) is completely unknown in the U.S. Many times, I wonder if my copy is the only one that reached the U.S., since a second-hand out-of-print copy costs $227 at Amazon. After my psychiatric training in Europe, I came to the U.S. in 1987. It was absolutely perplexing to learn that U.S. psychiatrists believed that interpreting proverbs was an important part of the mental status exam.  I was told that US psychiatric patients needed to interpret, “People who live in glass houses shouldn’t throw stones” in order to properly diagnose their mental illness.  I was very disappointed. I had never seen in any of the European textbooks I used in training a recommendation to use proverb interpretation and none of the US psychiatrists could articulate in a coherent way how a patient’s knowledge of this proverb would help me make a better psychiatric diagnosis.  Then I read a 1977 article by Nancy Andreasen, “Reliability and validity of proverb interpretation to assess mental status,” which demonstrated that my gut feelings were correct; interpreting proverbs did not help in any way to make a diagnosis. I admired Andreasen’s scales (SANS, SAPS and TLC) as the best elaboration of European descriptive psychopathology by US psychiatrists; in fact, seven of my articles in the 1990s focused on schizophrenia symptoms using her scales. Now I acknowledge that I should have disregarded my gut feelings and Dr. Andreasen’s article and listened to my first U.S. teacher. (He appeared not to be listening to my schizophrenia patients when I was trying to better assess their delusions and formal thought disorders. He got bored easily with them.  I thought he believed that, because the patient was psychotic anyway, who needed to pay attention to the details?  Only those Europeans.)  So now I know the proverb contains infinite wisdom, as I live in a glass house. I should not throw stones at Doctors Katz and Klein, who do not live in glass houses.

          Yes, the truth is that I have frequently performed factor analysis.  I recognize my sin.  If you go to PubMed and search for “de Leon J” and “factor analysis” you will see I have written 11 articles using factor analysis, many with Dr. Andreasen’s scales.  Yes, I am a psychiatrist, but I have “engineer genes” (see below). And, yes, I confess I have done many factor analyses, some by myself and many while supervising other psychiatrists or statisticians.  I have even “committed” the greater evil of being involved in confirmatory factor analysis with LISREL (most sinners do not pass beyond exploratory factor analyses).  I am happy to report that after going to Factor Analysis Anonymous I have stopped completely. Notice that PubMed shows that my last article containing factor analysis was published in 2003.  Since then my statistician and I are restricting ourselves to reputable statistical analysis, such as logistic regression or mixed models.  I have reached the mark of 10 years of sobriety without doing any factor analysis or even using this command in SPSS, although it is installed in my laptop.

          As almost no psychiatrist understands factor analysis, I am going to explain it for their tender young minds.  If you have a scale with 10 items, you can produce a correlation matrix of the correlations among all the items.  If items A, B, C are highly correlated you assume that they should be part of the same factor.  What are these factors?  They are mathematical artifacts that attempt to summarize the correlation matrix.  The problem is that you can choose factors that are independent or correlated (this is the secret of the rotation).  More important is the art of selecting the number of factors.  Is that a religion?  The literature describes a heated argument about what method or which prophet teaches you the best method to select the number of factors.  Is selecting the number of factors an art?  Some statisticians will tell you it is an art.  However, these same people will not acknowledge that art is not the strong suit of statisticians.  They tend to have “engineer minds” (if I am talking among psychiatrists I will say in a low tone, “They tend to have Asperger minds”).  This is only a comment on the religion and the art behind exploratory factor analysis.  I cannot provide an easy discussion for the tender young minds of psychiatrists on confirmatory factor analysis; that is only for grown-ups used to the evils of high-level mathematics. In my 1994 article with Victor Peralta and Manuel Cuesta I needed almost one published page to explain it.  

          As I am pretty sure that Dr. Katz has stopped reading my disorganized ramblings written purposely to put him to sleep, I can go ahead with summarizing the problem and throw the stones.  Factor analysis is an archaic and primitive method for summarizing scales.  Dr. Bech recommends item response theory (IRT) instead.  As Dr. Katz has no clue what IRT is and cannot comment on it, he takes exception to Dr. Bech’s criticism of factor analysis.  As someone who is no longer addicted to factor analysis, I can tell you that factor analysis is no good for your scales; please do not do that to them. You will need to believe me unless you want to try it yourself.  If you expect me to throw stones at IRT you need to wait a few months until I have had time to read Dr. Bech’s book and carefully digest it.  I have the same weakness as any psychiatrist except Dr. Bech in that I have no mathematical training.  But it appears that I am as abnormal as Dr. Bech because I understand statistics. Unfortunately, it requires a good deal of effort even though I have a mathematical brain (inherited from my father who was a professor of mathematics at an engineering school).  Since I am lazy, I have not studied enough mathematics. Did I inherit my lack of studying mathematics from my son?  As heredity from sons is not standard evolutionary theory, I am providing a reference (de Leon 2013) to support the fact that my son tends to be lazy but brilliant.  If that reference is published in a psychiatric journal, it has to be true. Thus, I know that my laziness comes from his genes, or is it the other way around: that I am hardworking and stupid?  You may need to ask my son.

 

References:

Andreasen N. Reliability and validity of proverb interpretation to assess mental status. Compr Psychiatry. 1977;18:465-72.

Ban TA. Neuropsychopharmacology and the Forgotten Language of Psychiatry Madness: From Psychiatry to Neurology. Madness: From Psychiatry to Neuronology. inhn.org.ebook. November 14, 2013.

de Leon J. Is psychiatry scientific?  A letter to a 21st century psychiatry resident. Psychiatry Investig. 2013;10(3):205-17.

Peralta V, Cuesta MJ, de Leon J. An empirical analysis of latent structures underlying schizophrenic symptoms: a four-syndrome model. Biol Psychiatry. 1994;36:726-36.

Scharfetter C. General Psychopathology: An Introduction, Cambridge University Press, 1980.

 

November 21, 2019