Basic ideas of scientific sampling

and also the use of dummy (usually integer valued) variables. Chapter 6 discusses the assumptions of regression, normality, constant variance, and independence of the errors, and how they can sometimes be dealt with. The close connection between linear regression and analysis of variance forms Chapter 7, which although true, since both are in the class of general linear models, need not be over emphasised. They really do, most of the time, play different roles in different situations. The final chapter is a resume of some of the more popular experimental designs, Latin Squares, randomised blocks, and factorials. Again the regression style of exposition reads a little uncomfortably. A pleasant readable book which I would say succeeds in its objective, but oh dear! that price, it will frighten off a significant section of the market.