Page:American Journal of Sociology Volume 2.djvu/763

 REVIEWS 747

used in destroying the power of a general impression and even to refute such a general impression by proving its opposite. While the manipu- lation of figures for political or rather partisan purposes has been a great abuse, yet statistics really performed the corrective service as one of its chief functions. Mr. Spahr comes preaching a new doctrine, and announces in his preface, that "he has come to believe that social sta- tistics are only trustworthy when they show to the world at large what common observation shows to those personally familiar with the con- ditions described." If it is true that "upon matters coming within its field, the common observation of the common people is more trust- worthy than statistical investigations of the most unprejudiced experts" (p. v.), it is at least pertinent to inquire what is the justification of this elaborate statistical presentation which the author makes. If it adds nothing to popular impressions, not even scientific accuracy, is it not reprehensible to give it the appearance of doing so ?

It is even possible that the author has indulged in some of the per- formances of the statistical prestidigitator, to whose activities statistics owe much of their disrepute. At least such an impression is possible when such statements as the following form a justification for general- izations covering the entire country. On page 65 the author says : (i) "The small holdings of real estate should be increased about one-half, because of the failure to record real estate in the rural counties. It is chiefly the small holdings of realty that fail to be recorded. (2) The small holdings of realty should be reduced about one-half, because the returns cover the gross possessions of the decedents." Why 50 per cent, should be added and subtracted instead of 20 per cent, or 100 per cent, is not suggested. Now while this particular readjustment does not affect the relative proportion in this particular table, it reveals the unscientific character of statistics drawn to accord with popular impressions. Nor does this criticism fail to recognize the fact that the author has made some special investigations into this field, though covering a limited area; hut that fact alone does not justify such sweeping assertions of such importance without elaborating the basis for the assumptions. In other words, there are too many occasions in the argument where the connecting link is an assumption (t. .;.. pp. 65, 68, 69, 97, 99, 100), often no doubt well founded ; but assumptions however well founded are not s

Some of the difficulties into which the author is led by such a method are revealed in the following instance. In his estimate of the