Finding
As we noted earlier (Packard and Boardman 1987, 1988), even statisticians disagree on what is the "best" al-
Paper
Reply to Technical Comments
Citations: 7
Full textAbstract
The primary objective of our earlier papers (Packard and Boardman 1987, 1988) was to demonstrate that dividing values for a physiological variable by corresponding measures for body size is not a generally useful way to "correct" the data for variation in body size within and among experimental groups. Such ratios do not remove effects of body size from the variable of interest, and the ratios often are very misleading. The use of ratios to scale data can lead to the appearance of an effect when none is present, or it can mask an effect when one actually exists (also Tanner 1949). A second objective of our work was to demonstrate the importance of plotting data and examining them visually. A study of the bivariate plots presented in our papers was essential for the proper interpretation of data provided in the several examples. We realize that many biologists feel uncomfortable with the idea of relying, in part, on subjective impressions gained from studying such plots, but such subjectivity has an important place in the interpretation of data (Warren 1986). We judge that Tracy and Sugar (1989) and Magnusson (1989) agree with these major points, but that they disagree to differing degrees with our advocacy of the analysis of covariance (ANCOVA) as an alternative to the use of ratios in analyzing physiological data. As we noted earlier (Packard and Boardman 1987, 1988), even statisticians disagree on what is the "best" al-
Authors
G. C. Packard, T. Boardman
Journal
Physiological Zoology