Page:American Journal of Sociology Volume 4.djvu/137

 METHODOLOGY OF THE SOCIAL PROBLEM 121

question : What are the di£fereiit processes that combine to make up the life of men in society?

Causal analysis is distinguished particularly by its arbitrary isolation of specific elements from the whole complex mass of facts under investigation. This isolation is with the aim of discovering the causal relations of the ele- ments thus separately considered. Elementary analysis goes no farther in changing the object of contemplation than to separate its component parts successively from each other (thus doing m thought what is done literally when the different parts of a manikin are in turn removed and laid aside in the process of illustrative dissection).

To return to the case of our tree in the gale : descriptive analysis does not furnish an instantaneous photograph, nor even a kinetoscopic view, but something midway between the two, i. e., a naming of leaf, branch, trunk, with the conditions of each, as though for the moment the remainder did not exist.

Causal analysis goes beyond this in misrepresenting its object, i. e., it ignores altogether the existence of certain parts. It does not then confine itself to demonstration of the remaining parts. On the contrary, it makes every possible attempt to change the conditions of their coexistence or sequence. Thus the device of isolation is reinforced by arbitrary variation of the elements. (For example, in the familiar case of iron filings agitated upon the surface of a glass disk, and the variation of introducing a bar of magnetized steel beneath the disk, with resulting arrangement of the iron filings into the so-called magnetic curves.) This reinforcement of causal analysis by arbitrary variation of the constituent elements may take place whenever the nature of the object permits either the total removal of certain elements or an alteration in their proportions. The analytic form of experi- mentation consists of such arbitrary variation. Whenever experiment can be used, it deserves preference above every other kind of causal analyis. It is the most direct way of determining the causal relation of the parts of a phenomenon. In certain subjects experiment is out of the question (as, e. g., in the case of certain questions of human physiology, and of the most general cosmological, biological, and historical and social problems). In this case the closest approach to experiment, or the best alternative, is observation of the variations among the elements of phenomena of like general nature. (Thus we cannot produce murderers for scientific purposes, nor may we produce the conditions which incite to murder. We must resort, in the study of homicidal tendencies, to comparison of the conditions under which homicides have occurred.) In other words, we must observe variations that take place, without the assistance of the observer, instead of making variations. The more closely these collected variations resemble the con- structed variations, the more nearly will they approach experimentation in value as the source of causal conclusions.

It must be observed in passing that the discovery of appropriate facts,