Premium
RELEVANCE OF REGRESSION TO THE MEAN TO EPIDEMIOLOGICAL AND CLINCAL STUDIES: A REVIEW OF THE LITERATURE
Author(s) -
ALTAN S.
Publication year - 1987
Publication title -
journal of sensory studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 53
eISSN - 1745-459X
pISSN - 0887-8250
DOI - 10.1111/j.1745-459x.1987.tb00183.x
Subject(s) - regression toward the mean , regression , regression analysis , statistics , value (mathematics) , econometrics , population , linear regression , psychology , baseline (sea) , relevance (law) , test (biology) , demography , mathematics , biology , political science , sociology , paleontology , law , fishery
Investigation of the relative heights of pairs of fathers and sons led the famous 19th century scientist, Sir Francis Galton, to the concept known as “regression to the mean.” Gallon observed that sons of extremely tall or extremely short fathers tended to be less extremely tall or short. The concept has since been ‐ found to apply to virtually any experiment requiring a test‐retest design. On the average, the bottom group on the first test will tend to increase, and the top group will on average decrease on the retest. For example, a clinical study requiring an initially high blood pressure baseline value (screening value), might decrease as a group on a subsequent retest, perhaps merely due to a regression effect. One way to account for possible bias due to regression effects is by using an appropriate control group. In the absence of a control group, James (1973) has developed some useful formulas for estimating the expected regression effect, when a truncated population is being sampled. Choice of an appropriate baseline value should also be guided by consideration of the potential effects of regression.