Page:American Journal of Sociology Volume 2.djvu/395

Rh the causes at work in any case into two classes, the accidental or disturbing or varying causes, and the essential or primary or fundamental causes, and affirms that if a sufficiently large number of instances be observed and the results averaged, the first group of causes will be more or less completely eliminated, and the effects of the second group thereby made obvious. If, for example, a sufficiently large number of electric-light plants all agreeing in the one characteristic of being conducted by the municipality, and another series all agreeing in the one characteristic of being under private control, be carefully observed and the results averaged, it is assumed by this law that the numerous differences of the members of each series among themselves would cancel in the averages, and that the difference between the two averages would express the difference of effect resulting from the one constant difference, viz., that in the mode of management.

But it is, perhaps, erroneous to speak of the statistical method. There is rather a series of methods all possessing the common characteristic just described and therefore statistical, but differing in many of their details. A student familiar with one group of these methods, as I am somewhat familiar with the methods of population statistics, is not thereby constituted a good guide through the mazes of another group, e. g., financial statistics. I must confine myself, therefore, to certain general statements and not attempt to enter into details.

A fundamental and much neglected requisite of sound statistical work is that of careful definition. In this respect the European statisticians are in advance of us. For reasons which it would be tedious to state here definition is far more important in studying such phenomena than it is in the field of the natural sciences. It is the only way in which we can avoid the difficulties involved in the constant flux of social phenomena and be certain that we mean the same thing from moment to moment. For each form of municipal enterprises investigated a whole series of more or less familiar terms will be found, the exact meaning of which for that investigation should be fixed, and