Enumeration Errors in Negro Population

Kelly Miller

The Scientific Monthly, Vol. 14, No. 2 (Feb., 1922), pp. 168-177

168 The Bureau of the Census was established for the purpose of enumerating the population of the United States, and for the collection and collation of other statistical data bearing on the social welfare of the nation. The government bases its calculations upon the information furnished by this bureau. The basis for congressional representation, military conscription and other federal regulations are based upon the census enumeration within the limits of the several states. Publicists and social philosophers base their conclusions upon the same data. It is, therefore, a matter of the greatest importance that the enumeration should be reliable and trustworthy. The Bureau of the Census ranks as a scientific department of the government. Constantly repeated errors of this bureau tend to impeach its scientific reputation and to vitiate the conclusions based upon its output. Numerous complaints have been made by competent critics not only repudiating the results, but also impugning the motive. Manipulation in behalf of sectional and partisan advantage has been freely charged. Senator Roger Q. Mills, in an article in The Forum, bitterly complained that the south was deprived of its due quota of representation by the imperfection of the enumeration of 1890. Indeed, the alleged inaccuracies of the eleventh census provoked a flood of condemnatory literature.

Various enumerations of the Negro population by the Census Office since 1860 have not been very flattering to the scientific reputation of that bureau. These enumerations have been not only inherently erroneous, but so conflicting and inconsistent as to demand calculated corrections. It may be taken for granted that the enumerations up to 1860 were reasonably accurate and reliable. The Negroes, up to that time, were in a state of slavery, and the master had merely to hand the list of his slaves to the enumerator, just as he would the list of his cattle or other forms of chattel. There was every facility and every reason for accurate returns. The Negro population up to 1860 was inflated by importation of slaves from Africa, and, consequently, it was impos169sible to check the accuracy of the count by the ordinary statistical tests. Beginning, however, with the census of 1870, this population has been cut off from outside reinforcement and has had to depend upon its inherent productivity for growth and expansion. It, therefore, becomes an easy matter to apply the ordinary statistical checks to test the accuracy of enumeration.

It is conceded that the enumerations of 1860, 1880, 1900 and 1910 were accurate within the allowable limit of error. According to these enumerations, the growth was more or less normal and regular, and conformed to the requirements of statistical expectation. But the enumerations of 1870, 1890 and 1920 are so flagrantly discrepant as to demand special explanation and correction. A miscount at one enumeration upsets the balance for two decades. If it be an undercount, it makes the increase too small for the preceding decade and too large for the succeeding one. Accordingly, the only consecutive decades upon which we can rely for accuracy concerning the growth of the Negro population would be the 1850-1860 and 1900-1910. In order to escape obvious absurdities, the figures for the other decades must be supplied by reasoned interpolations. The mere exhibit of the several enumerations by the Census Office will convince the student of their inherent improbability.

The irregularities of these figures are as whimsical as if produced by the sport of the gods. The normal growth of population uninfluenced by immigration or emigration shows a gradual increase in decennial increment and a gradual decline in the rate of increase. Wherever there is found to be a wide divergence from this law, it must be accounted for by special contributory influences. The column giving the decennial increments, instead of showing a gradual behavior, jumps back and forth with unaccountable capriciousness. A sudden drop from 803,022 to 438,179 is offset by an alarming rise to 1,700,784 for the next decade, when, lo and behold, there is a swift decline to 907,883 for the following ten years. We look aghast at the upward bound to 1,345,318, thence a downward drop to 993,769, followed by a still further 170 startling decline to 635,250. It makes the head swim to try to keep track of such whimsical variations. The decadal increase per cent, shows similar irregularities. The rhythmical rise and fall of these figures impresses one as the alternate up and down motion of boys playing at see-saw. Why should the ordinates of a curve, which should move smoothly downward, drop suddenly from 22.3 to 9.9, then rise to 34.9 and drop again to 13.9, then rise to 18.0 and decline again to 11.2 with a final slump of 6.5? Such variability has perhaps never been experienced by any human population. The internal evidence of error is overwhelming. The Census Bureau has sought to make corrections for the evidently erroneous enumerations of 1870 and 1890. But the equally discrepant figures of 1920 remain so far indisputed.

The census of 1870 has been universally discredited. The greatest error of enumeration falls, naturally enough, on the Negro race. This race had just been set free, and had not reestablished itself in definite domiciles. Political conditions in the South were in the flux and flow of readjustment. The machinery of the Census Bureau was not sufficiently efficient to cope with so complicated a situation. Statisticians, recognizing the evident error, have tried to correct the mistake by statistical computation. The Census Bureau estimates the error in the Negro population for the decade to be 512,163. An acknowledged error of a half million, it would seem, would put this bureau on the lookout for similar errors in the future. But the census of 1890 was notoriously faulty. Here again the undercount, it is obvious, fell mainly in the South, and largely among the Negro population.

The Census Bureau, in commenting upon the apparent irregularities of returns for 1890, states: “According to the returns, the rate from 1880 to 1890 was very much lower than even the last rate, that of 1870-1880, and the rate for 1890-1900 was much higher than during the preceding or succeeding decade.” Such abrupt changes in a class of the population which is not affected by immigration seem very improbable and almost force the conclusion that the enumeration of the Negroes in 1890 was deficient. In the special volume on “Negro Population of the United States 1790-1915,” the director further declares:

The presumption of an undercount at the census of 1890, therefore, rests upon the improbability of the decennial rates of increase themselves as developed from the census returns; the inconsistency of the indicated changes in the rates from decade to decade with the changes in the proportion of children in the Negro population, and upon the improbability of the decennial mortality indicated for the decades 1880-1890 and 1890-1900. …The number of omissions at the census of 1890 cannot be accurately determined, but it would seem to be a fair assumption that the decline in the rate of increase from 171 decade to decade was constant, and that the rate fell off in each of the two decades 1880-1890, 1890-1900 by approximately the same amount. On this assumption, the probable rates of increase for the four decades, 1870-1910, are 22.0, 17.9, 13.8, 11.2. … A rate of 17.9 per cent, for the decade 1880-1890 would give a Negro population in 1890 of nearly 7,760,000, which, in round numbers, exceeds the population as enumerated at the census of 1890 by 270.000. This is probably the number of omissions of Negroes at the census of 1890, on the assumption that the retardation in the rate of growth in the 20 years 1880-1900 was constant.

By making the estimated corrections for acknowledged error in the counts of 1870 and 1890, decadal growth from 1880 to 1890 would be reduced and from 1890 to 1900 increased, so as to produce reasonable conformity with the laws of normal growth. A gradual decline in the rate of growth from 22.3 per cent, to 11.2 per cent, in 60 years will prove that the Negro element conforms to the regular law of human population. This decline would appear even more gradual if we consider that the rate of 22.1 from 1850 to 1860 was contributed, in considerable measure, by African importation. The Census Bureau offers the following table with corrected numbers for 1870 and 1890:

According to the recent bulletin issued by the Bureau of the Census, the Negro population showed a surprising and unexpected decline during the last decade. In 1910 there were 9,827,763 Negroes, and in 1920 10,463,013, giving a decadal increase of 635,250 or 6.5 per cent. If these figures were added to the table corrected to 1910, the disparity would be as glaring as any which has yet come from the Census Bureau. The sudden drop in decadal increase from 993,769 to 635,250, or from 11.2 per cent, to 6.5 per cent., is so strikingly out of harmony with the more or less regular movement of the table as to call loudly for correction or explanation. The table shows a gradual decrease in the decennial increment from 1880 to 1910, a decline of 194,852 in three decades. But now we are called upon to accept a sudden decline of 358,519 in a single decade.

The decennial rate of increase dropped from 11.2 per cent, between 1900 and 1910 to 6.5 per cent, between 1910 and 1920, 172 whereas we should have expected a gradual decline of not more than 1 or 2 points. On the face of the figures it seems probable that the Census Bureau has again committed an error in the enumeration of the Negro population. As this bureau has admittedly committed grave errors in enumeration of the Negro population in two preceding censuses, it is but reasonable that the obvious discrepancy can be most reasonably accounted for by an error in the present count.

Aside from the internal evidence itself, there is sufficient reason to suppose that this count might have been erroneous. The mobile Negro population has been greatly upset by the world war. There was a mad rush of Negroes from the South to fill the vacuum in the labor market caused by unsettled conditions. Thousands of Negro homes were broken up and their members scattered without definite residential identity. In the cities especially, it seems probable that the count was greatly underestimated. The Negro migrants lived for the most part in improvised lodgings and boarding houses whose proprietors had little knowledge and less interest in the identity of the boarders. The census official, visiting such boarding houses with a large number of Negro boarders would, in all probability, receive an inaccurate underestimate by the ignorant and uncaring proprietors. As an illustration of such inaccuracy, I cite a quotation from an editorial of the Dispatch of Oklahoma City:

If the census enumerators over the United States were as careless in the count as they were shown to be by this publication during the poll of the population last year, the general charge is right that the black man has made a much larger numerical advance than the official, yet faulty, records show. It will be remembered that the Dispatch made the charge during the enumeration that there was a laxness and really seeming desire to overlook the black man in this city. Our charge was printed in the daily papers. To cap it all off, the irate enumerator in the section of the city where the Dispatch is located, appeared on the evening that the charge was published, and demanded of the editor the basis of the charge. We took him out into the 300 block on East 2nd Street and found 33 black men whom he had not counted, folk who told him so, and whose names he did not have on his lists.

If a presumption of undercount was justified by the statistical indication for 1870 and 1890, surely a like presumption would obtain for the census of 1920. There are but three methods of accounting for this sudden slump in the growth of the Negro population. First, an undercount of the Census Bureau, second, a sudden increase in the death rate, and third, a decrease in the birth rate of the Negro population.

It is known that the death rate of the Negro is decreasing rather than increasing under improving sanitary conditions and general 173 social environment. The Director of the Census states that ‘‘the death rate has not changed greatly.’’ Instead of adhering to the “fair assumption” of a steadily declining rate of increase, as was done for the faulty enumerations of 1870 and 1890, the Director of the Fourteenth Census accepts the violent leap from 11.2 to 6.5 and endeavors to vindicate the count of 1920, by assuming a sudden decrease in Negro birth rate.

On this point the Census Bureau explains:

The rate of increase in the Negro population, which is not perceptibly increased by immigration or emigration, is by far the lowest on record. This element of the population has been growing at a rapidly diminishing rate during the past 30 years, its percentage of increase having declined from 18 per cent, between 1890 and 1900 to 11.2 per cent, during the following decade and to 6.5 per cent, during the 10 years ended January 1, 1920. Such data as are available as to birth and death rates among the Negroes indicate that the birth rate has decreased considerably since 1900, while the death rate has not changed greatly.

The statement, “this element of the population has been growing at a rapidly diminishing rate during the past 30 years,” that is, since 1890, presupposes the accuracy of the census of 1870, which presumption the census office itself discredits in a previous statement. It entirely overlooks the fact that the rate rose suddenly from 13.8 for 1880-1890 to 18.0 for 1890-1900. With the indicated corrections the rate of increase has declined within the expected limits of fluctuation from 22 per cent, for the decade 1850-1860 to 11.2 per cent, for the decade 1900-1910, making a drop of 10.8 points in 6 decades. The sudden downward drop by 4.6 points in a single decade certainly calls for a more satisfactory explanation than a sudden and unaccounted for decrease in birth rate. The only statement which the Census Bureau vouchsafes to account for this rather startling conclusion is a very hesitant and uncertain one:

Such data as are available with regard to birth and death rate among Negroes indicates that the birth rate has decreased considerably since 1910, but the death rate has not changed greatly.

On examining the data on which this conclusion is based, we find that they are wholly insufficient to justify the sweeping conclusion imposed upon it. The mortality statistics are based upon returns from the registration area. Only five southern states are now included in the area, namely, Maryland, Virginia, North Carolina, South Carolina and Kentucky, from which birth and death rates are collected annually, and even these states were not admitted to the birth registration area in 1900. So that the computation of birth and death rates for the colored population of these states is neither adequate nor convincing.

174

Those are the only heavy Negro states within the registration area.

These states were not all included in the registration area for 1900. Mortality statistics in the non-registration area are notoriously inaccurate and unreliable. Birth registration is especially unsatisfactory even in the registration area.

Return of Negro births would naturally be most inaccurate. Negro births, especially in rural and small urban communities are not always attended by regular physicians or certified health officials. The midwife still plies her trade. There is a relatively large number of illegitimate births among Negroes. Official returns in such cases would not be apt to be rendered fully for prudential reasons. It is therefore evident that the rapidly declining birth rate revealed by the census is based upon noncomparable and inadequate data.

Even the apparent rapid increase in the white death rate awaits fuller explanation before the figures can be relied upon with assurance. It is curious to note that the birth rate among the whites in South Carolina fell from 32.3 in 1900 to 27.1 in 1919, the death rate rising but slightly from 10.4 to 10.6 during the same interval. And yet the white population of that state increased from 557,807 in 1900 to 818,538 in 1920. There was a vigesimal increment of 250,731 with little or no reinforcement from immigration. This unexplained increment in the white population seems also to discredit the reliability of the recorded mortality statistics within the states so recently added to the registration area.

It is well understood that these states, except South Carolina, 175 have shown a comparatively slow rate of increase in Negro population for 30 years preceding the census in question. The facts are indicated in the following table:

From the table it will be seen that the increase in Negro population in the southern states within the registration area has been considerably lower than that for the country at large. In Maryland, there is an actual decline in the Negro population of 1.2 per cent, from 1900-1910 and the small gain of 2.6 from 1880-1890. In Virginia the highest rate of increase during the past 30 years was 4 per cent. In Kentucky there was an actual decline for two of the three decades. The low rate of increase in the border states is due to the large emigration of the Negro from these states to the nearby northern states and cities. It is well known that the Negro who migrates to the North and the large cities is made up of younger people of both sexes who, if they had remained at home, would naturally tend to increase the birth rate.

The low birth rate revealed by the census in these states is due to the migration of the Negro population of reproductive age from those states within the registration area. This, of course, does not affect necessarily the birth rate of Negro population as a whole. A better view of the birth rate of the Negro population may be secured by considering the growth of this population in the more typical southern states not so much affected by migration during the same period.

Thus it will be seen that four heavy Negro states, with an aggregate Negro population of nearly four million, shows a rate of increase far greater than those in the registration area. The in176crease in those states was due wholly to the excess of births over deaths. But this does not tell the whole story. While the stream of migration was not so pronounced from these states as from the northern tier of southern states, still there has been a considerable northern movement for the past three or four decades.

From a comprehensive view of the whole situation, it seems perfectly clear that the sudden decline of the Negro population as revealed by the census of 1920 is due to miscount rather than to the declining birth rate. If we should estimate an error in count of 300,000, scarcely greater than that conceded by the Census Bureau itself for the count of 1890, the Negro population during the past 60 years would have followed more or less consistently the ordinary laws of growth. Let us accept the substantial accuracy of the census of 1860, 1880, 1900 and 1910 and estimate the error for 1870 at 512,163, for 1890 at 270,000, as conceded by the Census Bureau, and let us still further allow an error in the count, 300,000 for 1920, as here suggested. The growth of the Negro population since 1850 will be as follows:

The table makes the Negro population behave more or less normally, and is certainly more reasonable than the startling deviation revealed by the face of returns, and the explanation is more acceptable to reason than that urged by the Census Bureau, ctf a sudden and unexplainable decline in the Negro birth rate.

It is a source of surprise to note that the American mind seems to expect that any fact which affects the Negro will deviate from the normal course of human values. It is prone to accept with satisfaction wild assertions and unsupported theories, without subjecting them to the test of logic and reason. If it is seen in the Census, it is so. Any statement issued upon the authority of the government which seems to be belittling to the Negro will be seized upon by would-be social philosophers and exploited throughout the nation to the disadvantage of the race.

De Bow, relying upon the low rate of increase in the Negro population, revealed by the census of 1870, proved to the entire satisfaction of those who were satisfied with this type of proof 177 that the Negro could not withstand the competition of freedom and would, forthwith, fall out of the equation as an affected factor. The census of 1880, showing the unheard of increase of 34 per cent., set all of De Bow’s philosophy at naught. But thence arose another school of philosophers which declared that this unheard of increase in the Negro population threatened the numerical ascendancy of the white race, and, therefore, the black man should be returned to Africa from whence his ancestors came. The census of 1890 refuted this conclusion by showing only an increase of 13.8 per cent., but, no whit abashed, another type of anti-Negro propagandism arose, declaring that the rapid decline in the race indicated inherent, degenerative physical tendencies threatening to the health and stamina of the American people. The census of 1900, showing a rise of decadal growth to 18.0 per cent., produced a calm in the domain of social speculation. But the preceding prophecies of evil are still of record. It seems to be the nature of the prophet to ignore the failure of the fulfillment of his prophecies.

It is particularly unfortunate that such loose and unscientific propaganda can be bolstered up by data from governmental documents which the uninquiring mind is disposed to accept with the authority of holy writ. The calamity philosophers have already dipped their pens in ink to damn the Negro race to degeneration and death by reason of the latest census figures. The thought, and perhaps the conduct, of the nation may be misled on the basis of erroneous data, backed up by governmental authority.

The broader question arises in the scientific mind. If the data on Negro population furnished by the census can not be relied on, as is clearly shown by past enumerations, what assurance is there that collateral information, such as death rate, birth rate, occupation, illiteracy, etc., are to be given full credit and confidence. The Negro problem is the most complicated issue with which we have to deal. Straight thinking and sound opinion based upon accurate data are absolutely necessary to enable us to reach any conclusion of value. The Census Office has now become a permanent bureau, which, it is hoped, will take rank with other scientific departments of the government.

Statesmen and publicists should have serious concern about the accuracy of Negro statistics in view of the importance of the political and sociological conclusions based upon and derived from them.