This book is designed to serve as the text for a one-semester beginning course in economic or business statistics. The number of similar books currently available is large, and, since all are intended for the same purpose, the amount of material in common is necessarily substantial. This book differs from the others, then, not so much in content as in its basic organization and method of exposition. In this it is sufficiently novel to merit a few words of explanation.
The central problem of quantitative analysis in any social science arises from the complex and multivariate character of individual human and institutional behavior. A realistic approach to the subject requires that the student be introduced to this problem in its most general and complicated form as soon as possible. While many specialized multivariate techniques are far beyond the scope of a single semester, the most general and inclusive method is fortunately also the easiest. The cross- tabulation of cell means and their standard errors is easily understood, applied, and interpreted by the student as soon as he understands the nature of the arithmetic mean and its sampling distribution.
Not only is this method the easiest, it is also--given sufficient data-- the most sophisticated. Indeed, the student immediately meets such concepts as multiple correlation, interaction, and multivariate relationships combining numerical and categorical variables. Moreover he learns to know nonlinearity, heteroskedasticity, estimating bias, unequal cell sizes, and so on, as facts of research life rather than as vague threats lurking somewhere behind simplifying assumptions.
This book is tightly organized around the multivariate function