From Wright's path analysis, a theory of hypothesis testing for linear models of familial correlation is developed and applied to several bodies of data, for some of which the authors' conclusions are not supported. Much larger samples and better designs are required to discriminate hypotheses than to estimate parameters under a particular model. Possible extensions of the method are discussed.
|Number of pages||29|
|Journal||American journal of human genetics|
|State||Published - Dec 1 1974|