do so satisfactorily in the case of pneumonia. For, while the outdoor class of occupations has a rate of 181.5, the Industrial class is no higher (181.2), the Professional class is but little higher (187.1), and the Mercantile class is much lower (167.0). The Laboring and Servant class, however, has an enormously higher rate, 313.2. Table VI will indicate that this class is numerous enough to counterbalance with their high rate the somewhat low rate of the Mercantile class and to give the city the preponderance noted before. Farmers have a higher deathrate from respiratory diseases than the average of the outdoor class, their rate being 198. Indeed, of the fifty occupations with rates given, only eighteen had higher rates, while thirtyone had lower rates. These eighteen that are higher, however, are enough higher to not only counterbalance these thirty-one, but to make the city death-rate to males from respiratory diseases 156.6 higher than the country rate and 137.8 higher for pneumonia alone (See Table IV, cols. 1 and 3).*

In view of the fact that women have a much lower death-rate from respiratory diseases than men in most of the occupations where they share the same conditions, and considering that the age composition of the sexes in these occupations has but comparatively little effect, it must be concluded that men are more prone than women to these diseases from lack of constitutional vigor. Whether this lack of vigor results from an

*This is not wholly true. The deaths from pneumonia, etc., among males not in occupations probably has some effect.

inborn tendency or from the greater stress of masculine activity in both a business and a social sense cannot be answered here. The fact that women have a higher death-rate from these diseases in the city than they have in the country leads to an inference that the oft-mentioned artificiality, hustle and dash, and nervous strain of city life do have their weakening effect. And it may be that the greater strain, the more general exposure, and the more frequent dissipation of men in cities turn down ever so slightly the flame of their bodily vigor, a conjecture strengthened by the city preponderances, small though they be, in nervous diseases, suicide, venereal disorders, and alcoholism, so that the small degree of vitality which so often means the difference between life and death is wanting when a virulent malady like pneumonia assails. Again, it may be that Willcox is correct in his inference, previously mentioned, that women have greater constitutional resistance to disease. By some it is urged that the larger head of the male child. weakens him at birth, and, though his rougher life in youth may bring him greater physical strength, it never restores the lost constitutional strength. The greater death-rate among male infants who share precisely the same conditions with female infants seems almost conclusive evidence that men begin life with a far greater susceptibility to bodily ailment than women.

To recapitulate the conclusions which have been suggested, it seems that a general deficiency of organic vigor in males (from whatever causes) results in a high male excess of mortality from pneumonia in cities; that, whether or not this deficiency in organic vigor lies at the basis of the much greater excess from consumption, city conditions attending indoor labor have a profound effect. These two diseases are of such prevalence as to bring about an excess of male over female mortality, generally higher in cities than in rural sections. This excess is roughly in proportion to the amount of city life, though it is affected by geographical environment.

THE USE AND MISUSE OF STATISTICS IN SOCIAL

WORK.*

BY KATE HOLLADAY CLAGHORN.

The general theory underlying the use of the statistical method in social work is so plausible and attractive that few to-day would venture seriously to attack it. That every social condition is due to some cause or causes, that social remedies depend for their effectiveness on knowledge of causes, that this knowledge involves more or less investigation and examination are almost self-evident propositions, stated in this general way, and lead naturally to the employment of the statistical method as a form of research especially applicable where large numbers are to be dealt with, as in the case of social groups.

And so it happens that every phase of the growing interest in social conditions and social betterment is manifesting itself in the endeavor to produce its own appropriate form of social statistics.

Every matter on which legislation affecting social conditions is desired (and these are growing in number and extent day by day) is first referred for investigation, as a matter of course, to some committee or commission which, after laboring from six months to two years, produces its own bulky quota of columns and tables to be stacked away upon library shelves.

The official censuses, State and Federal, originally intended to be mere enumerations of the population, include, year by year, a wider range of facts bearing on social conditions.

Our great organizations of various kinds, public and private, dealing with the social group in its various subdivisions, as dependents, delinquents, defectives, and so on, are keeping more and more complete and careful records, to show not merely

*Read before the National Conference of Charities and Correction, Richmond, 1908.

administration efficiency, but also such facts as are thought likely to be of value in throwing light on the special problem involved in their work or even upon those of a more general

nature.

And, finally, every settlement worker, every college boy and girl, filled with enthusiasm for "research" and "statistics," is roaming up and down the land, with open note-book and freshly sharpened pencil, to glean such precious spears of statistical fact as may have been passed over in the lumbering progress of the official machinery.

But, in this enthusiastic advance upon the shining fields of social research, how much have we looked behind, to note the value of the crop we have been gathering?

In some of the older lines of statistical inquiry the process of criticism has followed well after the progress of growth. In the methods of census investigations, in records of institutions, improvements are constantly being made that are constantly adding to the reliability and usefulness of the statistical results.

But outside of these boundaries, among the host of newcomers into the statistical field with a fresh idea for a statistical investigation for every day in the year, it seems as if the preoccupation with each successive new scheme prevents any critical examination of those already brought to completion. These inquirers irresistibly remind one of fresh-air children on an outing, so entranced by the new riches before them that they grasp handful after handful, flowers and weeds together, only to drop them unregarded at the enticement of the next waving tuft of brightness.

There is, in particular, one branch of social inquiry in which there is especial activity at the present time and in which it seems there has been especially little taking account of what is really accomplished, and that is the investigation of the circumstances of the lives of the poor in their homes.

A favorite variety of this sort of investigation is that which concerns itself with the standards of living of families, as shown by the items of family income and expenditure or the family

budget. The difficulties of this kind of inquiry seem to be fairly well recognized in theory, but with apparently little effect in restricting the output in practice.

Every one to whom it would occur to make an attempt at statistical investigation at all is supposed to understand, as the very first principles of his work, that, in order to secure anything that may properly be termed "statistics," the matter dealt with must be capable of expression in quantitative terms, must consist of units that can be distinguished as like and unlike on some objective and verifiable basis, and that can be added, subtracted, and otherwise compared numerically. In the second place, information as to these quantities and relations must be accessible. In the third place, a sufficient amount of information must be gained to eliminate individual variations in the subject studied and errors of observation on the part of the investigator.

In social inquiry, generally, as compared with most other lines of research in which the statistical method is used, there is especial difficulty in meeting not merely one or two of these conditions, which might be counterbalanced by especially thorough compliance with the remaining one, but in meeting any and all of the three, owing to the number, obscurity, and complexity of the elements involved.

And this is especially the case in investigations of standard of living. Any one who has been behind the scenes of one of these investigations, and knows how the information is obtained, must indeed feel a sense of wonderment at its imposing array of columns, percentages, and averages which look so positive and convincing on the printed page.

In the first place, merely in arranging the concrete elements of the problem, the investigator has no end of difficulty in deciding upon the "units" that are to be counted. They must be distinct, they must always have the same meaning in the same investigation, and they must be significant. It is necessary to reduce all sorts and kinds of commodities to some common basis of kind and quality and price. The concrete articles of food must be divided into general classes and dis