The Problem of Historical Statistics

Following is one of the head-notes from the Introduction to the first (1949) edition of "Historical Statistics of the United States 1789 – 1945.” Perhaps it could be helpful in understanding the significance and importance of having a vast reservoir of historical information. Actually, this came late (1949) in the sea change of information gathering that began with plans for a national Accounting System very early in the Great depression.

Copied directly from the 1949 edition.

The Problem of Historical Statistics

The statistics of the Nation are an important and even indispensable tool in the proper portrayal of the status of the United States in various subject fields at various periods in time. There are surprisingly few general fields in which existent figures cannot supplement or clarify the qualitative historical records—figures which were compiled year by year during the course of events, or were reconstructed later on the basis of existing statistical evidence.

The extent to which statistical data are cited or taken into account in historical writings is frequently dependent upon the ready availability of the needed data to the writers. An understandable lack of knowledge as to the existence of historical statistics in a given field, and the relative inaccessibility of the volumes in which they may be found, combine to prevent more widespread and effective use.

True, in some subject fields statistical time series are entirely lacking, particularly figures already arranged year by year. In many important fields, however, the past publications of the Nation, public and private, contain a wealth of data periodically compiled which reflect the fact that “a strong passion for statistics early developed itself in the life of our people….” (The phrase in quotes was cited as being from a speech by Francis A. Walker, superintendent of the Ninth and Tenth Censuses of the United States [1870 and 1880], delivered before the International Statistical Institute, Chicago, 1893.)

Sources of data. Among the numerous sources of historical statistics of the United States are the annual reports of the various Departments, Bureaus, and other agencies of the Federal Government, reports of special Federal commissions established from time to time, the volumes of the various censuses of the United States, printed debates of Congress, the published reports of Committees of Congress and the transcripts of hearings conducted by them on important legislative measures, the published reports and documents of State Governments, the statistical publications of private research foundations and organizations and of universities and colleges of the Nation, and the great mass of statistical and other volumes printed privately by other organizations and individuals.

Difficulty of accessibility. The accessibility of these great masses of historical data to those who wish to use them is another matter. As matters stand, Senators and Congressmen, public officials, economic, social, and political historians, research workers, teachers, students, journalists, and authors, to mention only a few groups, who wish to consult the historical statistics “available” in published form on a given subject are faced with three major difficulties:

First, the determination of the existence of the data and the identification of the exact public or private document or volume in which the data may be found. Frequently, this requires a knowledge of the responsibilities of government bureaus in years long past, and the scope, coverage, and formal description or title of their official reports. The exact material which is desired may already have been compiled, but it may well be buried in an obscure special report or in the published documents of an early Congress—publications which few libraries may have on their shelves.

Furthermore, the present staff of the Government Bureau now responsible for the given subject field may have only a meager knowledge of the detailed statistics contained in the early reports and published documents of their own and predecessor agencies.

Second, once an exact published source has been determined, the data may not be found already arranged in the form of a time series.

In some historical fields, such as banking, merchant vessel tonnage, and the fiscal affairs of government, current or fairly recent government documents carry extensive statistical summary tables which provide the more important data carried far back in time year by year. In other fields, the annual reports and other documents present figures only for the most recent year. Accordingly, historical tables must be constructed laboriously by reference to as many volumes as there years to be considered. This requires access to a collection of such volumes possessed by few libraries.

Even in Washington, D.C., at the seat of government, where the extensive collections of the Library of Congress may be supplemented by reference to Bureau and departmental libraries and records, the compilation of long-term time series by reference to individual annual reports is a laborious and time-consuming task.

Third, identification of changes in concept and coverage over a period of time is important since such changes may affect vitally the interpretation of the statistics for a span of years. Coupled with this is the need for definitions of terms employed in published historical tables, definitions which may be in a separate publication or may never have been published.

A ready “solution” of the first two difficulties by location of a previously compiled time series may prove deceptive. As a general rule, historical tables in government documents represent compilations of figures with a minimum of text and tabular notes. Definitions, where given, are usually for the current year; qualifications of data, particularly warnings as to changes in concepts and collection methods over the years, are often inadequate.

Grave risks are entailed in attempts to compile long-term time series of annual data by reference to successive issues of the Statistical Abstract of the United States. In each issue, many revised figures appear. Generally, these are for the immediately preceding year or years, but revisions in the more historical data are not uncommon. Statistics for the specific years for which data are shown in the most recent Statistical Abstract may be used with no greater precaution than is necessary when making use of any highly abbreviated presentation of historical statistics. However, the attempt to fill in data for omitted years by reference to earlier issues may well result in serious lack of comparability through time.

In view of the broad difficulties outlined above, it is apparent that no single reference volume can offer a complete solution to the problem of historical statistics, as such. At best, it can provide a selection of statistics in a number of broad subject fields and constitute a guide to the more basic and detailed sources of data—functions which comprise the objectives of Historical Statistics of the United States.