The Statistics of Robert Kellner: A Glossary of Terms

THE STATISTICS OF ROBERT KELLNER: A GLOSSARY OF TERMS JAMES S. GOODWIN* Robert Kellner, M.D., Ph.D., died November 15, 1992, following an extended illness. He was seventy. A native Czechoslovakian, Kellner grew up in Prague under the German occupation. As a teenager, he was imprisoned twice due to his resistance to the occupation. He escaped to Yugoslavia and joined the Czechoslovakian battalion in North Africa. He was later transferred to the British Royal Air Force, where he served until the end of the war. He entered medical school in Czechoslovakia after the war, but fled to England in 1948, where he completed his medical education. Kellner's productive and distinguished career, first in the United Kingdom and later in the United States at the University of New Mexico, covered many aspects of research in the behavioral sciences. In 1988 he was awarded the first C. Charles Burlingame Award for psychiatric education and research. His students and colleagues will remember and cherish his gentle skepticism, typified by his glossary of statistical terms. It is difficult for scientists to keep up with advances in their field and still have time to educate themselves in other areas. The purpose of this communication is to revise basic knowledge; the present issue deals with definitions of commonly used terms in statistics and behavior sciences. It is intended as a revision for physicians, but it may also interest others who need to interpret research reports. Regression to the mean: In times of distress a person may regress to earlier , infantile forms of behavior. The term indicates a state of hostility and vindictiveness in times of frustration. Multiple regression: As above, except that several individuals display childish and obnoxious behaviors simultaneously, such as during the meeting of a committee. Standard error: The number of errors expected by members of a group in one year. For example, a surgeon who makes twice as many mistakes as colleagues would be assessed as having two standard errors.»Center on Aging, University of Texas Medical Branch, Galveston, Texas 77555.© 1994 by The University of Chicago. All rights reserved. 003 1-5982/94/3702-0847$01 .00 Perspectives in Biology and Medicine, 37, 2 · Winter 1994 285 Mean error: Originally indicated malice disguised as a mistake. (He said it was an accident, but it was rumored among the village folks that he committed a mean error, that it was arson—Edward J. Jones, The Blessings of Village Life, 1782, London, p. 218). In medicine the term has come to indicate an error which has serious consequences, such as a malpractice suit. Base rate: The number of unethical or disreputable acts in any one year. The base rate of two, three, etc., is calculated the same way as the standard error. Standard error ofthe mean: Spiteful or ruthless behavior toward individuals who appear to be unimportant. This sometimes has unfortunate consequences because some of the victims later gain power and become vindictive toward the perpetrator. Standard deviation: The number and kind of strange or disgusting behaviors which are found in a substantial proportion of randomly selected individuals; although undesirable, they tend to be tolerated by contemporary moral standards. Norm: An individual who has one or two standard deviations. The term is derived from a fictional, compliant, and credulous character (CP. Evans, The Adventures of Norm Atkins, New York: Simon and Schuster, 1937). Mean deviation: Undesirable acts, more disapproved of than the standard deviations; they are unusual, malicious, and vile. Analysis of variance: Psychoanalysis of individuals with mean deviations. Most analysts believe that it is essential to deal with the anal traits of these patients. This is also recognized by some laymen who refer to these patients by a vulgar and derogatory term which refers to their anal character. Analysis of covariance: A complicated system of psychoanalysis in which several anal traits are analyzed simultaneously. It should be attempted only by experienced analysts who have had personal analysis of covariance . There is no conclusive evidence as yet that this treatment is effective . F-ratio: Mainly used in sexology. The ratio of heterosexual acts of two or more consenting adults in private to all other sexual activities, including standard deviations. 286 James S...