A more sensible test would be that of the grade of well-being of the family, which is also wanted for its own sake, as the final end of a budget investigation. But that, too, is difficult, if not impossible, to get statistically. It is not necessary to adduce elaborate proof of this point. Let any one who doubts the difficulty set to work to draw up a schedule to show degrees of physical well-being, and then consider how the information is to be secured and how accurate it is likely to be.

Or, to save labor, let him consult the Report of the New York Committee on Physical Welfare of School-children,* which did attempt to connect circumstances of life with well-being by the statistical method, and see what results are obtained.

That committee, reversing the process of the budget investigator, took a grade of well-being as its starting-point, and attempted to trace back to the circumstances of home life which might have caused it. Fourteen hundred school-children in the kindergarten and first two grammar grades, of different nationalities, from schools in all parts of the city, who were found by the school physicians to have defects of vision, hearing, breathing, teeth, and nourishment, were taken as the basis of the investigation, which consisted of inquiry into their home conditions "in order," the report states, "to ascertain whether their need arises from deficient income or from other causes."

At the very outset the test of "well-being" is seen to be a most uncertain one. While, at first glance, the idea that twothirds of the city's school-children are physically "defective" is a startling one, it depends, after all, upon the kind and degree of the defects how serious the condition is. The kinds of defects included in the class investigated by the committee are precisely those in which the degree of importance may vary most widely and is likely to be differently judged by different examiners.

The main items of defect shown in the fourteen hundred children studied were malnutrition, enlarged glands, eye defects, nose defects, throat defects. The largest single item was for "bad teeth," which was present in 74.9 per cent. of the

*Report on Physical Welfare of School-children. Pub. Am. Stat. Assn., June, 1907.

fourteen hundred children. This is obviously a defect as to the significance of which there may be great difference of opinion. Those who hold that it is of especial importance in its bearing on malnutrition and other defects will not get much support from the relation of defects shown in this report, as, in the 145 cases of malnutrition found among the 1,400 children (only 10.4 of all), only 73.1 were reported as having bad teeth; while the remainder not afflicted with malnutrition showed 75.1 with bad teeth. In fact, 291 of the children (21 per cent. of the 1,400) had no defects but bad teeth. Suppose we decided that "bad teeth," unaccompanied by any other defect, was not a very alarming condition, and subtracted the number showing that defect only from the defective class. Doing this and assuming that the 1,400 children with defects showed the ratio to children without defects given by the school authorities as 2 to 1, it is found that the proportion of "defective" children in the school population would drop to 53 per cent., which is quite a different story from 663 per cent.

In like manner the other items could be varied and the percentages changed. That there is a grave possibility of change in degree of defects, due to differences in examination, is more than hinted at in the report, where a comparison is made between the percentage of defects shown in the first examination of 1,400 children in October and in a re-examination of 990 of the same children in the following March and April. The remaining 410 had moved away,-a selection of sufficiently random character not to affect the proportions, which were as follows:

"Almost to a child," says the report, "conditions were found to be more serious in April than were recorded in the preceding October, November, and December." But "no one knows,"

the report continues, "whether these changes are due to actual deterioration or to the probability that a school physician, reexamining specially selected children, would be more thorough than when making original examinations. No allusion is here made to seasonal change, but this also would have to be taken into account."

Whatever the reason, it would seem that percentages which could vary so widely as from 45 to 70 for "enlarged glands," from 28 to 47 for nose defects, and from 30 to 45 for throat defects in six months, would be of little value either to show prevalence of defects or to make a basis for showing their causes.

As a matter of fact, the results of this research were mostly negative, as practically admitted by the committee themselves, due, however, not merely to the indeterminate character of the class taken for investigation, but to serious defects in method, which could, it seems, quite well have been avoided.

In the investigation of home conditions a large number of questions was asked, covering race, time of residence in New York State, in the city, income, occupation, members of the family working, kind and number of rooms occupied, rent paid, character of meals eaten by the child in question and his hours of sleep, disease and death record of the family, with record of circumstances accompanying the birth and infancy of the child, and many other matters.

The material thus obtained was embodied in thirty-eight tables of great length and detail, in comment upon nearly every one of which, one after another, the report itself states that the particular circumstance there tabulated did not seem to account for the defects.

This is not surprising when it is seen that not one of these tables gives comparisons with the same set of conditions for the school-child without defects, to see whether the circumstances shown are in a differing proportion for the defective child, and consequently may be presumed to have some relation to the defects.

In only four of the thirty-eight tables are the different defects tabulated separately. Of the remaining thirty-four,

twenty-two give classifications of circumstances of home life by race, nine by income, and three by both race and income for the class of defectives as a whole.

If this were a study of race traits and habits or of variations in manner of life characteristic of various income-groups, the tables would be to the purpose. But, after uniting heterogeneous defects in one arbitrary class and failing to give any standard of comparison with the normal child, in what possible way can the diseases and mortality of fathers and mothers, number of children born, mortality and diseases of children, housing conditions, income, etc., be shown to have relation to the defects given?

The nearest approach to a positive result is obtained from a couple of tables where malnutrition is tabulated separately and compared with the other defects, for several circumstances, income among them. Here we have some indication of circumstances that are presumably related to this one defect; but how much more light would have been thrown on even this one question, had the basis of comparison been, not children with other defects, but the average child, or, even more significant, the child without defects.

This lack could have been easily supplied for all the tables, and the value of the report increased tenfold, by making a parallel investigation of home conditions for seven hundred families in which the school-children showed no defects; that is, the remaining one-third, which is said to be the proportion of those without defects among New York City school-children.

Such an investigation, after all, is but a crude method of finding the causes of physical defects. Each defect may be the result of one of many coexisting circumstances or of any variety of combination of these circumstances; and each may in its turn interact with any of the others in a way that requires the utmost patience and nicety of physiological research to disentangle, taking up one defect at a time and one supposed cause at a time. Perhaps as long as this is the case it would be advisable for the lay investigator to leave such questions for the medical profession to solve.

Are the results gained in such inquiries undertaken under such difficulties as these worth the time and money spent on them? The most that budget investigations have determined, even if we assume them to be accurate, is that some people can live on a certain amount of money at a given time and place; but they seldom afford a guide as to what other people can or ought to do. The budget inquiry of the New York Conference was undertaken with the eminently practical purpose of throwing some light on the question of adequate relief, by showing how much money we ought to assume would be sufficient for a family to live on in a fairly normal way.

The conclusion drawn, however, is from an average based upon different race-groups, each of which presumably differs from the others in respect to the fundamental necessaries of normal living. For this purpose, indeed, the statistical method is especially weak, for it is a characteristic of the average not to fit individual cases.

The discrepancies shown in the different estimates that have been made of the minimum living wage indicate the difficulties of getting at the standard of living,by means of statistics. And this, quite leaving out of the question the variation in purchasing power of money between one period and another and one locality and another, which of itself introduces an element of uncertainty so great that as a guide to practice a budget investigation is out of date about as soon as completed, and inapplicable besides to other communities than that for which it was taken.

Are there, besides the lack of results, certain possibilities of actual harm in the present wholesale recourse to this form of inquiry? It always does harm to give out misinformation in the guise of information, and matter presented in the form of long columns of figures has in itself a convincing look. People in general will not criticise or even read statistical matter, but they are, notwithstanding the numerous jokes about statisticians and liars, tremendously impressed by it. This is what makes the irresistible tendency of the investigator to generalize on insufficient basis so dangerous.