Acton F.S. Analysis of Straight-Line Data

Reprint edition (originally published 1959). — New York: John Wiley and Sons; London: Chapman and Hall, 1984. — 275 p.The straight line occupies a central position in the presentation and interpretation of experimental results. Frequently experimenters devote considerable time and ingenuity to the dcvelopment of reasonable (and sometimes unreasonable) procedures for linearizing their data. It is the objective of this hook to guide the experimenter-analyst through the maze of regression techniques, each with its om set of assumptions, that has accumulated over the past eighty years.
The author gets off to an excellent start with a hrief discussion of the choice of a statistied model on which to base the analivsis. Most experimenters should pale considerably when, after a listing of seven hypothetical experiments, the author states "...it u-auld he sheer folly to apply cleaaical least-squares fitting to most of these data." Thia phrase should frighten most people into reading the remainder of the hook.
Well over a third of this hook deals with the so-called classical model in which the independent variable (z) is considered known (essentially) without error and (almost) all of the experimental variation is assumed tobe associated with the dependent, variable (y). Considerable emphasisis
placed on confidence intervals, and several non-pitramctrie methods are discussed and exemplified/