Relation of the Federal Goremment to Research 



221 



lytical research has declined both in quantity and 

 quality. 



Question of Future Policy 



With the rapid growth of social-science research in 

 universities, business, and privately endowed research 

 organizations generally, there is less dependence on the 

 Bureau of the Census for analytical research which 

 has the general aim of making additions to knowledge 

 than was the case a generation ago. Even as late as 

 the first decade of this century, there was hardly any 

 university or private organization equipped to do ex- 

 pensive analytical work with social and economic data. 

 We may ask : How does this change affect the Bureau's 

 responsibility ? 



Two extreme points of view may be presented. 



1. The Bureau of the Census should uo longer do research 

 of this type, except for brief interpretjitive texts accompanying 

 its basic routine tables. 



2. The Bureau should do an increasing amount of special 

 research on social and economic problems, documenting par- 

 ticular historical trends and developin;; and testing hypotheses 

 about social and economic relationships. 



In support of the first position several kinds of argu- 

 ments may be presented. 



(o) It may be argued that research of this character should 

 not be done by the Government if it can be done better bj- 

 outside scholars and organizations. The most effective re 

 search, it is said, grows out of pressure to solve specific prob- 

 lems. The university research man is in closest touch with 

 theory and is most likely to feel a pressure to put competing 

 theories to a quantitative test. The research organizations in 

 business may have behind them even greater pressure to solve 

 particular problems. Moreover, the university or other out- 

 side research man might have a technical knowledge of the 

 special field in which the problem lies, superior to the knowl- 

 edge which a career man in an agency like the Bureau of the 

 Census could have. 



(6) It may be held that such research is a proper function 

 of Government, yet belongs not in an organization like the 

 Bureau of the Census but in some central planning or research 

 organization, such as the National Resources Committee might 

 become. The Bureau of the Census is only one of more than 

 90 Government agencies which collect and compile statistics. 

 Although its coverage embraces almost every field of social 

 and economic life, there is hardly a .single field in which it 

 is the sole reporting agency. The facts bearing on a particular 

 historical investigation or social or economic hypothesis may 

 need to be assembled from the basic statistics collected by sev- 

 eral agencies. Planned and executed in a central agency con- 

 cerned with broad problems of immediate or ultimate impor- 

 tance to the country, such research would have a good chance 

 of infiueneing governmental policy. Moreover, if research men 

 were concentrated in a central office organized mainly for 

 analytical work and freed from the routines of superintending 

 the mass production of data, there would be a critical inter- 

 stimulation in such a staff which would insure the highest in- 

 tellectual standards and the maximum of scientific esprit de 

 corps. 



(c) An agency, like the Bureau of the Census, whose funda- 

 mental task is that of providing the raw material for both re- 



search and administration, runs some risk of weakening its 

 reputation for authoritative accuracy if It indulges too much in 

 research requiring refinement upon refinement of the data. .\s 

 research proceeds upward from the primary tabulations, and as 

 operations — sometimes involving I he theory of probability — -are 

 applied, there often enters an increasing subjectivity and per- 

 sonal judgment. 



(d) At the present time, fact-collecting agencies like the 

 Bureau, have not nearly enough funds to provide the requisite 

 raw data. If substantial additional money were not forth- 

 coming, additional research of the type here discussed might 

 draw off funds indispensable for basic reporting. 



In support of the opposite extreme position, namely 

 that the Bureau of the Census should do an increasing 

 amount of special research on social and economic 

 problems, documenting particular historical trends and 

 developing and testing hypotheses about social and 

 economic relationships, several arguments may be 

 offered. 



(a) The Bureau is a great storehouse of unpublished data 

 and those in charge of deciding what to tabulate and which 

 of the tabulated material to ijublish should be in an ideal 

 position to make intelligent and efficient use of these source 

 materials in further research. It is easy to cite examples of 

 errors committed by research people who did not know how 

 the field and editing instructions behind a given body of data 

 actually were interpreted. 



(b) Continuing research experience and close contacts with 

 the live problems of the day in sociology, economics, and busi- 

 ness are desiralile if division chiefs in the Bureau are to make 

 wise recommendations and decisions about what questions to 

 ask on the basic Census schedules, what definitions to make, 

 what cross-classifications to tabulate, and what to publish. 



(c) If it he conceded that research on the accuracy of its 

 basic data is an obligation of the Bureau of the Census, one 

 may point out that one of the most effective ways of detecting 

 inaccuracies and other defects in data is to conduct some re- 

 search with the aim, not directly of discovering the defects, but 

 rather of making additions to scientific knowledge. 



(d) If the Bureau of the Census is to offer a career to profes- 

 sionally trained men of high intellectual qualities, it nuist pro- 

 vide for some stimulating intellectual work, which will enhance 

 their reputation as social scientists. Otherwise, the best men 

 will not stay in the Bureau, and the basic data of the social 

 sciences will be collected and tabulated by routineers without 

 incentive for providing data of maximum accuracy and utility. 



(e) Often, without interrupting the ordinary routines of 

 tabulating regular census data, it is possible to provide for a 

 particular type of cross-tabulation which would illuminate a 

 special research problem. Thus, the Bureau of the Census, 

 if it had its own research men watching for such opportunities, 

 might make at practically no expense some scientific contribu- 

 tions whose cost would be prohibitive if carried on as an inde- 

 pendent operation. In view of the overhead costs of a decen- 

 nial census proper (nearly forty million dollars in 1930), the 

 slight cost of subjecting some of this vast material to special 

 research is insignificant as compared with the increased social 

 and economic value which a few additional operations on the 

 data would give to the whole census. 



Weight of argument in favor of research xcifhin the 

 Bureau. — In the writer's judgment, the weiglit of the 

 argument is decidedly in favor of research within the 



