THE CENSUS AGE QUESTION: A REPLY.

By WILLIAM B. BAILEY and JULIUS H. PARMELEE.

Before 1900 not more than one question regarding age had ever been included in the population schedule of the various American censuses. The question had always related to age at last or nearest birthday. For the Twelfth Census, however, an additional age question was inserted-"year and month of birth." As it was the aim of the Census Bureau

keep the number of queries on the population schedule for the Thirteenth Census as few as possible, it was inevitable that one of the two age questions utilized in 1900 should go. To assist in reaching a decision as to which question could best be spared, a study was made by us of the contention that the addition of the date-of-birth inquiry in 1900 had been responsible both for the increased general accuracy of the age returns, and for the decreased concentration on multiples of five. This study seemed to show that the effect of the addition of the date-of-birth query had been somewhat overestimated, and it was decided, therefore, to retain only the inquiry regarding age at last birthday.

The principal facts discovered by our study appeared in the June, 1910, number of these publications, and brought forth keen criticism from Prof. Allyn A. Young in the December, 1910, issue. Such part of Professor Young's article as deals with the inferences or opinions expressed by us we prefer to leave untouched till the age returns of the Thirteenth Census become available. It will be possible within a few months to throw the light of considerable new material on the subject. We shall take up here only two of the specific criticisms made by Professor Young-criticisms that seem to us to be especially faulty or wide of the mark. Professor Young asserts that we failed to give sufficient weight to European experience, and that in our

discussion of the artificial shifting of ages to ages ending in 4 and 9 the statistical evidence proved too much.

Our first paper was prepared on the principle that the reliability of the replies to the double age question in 1900 could be tested only by a study of the evidence directly afforded by the Twelfth Census population schedules. The conclusions set forth in the paper were based, not on European practice or on the recommendations of a statistical congress which may not have been familiar with American census problems, but on the returns actually received and handled by the American Census Bureau in 1900.

On page 368 Professor Young questions our conclusion that the date-of-birth inquiry seems to have been credited · with more than its real effect in the improvement of our age statistics. We are forced to an emphatic dissenting opinion. A certain amount of decrease in concentration on multiples of 5 has occurred regularly for several decades, and could be expected to occur between 1890 and 1900, whatever the form of the age question. On the face of it, therefore, it would not seem that the decrease in concentration in 1900 of itself established the efficacy of the two-question policy, especially as the decrease in 1900 was less than the decrease in 1890. We pointed out, moreover, that part of the 1900 decrease was due to artificial adjustments in the age returns which we believed to be unjustified. Making allowance for this artificial decrease, we find the decrease in concentration on multiples of 5 in 1900 as compared with 1890 even less. Thus the process of lessening age concentration was considerably retarded in 1900, and no great virtue seems to be indicated for the additional age question in that year.

Professor Young goes a step farther on page 368 and accuses us of having proved too much when we pointed out that the artificial shifting of ages brought about by editorial adjustments of the age returns on the census schedules was responsible for a decrease in concentration on multiples of 5 amounting to 2.9 per cent. In a footnote he estimates the increase that would have taken place in 1900 in the proportion of persons reported at ages ending in 9 had such increase been

in the same ratio to the decrease in relative number of ages ending in 0 as was the 1890 increase in the proportion at ages ending in 9 to the 1890 decrease in the proportion at ages ending in 0. Unfortunately Professor Young's arithmetic deserted him in making up this estimate. Three of the four percentages in his footnote* are incorrect, and the increase estimated by him as 0.8 per cent. should actually have been stated as 9.4 per cent. The corrected footnote reads as follows:

"A decrease of 8.7 per cent. in the relative number of reported ages ending in 0 in 1890 as compared with 1880 was accompanied by an increase in the relative number reported as ending with 9 of only 6.6 per cent. On this basis, the decrease of 12.4 per cent. in the number of reported ages ending in 0 in 1900 as compared with 1890 would have been accompanied by an increase of 9.4 per cent. in the relative number of reported ages ending with 9."

This 9.4 per cent. plus the 2 or 3 per cent. estimated by Professor Young† to represent the artificial shifting of ages from ages ending in 0 to those ending in 9, fully accounts for the 12 per cent. of increase that actually took place in 1900 in the relative proportion of ages ending with 9. The number of ages ending in 9 was below normal both in 1900 and 1890. Their approach toward the normal number was no greater, relatively speaking, in 1900 than in 1890, allowance being made for the artificial shifting already mentioned. In this respect the addition of the date-of-birth question would not seem to have proved of great advantage. Our contention, therefore, instead of "losing what had been its statistical support," is on the contrary strengthened by applying Professor Young's own mathematical test.

*It is but fair to Professor Young to state that when his attention was directed to the errors in this footnote he made the same changes as those given here.

† His language is as follows: "I estimate that the artificial shifting described above (if it was as frequent as Professor Bailey and Mr. Parmelee estimate) was responsible for between 2 and 3 per cent. in this increase."

REVIEWS AND NOTES.

SICKNESS, AND MORTALITY EXPERIENCE OF THE LEIPZIG COMMUNAL SICK FUND, 1887-1905.*

The Leipzig communal sick fund embraces several local sick clubs, the aggregate membership in 1909 being 172,617 persons. The fund includes both compulsory and voluntary members, and is by far the most important institution of its kind in the world. It is very gratifying, therefore, to be able to state that the extended experience, mortality and sickness, of this institution for the period 1887 to 1905 is now available to students of life and sickness insurance. The investigation of this experience has been made in great detail and in strict accordance with the best methods of actuarial and statistical science. Sex, age, and occupation are the three primary distinctions maintained throughout the investigation and all the essential facts are also reported separately for the compulsory and voluntary memberships.

The detail work of this great investigation was done by the Imperial Statistical Office with the coöperation of the Imperial Board of Health. The cost of the undertaking which amounted to about $77,350 (325,000 marks) was met by a special appropriation voted by the Imperial Parliament, March 23, 1903.

In any brief review it would be impossible to deal with many of the detail facts disclosed by this experience, and I shall therefore merely point out some of the many lines of inquiry along which the investigation was conducted. The total experience reported upon consisted of 996,445 males and 288,131 females exposed to risk an average of one year. Of the total membership, 1,212,256 were compulsory and 72,320 were voluntary members, and the distribution of the exposure by sex and class of membership was as follows:

* Krankheits-und Sterblichkeitsverhältnisse in der Ortkrankenkasse für Leipzig und Umgegend. Berlin, Carl Heymans Verlag, 1910.

Exclusive of persons employed on railways and in navigation, practically the whole industrial population of Leipzig and vicinity was included in the inquiry. It so happens that the more dangerous and unhealthful trades are included only to a limited extent, as these trades are not widely represented in Leipzig and its environs. The investigation, however, offers to the student the great advantage of presenting the facts in detail both for groups of occupations and for the specific occupations included in the groups. As an indication of the detailed nature of the published results, it may be stated that 323 specific causes of sickness and death are included, and 108 specific occupations of males and 79 specific employments of females are presented in correlation with age and class of membership distinctions.

The report is published in four large volumes aggregating more than one thousand pages. The first volume of 212 pages is entirely devoted to introductory and explanatory matter, summary tables, important conclusions, etc., and the other three volumes contain only the fundamental detail tables, arranged, however, in a systematic manner according to a simple scheme fully explained in the first volume. For example, the second and third volumes give the detail statistics of male members, while the fourth volume contains only statistics of the female membership.

Throughout the investigation careful distinction is made of sickness cases and days of sickness, special emphasis being placed upon the average duration of sickness cases. This information is of special value as it is given by causes of sickness and by specific occupations. The ratios of cases of sickness and of days of sickness per annum per 100 members, with distinction of sex and class of membership-compulsory and voluntary, are presented in great detail with due regard to age periods of life. Wherever the numbers are large enough to justify it, the age divisions are by five year groups from 15 to 74 with two other groupings, one for those under 15 and the other for those 75 and over. The mortality data are presented in actual number of deaths and per 100,000 observed years of exposure. Ratios of deaths to sickness cases and to days of sickness are also presented in detail by age, sex, occupation, class of membership, etc. Great emphasis is also placed upon industrial accidents, particularly accidents met with in the specific employment in which the insured was engaged at the time of his injury. An exceptionally valuable body of facts is presented in this connection for the duration of the disability is given together with the ratio of accident disability to the total sickness disability, which includes disability resulting from accidents.

In volume three a special series of tables are presented for what are termed "Alkoholiker" or cases where there was clear indication made on the sickness card by the medical attendant that the case was an excessive or immoderate drinker. Among other important conclusions deduced from these tables may be mentioned an excessive mortality rate at all ages, an excessive sickness rate, both by cases and days, and an excessive industrial accident frequency. So important were the latter differences shown to