Psychological Factors in Marital Happiness: By Lewis M. Terman, assisted by Paul Buttenwieser, Leonard W. Ferguson, Winifred Bent Johnson and Donald P. Wilson. New York and London: McGraw-Hill Book Company, Inc., 1938. 474 pp.

This volume is the outcome of several years of research aiming to define statistically the relationship of some four hundred variables to the marital happiness scores of 792 couples. The exclusive source of data for the study was provided by written answers to a lengthy

questionnaire embracing three main groups of variables: personality factors, background factors, and specific sexual adjustments. The total happiness score of each individual was computed as a numerical index representing a composite of numerous items derived from responses which received graded weights on the basis of intercorrelations, an individual's subjective estimate being only one item correspondingly graded.

In spite of the extreme methodological caution with which the authors seek to close in on their problem, the significance of their results is frequently undermined by the soft paternalism with which they handle the difficulties inherent in the questionnaire method, a method—at least in its present crude state—combining all the liabilities of the subjective approach with few of the advantages. The main trouble seems to arise from an insufficient preoccupation with what is perhaps the most crucial aspect of a venture such as this: the riddle of when is a question not a question or exactly what does each question presume to measure?

This neglect is fortunately not equally crippling to all sections of the work since of the several hundred questions to which the respondents handed in written answers, there were many where the answers could be accepted as adequate indices of definitely measurable quantities, and where the answers could be counted upon to be reliably straightforward, free from entanglements and capable of being lumped together for purposes of statistical comparison. Thus, questions having to do with age, occupation, income, number of siblings and even items like frequency of intercourse, for example, would fall into this category. Assuming a high degree of honest accuracy