disease or degeneracy are important in connection with certain problems, such as those relating to social pathology; but the phenomena themselves are not yet thoroughly understood, and they are presented as subjects requiring further consideration and investigation. The distinction between evolution and growth, though a real one, is not always necessary, for in social fields the two processes seem often to represent differences of degree rather than of kind. And in any case the processes of evolution and of growth seem to be achieved in much the same manner. The term progress will continue to be used to include both processes.

Methods of Reasoning. Deduction and induction. Finally we may consider briefly some of the most profitable methods of procedure in the study of social phenomena. Scientific methods are usually classified as deductive or inductive. Deduction formulates particular laws from a general hypothesis by abstract reasoning, while induction uses concrete facts as a basis and makes its generalizations from them. Inasmuch as the deductive method has led scientists astray into the formulation of laws far removed from the actualities of life, it is at present in disrepute; and induction is supposed to be the only dependable scientific method. While this conclusion may be sound as a general proposition, the idea at present tends to be carried to an extreme; and hence the warning may be pertinent that deductive reasoning is not necessarily a useless scientific instrument. Pure induction itself may have its own dangers and defects, for it sometimes causes scientists to be content with the mere accumulation of masses of facts with little attempt at or success in their coördination or in the formulation of principles. In short both methods are capable of abuse. The danger of deduction would be its liability of producing a mass of nebulous theory suspended above the earth of facts; while the danger of induction would be its liability to accumulate a wilderness of facts leading to a desert of truths. The safest and at the same time the most fruitful method in science consists of a combination of induction and deduction, comprising three distinct parts. First, a theoretical but tentative hypothesis should lead next to an examination of the facts pertinent to the hypothesis, and lastly a permanent conclusion should be deduced from the facts. In many problems of the natural sciences the study of the facts is of overwhelming importance and the first and third parts of the process come to be almost negligible, so that the method may be said to be purely inductive. In the social sciences, however, where the forces are many and interrelated, the conclusions to be drawn from the facts are not so evident; and the selection and interpretation of the facts themselves is such an important part of the process that the method is manifestly deductive to an important degree. The truth is that, as the data of the sciences become more complicated, the theoretical side of the investigation has to be resorted to more and more. It is also true that a new science requires the deductive method more than an advanced one, for a new science is in the process of determining its boundaries and preparing its ground. Until tentative theories have been formulated the mere collection of facts is likely to be without system and without positive results. However, while all the social sciences will continue to use deduction to a greater extent than will the natural sciences, sociology is one of the concrete sciences; and, as it matures, it will necessarily place its chief reliance upon induction.

The method of difference. Induction as a system of procedure requires further explanation, because it is a name for a general process and includes a number of specific methods. The two which are of greatest interest to the social scientist are the method of difference, and the method of agreement, the most common form of which is that of concomitant variation.

One use of the method of difference consists in the study and comparison of two periods, one of which includes and one of which does not include a given factor. An example of this method may be found in the study of a society before and after a specific law has been put into operation. It is taken for granted that no other causal factor has been at work, and that therefore the changes which occur are the result of this one factor. In such a supposition lies the weakness of this method, because time is needed for the working out of any new force and time almost certainly brings other new factors into the situation.

.

Another application of the method of difference consists in comparing two similar societies or groups, one of which has a given factor at work while the other has not. This method requires the exercise of the greatest care in comparison, for no two societies are so alike as to vary in only one respect. In both cases deductive reasoning must be employed continually to show the probable relationship of causes and effects.

Sometimes both kinds of comparisons may be used to advantage on the same problem, the former furnishing positive and the latter negative testimony to the demonstration. In fact to be convincing the method of difference frequently requires a checking up by negative results taken from the study of a society where the particular factor in question is absent. If in this case the original results do not appear, then the probability that cause and effect have been found in the first case is very much strengthened. An example will make clear the complete process of induction by the method of difference. General observation leads to the supposition that drunkenness is an important cause of crime and therefore it is anticipated that a prohibition law will greatly reduce the amount of serious crime in a community. This supposition represents the tentative hypothesis which requires verification from a study of the facts. Turning next to the facts we may study the conditions existing within a state before and after the passage of a prohibition law. For example, in Colorado within two years after the state prohibition law went into effect the number of prisoners in the penitentiary was reduced by about one-third. Thus far the original hypothesis is supported by the facts. Careless investigators are usually satisfied to stop at this point with a single case of confirmatory evidence. To complete the demonstration however the negative test should be used by studying the situation in states having no prohibition law. The situation in California, which was then under the license system, failed to verify the preliminary conclusion, for during the same period the number of prisoners there decreased to almost as great an extent as was the case in Colorado. Thus doubt was thrown on the original supposition with regard to cause and effect, making necessary the testing of new hypotheses concerning the

primary cause of serious crime. Such an instance makes evident the desirability of negative as well as positive tests of the facts when inductive reasoning is employed.

The method of concomitant variations. The method of concomitant variations consists in comparing two variable social factors or conditions over a period of time. If the two factors rise and fall, or appear and disappear, together, the assumption is that they are in some way related. Either they are cause and effect, or both are the results of some common causal influence. This method is ordinarily used when large numbers of instances, as represented by statistical averages, are available; but the method may also be used with a few cases requiring no statistics.

The weakness of this system lies in the tendency in its use to assume a causal relation between the variants when no such relation actually exists. Sometimes also, even when cause and effect are evident, it is not clear which factor is cause and which is effect. While the method of concomitant variations often gives surprising and incontrovertible results, to be safely used it must be accompanied throughout by deductive reasoning reenforced by a broad knowledge of social forces.

An illustration of this method may be taken from the one given above with regard to the causes of crime. Another preliminary hypothesis could be made, – namely, that serious crime was the result of economic conditions as shown by the extent of unemployment. To test this possibility the investi- . gator would naturally shift his method from that of difference to that of agreement. If the statistics of unemployment varied constantly with the amount of crime in a variety of places, and at different times, the evidence would be very strong of a causal relation, because it would not be difficult to show reasons why unemployment might lead to crime.

Sources of Material. To be distinguished from methods of proof are two important methods of collecting and presenting material: the statistical and the historical.

1

See Durkheim's “Le Suicide" as a model example of

use of different kinds of concomitant variations. It furnishes also an example of the way in which the conclusions from this method of induction may be vitiated by an inadequate use of deductive reasoning.

Statistics. If the phenomenon to be studied is one which is continually recurrent in present day society the data may often be obtained in the form of statistics; and, if these are reliable and properly interpreted, they form an excellent basis for proof because they are so inclusive. The figures alone however may be very misleading, because they sometimes combine things dissimilar in nature; hence the more statistics are illumined by a wide personal knowledge of conditions the better.

History. If the phenomenon to be studied is one in which changes occur slowly, as in an institution, or one in which complete development occurs only occasionally, historical data are usually the only kind available. The historical method includes not merely the comparison of conditions in different historical periods in a given society, but also the comparison of conditions in societies far removed from one another, and often of those in different stages of development. Sometimes a phenomenon occurs so rarely that if all known cases are collected hardly enough evidence is found to afford proof of any theory; and such material is then presented merely as evidence of what has happened under the conditions given, rather than as a demonstration of what will inevitably take place under similar circumstances.

Either statistical or historical material may be used in the method of proof by means of concomitant variations, and occasionally they may be combined. For example, the marriage rate may be found to vary with prices; in which case we have concomitant variations of two sets of statistics. Or again, it may be found that the suicide rate varies with periods of war and peace; in which case we find sets of statistics varying with historical periods.

Observation. In addition to these formal sources of material, studies of social life, provided they are the results of trained observation, are of sufficient importance to deserve special mention. The value of individual observation will naturally vary with the nature of the problem studied. Its chief danger is that conclusions may be hastily made from cases too limited in number. This error is so common that in social phenomena covering large areas, or including large numbers, the method