Popular Science Monthly/Volume 40/November 1891/Lessons from the Census II





TO my own mind, the Federal census system is faulty in many features. It is bungling, unwieldy, and unproductive of scientific results. It is the legitimate growth of time and the honest endeavor to secure broader and broader results to satisfy the growing demand for information concerning all the conditions of the people, and it is perfectly natural that the additions from time to time should have resulted in the present system. The system should be changed radically before another census period comes around.

To be specific in the condemnation of our system, attention should be paid, first, to the method of enumeration. Vicious as it is, it is a vast improvement upon that existing prior to 1880. There are four methods of enumeration, or rather four methods of enumeration have been tried on pretty extensive scales. The English method consists in securing all the facts called for under the law in one day. For this purpose a vast army of enumerators is appointed from the central office.[1] The organization under the British Census Act is under the control of the Local Government Board, and the immediate chief is the Registrar-General. Local registrars of births and deaths must divide their subdistricts into enumerators' divisions, in accordance with instructions from the Registrar-General, and subject to his final supervision and approval. Every registrar of births and deaths must furnish to his superintendent registrar lists containing names, occupations, and places of abode of a sufficient number of persons qualified, according to instructions, to act as enumerators within a subdistrict, and such persons, if approved by the superintendent registrar, shall be appointed enumerators for taking the census. The board causes to be prepared a table of allowances to be made to the several enumerators, registrars, superintendent registrars, and other persons employed in taking the census; and such table, when approved by the Treasury, is laid before both Houses of Parliament for their action. Under the act the schedule comprehends eleven inquiries, relating to the members of the family, visitors, boarders, and servants who slept or abode in the dwelling on the night of Sunday, April 5, 1891, and the schedule was called for on Monday, April 6th, by the appointed enumerator, whose business it was to see that the schedule was properly filled by the head of the household, and, if not, to cause it to be so filled. This method seems to be the one that attracts the attention of statisticians as the ideal method. Under it, however, much complaint exists in Great Britain, not only as to the processes of carrying out the law, but relative to the inaccuracies in the returns; and I have been informed that much difficulty is experienced in obtaining well-filled schedules. It is unreasonable to suppose that in a population varying widely in the intelligence of its individual members a schedule can be properly filled or so well filled as to secure a reasonably scientific result. The English census has been extolled for its accuracy. I do not believe it is any more accurate than any other census taken by other methods. I have before me a discarded schedule—that is, an improperly filled one—left with an intelligent mechanic, well educated, of wide experience, a machinist by trade, and perfectly competent to write an article for a magazine; and yet he could not, or did not, properly fill the schedule left with him, and on an examination of it it is not strange that he did not. When the difficulties of filling the simple English schedule are considered, it becomes preposterous to suppose that the expanded schedule under the Federal system could be filled under the English method. This has been tried, and in a State where the population has been taught to consider the value of statistics—the Commonwealth of Massachusetts. In 1875 the English method was adopted; the schedules, comprehending all the inquiries at that time called for by-law, were left with the heads of families, with clearly defined instructions, sample sheets, etc., all in accordance with the recognized English method; and from that community, which, it is reasonable to suppose, could fill the census schedules if any community could do it, but thirty-seven per cent of the returns were in a condition for use. The balance had to be corrected or made entirely by the enumerators. That method was therefore abandoned in subsequent censuses for the State of Massachusetts. With the sparsely settled population of the United States, and with the broad schedule of the Federal census, covering as it does twenty-four inquiries, it would be absurd to attempt to take the census under the English system.

In Germany the labor of enumeration is performed by persons who, in consideration of the public utility of the work, do it without compensation.[2] It has been thought that this feature could be embodied in the United States census to a certain extent, or at least supplemented by the employment of school-teachers in the enumeration. The German method involves, of course, the creation of exceedingly small enumeration districts, after the English method, a block in a city or a portion of a street in a town or village being allotted to some patriotic citizen who would without compensation see to it that the schedules were properly filled. It is doubtful if this method could be made useful in the United States. Our people are too busy—at least those competent to take charge of such work—to induce them to enlist. The great difficulty even now is to secure men for a week or a month's service under the Census Office.

The third method of enumeration is that practiced in the State of Massachusetts, and certainly the scientific results of the censuses of that State would indicate the value of the method employed. Since 1845 the Commonwealth of Massachusetts has taken a census regularly, on the mean year of the Federal censuses. It started its census work in 1837 by an account of its manufactures, etc.; but its first enumeration on any broad scale was in 1845, through the assessors of cities and towns. In 1875 the field work was done by enumerators appointed by the census authorities and paid by the day, and they were instructed to secure full and complete results without regard to the time taken. For the population the English method was used, as already stated. The manufactures and agricultural products were secured on individual schedules, statements being certified to by proprietors. In 1885 the card schedule for population was successfully introduced, the other features of the 1875 system and per diem compensation being retained.

Under the Federal system, which I have said is so faulty, all data are collected, so far as population, agriculture, and the general statistics of manufacture are concerned, by enumerators selected by the supervisors and appointed by the Superintendent. The supervisors under the eleventh census are fairly compensated; the enumerators are not. The compensation for enumerating the population under the existing law is in most of the country two cents for each living inhabitant, two cents for each death reported, fifteen cents for each farm, twenty cents for each establishment of productive industry enumerated and returned, and five cents for each surviving soldier, sailor, or marine, or each widow of a soldier, sailor, or marine returned. In some subdivisions the allowance for each living inhabitant may be increased, but the compensation allowed to any enumerator in any difficult district shall not be less than three dollars nor more than six dollars per day of ten hours' actual field work, when a per diem compensation shall be established by the Secretary of the Interior instead of a per capita; nor, where the per capita rate is increased, shall it exceed three cents for each living inhabitant, twenty cents for each farm, and thirty cents for each establishment of productive industry; nor shall claims for mileage or traveling expenses be allowed any enumerator in either class of cases, except where difficulties are extreme, and then only when authority has been previously granted by the Superintendent of the Census. The allowance relative to inhabitants and deaths is the same as under the tenth census. There is an increase of a few cents in the compensation for enumerating farms and establishments or productive industry. It may not be possible nor wise to change this method, but it is possible and wise to make the compensation fair and just. Under these rates it is almost impossible for an enumerator to earn a fair day's wage if he does his duty. In localities where the population is dense, he can earn three or four dollars per day. His ambition is—and human nature prompts it—to secure as many names as possible, and in too many instances he will do this at the expense of accuracy; for accuracy consumes time. Furthermore, he may be inclined, in the very worst localities, in the slums of great cities, to omit, for personal reasons of convenience or otherwise, to enumerate all the people, being contented with taking the population in sight; in other words, two cents a name might not induce him to enter all the dens of the slums of a great city for the sake of accuracy. In sparsely settled localities even three cents a name (the per capita rate, it must be borne in mind, covers all the multitude of facts called for on the population schedule) will not enable an enumerator to earn a living for the time employed, and he is often inclined to take the statements of neighbors rather than to travel a mile or two to secure accurate statements relative to half a dozen persons. In enumerating establishments of productive industry, the compensation allowed by law will not enable an enumerator, either honestly or dishonestly inclined, to secure any very valuable results. It is quite impossible to fill out a manufactures schedule completely and with fair accuracy for twenty cents. A man could not earn one dollar a day if he did his duty, and on the enumeration of farms he could not earn seventy-five cents a day. The complete agricultural statistics under the census of Massachusetts in 1885 cost about one dollar per farm, instead of fifteen or twenty cents.

The difficulty which Congress would have to meet in adjusting this matter of compensation is twofold. If a very large body of enumerators, like that employed under the elventh census, nearly fifty thousand, should be enlisted on a per diem compensation, the fear would be that there would be men enough in that vast army who would delay their work for the purpose of increasing their earnings to swell the cost of enumeration to enormous proportions although reasonable accuracy would thereby be secured in every direction. On the per capita basis the question would be whether accuracy should be sacrificed for the sake of a lower cost. The evils of the present system are so great, however, so far as compensation is concerned, and the results of the census vitiated to so large a degree, that it would seem to be wise to adopt a system of compensation which should secure fair accuracy in the results even at an increase in the expense The country grows so rapidly, and the wealth and business increase so largely, that the total expense of a census should not be considered when the accuracy of the same is at stake.

Another fault of the present system, to my mind, lies in the organization of the field forces. It is perfectly natural that the Census Office, and that Congress, even, should seek a speedy enumeration of the people; but it is submitted that if an instantaneous enumeration can not be had—and it is clearly demonstrable that it can not in this country—then whether it take a week or two weeks, or even three or four, to complete the enumeration becomes a matter of lesser consideration. It might, therefore, be wise to make larger districts and use a less number of enumerators rather than to extend the method by decreasing the size of the districts and increasing the number of enumerators, as is the present tendency. An enumerator, working for a few days, acquires speed and accuracy as a matter of experience, and his second week's work is of vastly greater value than his first few days' service. It might be well, therefore, to so subdivide the country into enumeration districts that each enumerator would have at least four or five thousand people to enumerate, instead of an average of two thousand, as under the present method. If the districts were enlarged, the number of supervisors should be greatly increased. The present law provides for one hundred and seventy-five supervisors; that of 1880 provided for one hundred and fifty. It would seem to be a prudent measure to provide for at least one thousand supervisors, which body, with a reduced number of enumerators, could take greater pains with all parts of the enumeration; and if supervisors could be selected with special reference to their fitness and enumerators could be tested by the use of a preliminary schedule relating to their own families and perhaps one or two neighboring families, results would be secured which would defy criticism. With such changes there should come a change of date for the enumeration. The count of the people is now made as of the 1st of June—under the present law, the first Monday in June. The changes in the habits of the people necessitate a change of date. More and more every year people leave the town for the country, and this change occurs about the time of the enumeration. The date should be changed to a period of the year when the population is more thoroughly fixed or more thoroughly housed in permanent homes. Could the date be carried forward to the autumn, a great gain would be made in the accuracy of the enumeration—not perhaps in the total for the whole country, but in the total for each State and city. Certainly the results would be far more satisfactory to all concerned, even though the change in the total population of the United States did not exceed a few thousand. Each State wants its own: political and social reasons demand that this should be so.

Perhaps the very worst form of the present system is the temporary nature of the service. As the census year comes in sight each decade, a Census Office is created by law, the organization to be taken entirely from new material, from the head to the foot. Of course, the aim always is in securing a superintendent to select some one who has had more or less experience or is supposed to be more or less competent in census work; but then comes the greater difficulty, the selection of the forces. A good business man at the head of the Census Office—one of excellent administrative and executive abilities, without knowledge of statistics—would handle a census, in all probability, as well as or better even than a statistician without business qualifications; but the organization demands skillful men at the head of divisions and skillful and trained statisticians as assistants. Every superintendent endeavors to draw into his service a certain number properly qualified, statistically speaking, for the service required; but everything must be drawn together hurriedly—a great bureau, the largest in the Federal Government, created in a brief period, and the work carried on with the greatest rapidity. With the vast expansion of census inquiries, in connection with the necessarily speedy organization, it is absurd, without regard to the qualifications of the head of the office, to expect valuable results for the money expended. It is not in the power of any superintendent, no matter what his experience, no matter what his qualifications may be, to take a very satisfactory census under the conditions involved in our Federal system. The attempt is made to create a vast official machine, and then to at once collect material involving in its collection answers to thousands of inquiries by a force of nearly fifty thousand men in the field and an office force of five thousand, the whole work to be completed within a year or two, and the data to be collected under a system of compensation which does not allow, or certainly does not induce, accurate work. The result is that the Census Office is, within a few months after the date set for enumeration, literally "snowed under" with raw material collected by crude and, in a large majority of cases, inefficient forces, to be digested and compiled for printing by another force nearly as crude as the field forces. It is not in the power of human capacity to carry out scientifically the work of the Federal census. It never has been done; it never can be done until the system is changed. This does not involve any criticism as to the growth of the system nor of the men who have so ably administered it. The point I make is that the census system has grown to be unwieldy in natural ways, and that it is time to correct it, and the very first step toward correction lies in the direction of the establishment of a permanent Census Office, under which there ought to be a constant force of trained and experienced statistical clerks, and the collection of facts distributed over the ten years instead of being crowded into a few months. This change of itself would correct many of the faults of the present system. The facts relating to population and agriculture might be collected in the fall of the census year when the new agricultural crops would be considered instead of the old, as under the present system, and then the data relating to manufactures and all the other features necessarily involved in the census could be taken up year after year and carried each to a successful conclusion. This would involve the employment constantly of a much reduced office force, and a field force, except for the enumeration of the population, gradually becoming more and more skillful. The expense during the whole ten years would be somewhat larger than is now involved, but the results would be of such infinitely greater value that the increased expense would not be a matter for a moment's consideration. My suggestion, then, for future census work would be, first, a permanent Census Office, involving an efficient field force, under the most liberal provisions as to supervision, and an organization of an office force so adjusted that it could be made elastic and yet preserve the functions required to secure accuracy and completeness; second, an adjustment of compensations for field work that would secure complete and accurate returns in all the departments of census work.

It may be argued that there would be nothing for a permanent Census Office to do a great part of the time. In answer to this it can be said, that if the regular work of the census should leave the force in comparative idleness, it might be employed in tabulating some of the results of previous censuses which it was found necessary to abandon; for instance, in 1880, although the facts were secured by the regular enumeration, no tabulation was made of the single, married, widowed, and divorced. The questions now agitating the public mind relative to marriage and divorce are only half discussed, because the facts for the whole country can not be ascertained. This is only one feature. A tabulation of the facts relative to conjugal condition, as indicated, for the year 1880 would be vastly more valuable, even now, than it would have been in 1880. And so of other features. By picking up such abandoned results, a reasonable force in the Census Office could be constantly and profitably employed, with increasing skill, so that when the results of new enumerations came into the Census Office, a trained force sufficiently large to influence the whole body of new appointees would be in readiness.

If, in addition to the changes suggested, the several States could be induced to co-operate with the Federal Government, a great advantage would be gained. The States might undertake the collection of the statistics of population, manufactures, and agriculture on as extended a basis as individually they might choose, but guaranteeing to furnish the Federal Government with certain clearly defined and uniformly collected data, for which the Federal Government should provide reasonable compensation. Under some such adjustment the statistical work of the United States Government and of the individual States could be brought to a very high state of perfection, with the burden of expense so divided and adjusted that it would not be considered as a stumbling-block in the way of progress.

One of the most encouraging movements of the present day is that of the trade and business organizations of the country to secure a perfected and scientific statistical service in this country. This movement commenced during the closing days of the last Congress, through memorials from boards of trade, presented by the National Board of Trade, asking that the question of the establishment of a permanent Census Office be considered by the Secretary of the Interior and a report made to the Fifty-second Congress. The matter is therefore open for consideration by the public and by Congress, and, whether a permanent statistical service is provided for or not, great good must come from the discussion, and ultimately the faulty features of the present system be removed.

  1. In an article in the North American Review for June, 1889, I stated that the English census was taken through the constabulary. I made this statement on most excellent authority. It was, however, an error.
  2. The History, Theory, and Technique of Statistics, by August Meltzen, Ph. D., professor at the University of Berlin. Falkner's translation.