SS .1 *<°\ c <^ ''Vis o« ' Census Tract Papers Series GE-40, No. 8 Small-Area Statistics Strengthening Their Role in Federal Government and Their Use in Criminal Justice Programs U.S. DEPARTMENT OF COMMERCE Social and Economic Statistics Administration BUREAU OF THE CENSUS Census Tract Papers Series GE-40, No. 8 Strengthening Their Role in Federal Government and Their Use in Criminal Justice Programs Issued May 1972 U I 3 Papers Presented at the Conference on Small-Area Statistics American Statistical Association Ft. Collins, Colo. August 23, 1971 * f «TtS O* r U.S. DEPARTMENT OF COMMERCE Peter G. Peterson, Secretary James T. Lynn, Under Secretary Harold C. Passer, Assistant Secretary for Economic Affairs and Administrator, Social and Economic Statistics Administration BUREAU OF THE CENSUS, George Hay Brown, Director BUREAU OF THE CENSUS George Hay Brown, Director Robert L. Hagan, Acting Deputy Director Paul R. Squires, Associate Director GEOGRAPHY DIVISION Morton A. Meyer, Chief Library of Congress Card No. A68-7769 SUGGESTED CITATION U.S. Bureau of the Census, Census Tract Papers, Series GE-40, No. 8, Small-Area Statistics: Strengthening Their Role in Federal Government and Their Use in Criminal Justice Programs, presented at the Conference on Small-Area Statistics, American Statistical Association, Ft. Collins, Colo., August 23, 1971. U.S. Government Printing Office, Washington, D.C., 1972. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402. or any of the Department of Commerce Field Offices. Price 55 cents. PREFACE The papers published in this report were presented at the Small-Area Statistics Conference at Colorado State University in Fort Collins, Colo, on August 23, 1971 during two sessions of the annual meeting of the American Statistical Association (ASA). This conference was sponsored by the ASA Committee on Small- Area Statistics with the program planned by Albert Mindlin, Chairman of the Committee. The first session featured several speakers who have served as consultants or contributors to the President's Commission on Federal Statistics and dealt specifically with Strengthening the Federal Role in Small- Area Statistics. Albert Mindlin chaired this session and has provided an introductory statement. The second session on Small- Area Statistics in Criminal Justice was developed from the Federal-State-local cooperative program of the Law Enforcement Assistance Administration (LEAA) of the U.S. Department of Justice. The program was arranged and chaired by George Hall, Chief, Statistics Division, LEAA. This report was organized and prepared under the direction of Robert C. Klove, Research Geographer, Office of Associate Director for Statistical Standards and Meth- odology, Bureau of the Census. Ill CONTENTS Page STRENGTHENING THE FEDERAL ROLE IN SMALL-AREA STATISTICS-REPORTS TO THE PRESIDENT'S COMMISSION ON FEDERAL STATISTICS Introduction to the Morning Session 1 Chairman: Albert Mindlin, Government of the District of Columbia Improvements in Federal Statistical Programs for Small Areas 2 Albert Mindlin, Government of the District of Columbia Institutional Aspects of Subnational Data Collection and Usage 13 Wilbur A. Steger, President, CONSAD Research Corporation Intrastate Statistical Coordination 17 Herbert Alfasso, Director, Statistical Coordination, New York State Division of the Budget Fedcral-State-Local Cooperation in Statistical Data Production 28 Morris B. Ullman, Consultant, President's Commission on Federal Statistics Comments on Papers on Small-Area Statistics 31 Daniel B. Rathbun, Executive Director, President's Commission on Federal Statistics SMALL-AREA STATISTICS IN CRIMINAL JUSTICE-ACTIVITIES OF THE LAW ENFORCEMENT ASSISTANCE ADMINISTRATION The Law Enforcement Assistance Administration Program 35 Edwin R. La Pedis, Law Enforcement Assistance Administration Statistics Division Support for Criminal Justice Programs 37 Patti Rowan, Law Enforcement Assistance Administration San Jose Pilot City Criminal Justice Statistical Information System 40 Paul K. Wormeli, Vice President, Public Systems, Inc. and Robert L. Marx, Vice President, Public Systems, Inc. IV Strengthening the Federal Role in Small-Area Statistics -- Reports to the President's Commission on Federal Statistics Digitized by the Internet Archive in 2012 with funding from LYRASIS Members and Sloan Foundation http://archive.org/details/smallareastatistOOconf INTRODUCTION TO THE MORNING SESSION Albert Mindlin Government of the District of Columbia The morning session of the 1971 ASA Conference on Small-Area Statistics dealt with the President's Com- mission on Federal Statistics. All of the papers dealt with reports presented to the Commission on improve- ments in the Federal statistical system for small areas-- SMSA's, counties, cities, neighborhoods. Mindlin presents the results of a nationwide survey of small-area data needs as perceived by their most direct local users. The American Statistical Associa- tion and the American Institute of Planners sent to hundreds of practitioners a call for comments, ex- periences, frustrations, problems, and recommendations. Nearly 100 replies were received. The author organized the replies, classified them and reduced the material to a coherent package. He then selected 12 major problems with associated recommendations, that the replies in- dicated were of top priority. On these 12 topics he expanded the survey replies by additional contacts with Federal agencies and further research. Steger attempts an estimate of the sheer magnitude of resources currently being spent to produce and analyze subnational statistics, then proceeds to a brief discussion of some basic institutional trends in sub- national statistical activities that he forsees for the 70' s, and finally raises several global issues that he believes would have to be examined in any thorough study of a Federal statistical system for small areas. Alfasso is concerned primarily with the evaluation of State statistical coordination functions and agencies, following the recommendation for the development of such agencies which was formulated at the National Governors' Conference on Comparative Statistics in 1966. Based on the responses of 37 States to a questionnaire submitted to all the States, plus an in-depth study of three States, as well as his own work in this field, he describes the present status of such functions on the State level, the stake of the Federal Government and its present role, and develops a set of action recommendations for both Federal and State Governments. Ullman also addresses himself to the issue of Federal- State-local cooperative programs. Having had pre- Conference access to the ASA-AIPand Alfasso papers, he discusses various points made in those papers, pointing out some problems and suggesting alternatives. The papers taken together present an effective picture of most of the material on small-area statistics that was submitted to the President's Commission on Federal Statistics. Daniel Rathbun, Staff Director of the Com- mission, was present at the Conference as a discussant. His remarks also were captured for publication. IMPROVEMENTS IN FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS (Recommendations made to the President's Commission on Federal Statistics submitted by the Committee on Small-Area Statistics, American Statistical Association and the Census Committee Information System Department, American Institute of Planners) Albert Mindlin Government of the District of Columbia INTRODUCTION On August 12, 1970, the President appointed a Commission on Federal Statistics. The overall charge to the Commission was to undertake a comprehensive review of the statistical programs of the Federal Government and recommend improvements. The Com- mission solicited from the statistics using community recommendations for improvement in the Federal Government's role in statistics. The pertinent com- mittees of the American Statistical Association and the American Institute of Planners have developed in the present document a joint response to this solicitation in the specific field of small areas- -metropolitan areas, counties, cities, neighborhoods, census tracts, blocks, and other small areas. The urban revolution in American society that has resulted in massive shifts of population to metropolitan areas, in kaleidoscopic mobility and escalation of urban problems to the point where they are now and in the foreseeable future a dominant concern of domestic American politics, is too well known to require elaborate statement here. Equally well known is the irreversible historical trend toward social and economic improvement and security through governmental effort. State and local government expenditures, exclusive of Federal grants, rose from $43 billion in 1959 to $97 billion in 1969, an increase of 130 percent in 10 years. Even in constant (1967) dollars, the increase was nearly 80 percent. Federal grants to State and local governments rose from $6 billion in 1959 to $20 billion in 1969, over 230 percent increase, or 157 percent in constant (1967) dollars. Indeed, the increase in Federal aid is accelerating. Between 1968 and 1969 the increase in current dollars was 5 percent; between 1969 and 1970, it was 20 percent; between 1970 and 1971, it has been estimated by the Office of Management and Budget at 25 percent; and between 1971 and 1972 at 27 percent- - from $20 billion to $38 billion, nearly 100-percent increase in 4 years. The information about the community that is needed to research, plan, administer, and evaluate the vast governmental community programs that are by now a permanent feature of American society, has escalated in parallel with the new activities themselves--and by community is not meant national or regional, or even State information, but information on a small-area level- -county, city, neighborhood- -and not only in the large cities, but in the medium cities and small towns as well. Vast Federal expenditures rest on microdata about thousands of small areas. For any sensible efficiency, not only of local operations, but of operations of the Federal Government, the Federal statistical system must become sensitized and responsive to the need for comprehensive, accurate, and timely information on small areas. This report is the result of a special call for rec- ommendations which was sent by the AIP and ASA to numerous organizations and persons all over the country actively engaged in the use of Federal small-area statistics. Ninety-four replies were received. These replies represent substantially more than the 94 letter signers. Many of the organizations circulated the call and incorporated comments of numerous agencies into a single reply. Most of the replies were signed by directors or research directors of agencies. By letter- head, the replies came from 21 city planning and other city agencies, 8 counties, 18 regional organizations, 10 States, 10 Federal agencies, 18 nongovernmental organizations, and the remainder from individuals. Some of the replies were from well-known authorities in the field of small-areas statistics who well understand the inner workings of the present Federal statistical system, including its financial, organizational and tech- nical problems and limitations. However, the bulk of the replies represented agencies and persons all over the country that do not have this professional sophistication, but that are immersed in the daily practical problems of the urban revolution and that are operating State, regional, county, city, and neighborhood programs that rely heavily on data. Thus, with some notable exceptions, the replies were not expressed in terms of fundamental issues of the Federal statistical system. Rather, they were expressed in terms of the real, practical, immediate problems and frustrations with Federal statistics that are encountered every day by State, regional, and local users, users who are in direct interface with the ultimate target of the whole system- - the American citizenry. However, the implications of many of their replies do indeed involve some fun- damental issues of the Federal statistical system. The replies contained over a hundred separate rec- ommendations. As would be expected, there was a heavy overlap of recommendations. Certain recommendations were made again and again, reflecting common problems and priority concerns. It is not feasible in this paper all of the recommendations. We urge to present the Commission to peruse these communications, in order IMPROVEMENTS FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS to capture their flavor and atmosphere, and also in order to consider the many recommendations that we are unable to include in this summary. We have chosen to summarize 12 major problems with associated recommendations, that seemed to us to be of top priority to the small-area statistics user community, as judged by the frequency and vigor with which they were stated. These are: 1. The need for a quinquennial census. 2. The problem of timeliness of Federal statistics and the related needs for increased Federal assistance in local surveys and an expanded Current Population Survey. 3. The need for adequate ongoing housing data. 4. The need for improved small-area statistics of crime and the criminal justice system. 5. Improved labor force statistics including em- ployment data by occupation and place-of-work data, and work- residence patterns. 6. The need for current small-area statistics on income. 7. The need for improved geographic identification. 8. The need for wider knowledge of Federal statistics for small areas, the characteristics of Federal data files, and the nature of the Federal statistical system. A QUINQUENNIAL CENSUS The overriding issue mentioned, far more often and with more vigor (one might almost say more agony) than any other is the need for a quinquennial census of pop- ulation and housing. So much has been said and written on this subject in recent years that it seems hardly necessary or even possible to say anything new. Therefore, only a few remarks are made here to attempt to convey the importance and urgency of this issue. It is no ex- aggeration to assert categorically that the U.S. Census of Population and Housing constitutes the informational backbone of most social planning and evaluation carried on throughout the country. It constitutes the principal data base supporting most Federal, State and local programs, enormous Federal expenditures, the dis- tribution of these resources among thousands of very small areas such as local neighborhoods in hundreds of cities, and the evaluation of the effectiveness of these efforts and expenditures. Under the dynamic cir- cumstances of contemporary American life, to obtain only once every 10 years the vitally needed social and economic information, underpinning public and private policy, in the subject and geographic detail required, is a glaring anachronism. The Bureau of the Census now conducts several censuses- -population and housing, busi- ness, manufacturing, governments, transportation, and agriculture. Every one of these except the population and housing census is authorized and conducted each 5 years. Yet for the success of the vast and complex programs generated by the urban revolution, such programs extending into the deepest neighborhood and the smallest town everywhere in our land, the census of population and housing is by far the most important of all information sources. There is no remotely adequate substitute for this census in the foreseeable future either in other Federal programs or in State or local programs. In light of this most important of all needs of small- area statistics users, a need that has been heavily documented in multiple congressional hearings and hundreds of pages of congressional testimony extending over several years, it is shocking to learn that the present national administration has recommended against a middecade census on the ground that it would be too expensive. The proper use of literally billions of Federal dollars depends on accurate small-area statistics. By the last half of the decade many of the statistics for small areas from the decennial census--the only source of the vast bulk of required data- -are highly unreliable. The out-of-pocket cost of a middecade census must be balanced against the vastly greater cost of misdirected programs and misspent funds stemming from poor statistics. If the President's Commission on Federal Statistics can play a major role in persuading the Federal Government to take the census of population and housing quinquennially and if it accomplishes nothing else, it will have made a major contribution to the modernization of our Federal statistical system. MORE TIMELY STATISTICS The second commonest problem expressed in our survey was widespread dissatisfaction over the lack of timeliness of Federal statistics for small areas. There are two dimensions to this problem: the pro- longed delay in processing and releasing the results of surveys and other data collection programs; and the infrequency of important surveys or other data programs. With respect to the first dimension, again and again complaints were expressed over the delay in release of 1970 census data. Indeed, it is now well over a year since that census was taken, yet only the barest trickle of information has been released. There is a widesprad belief that the 1970 data are much behind earlier censuses in timeliness of release. It should be noted in this connection that interest in census data has greatly increased, not only for the reasons stated in point 1, but also because far more local resources were involved in preparing the 1970 census than any previous census, particularly in the preparation of the Address Coding Guide. Hence, there was a greater sense of participation. Also there has been substantial publicity about the summary tapes. So there has been much more sen- sitization and eagerness for the results than ever before. Impatiently we wait and wait- -and the data get older- - and older. Indeed, the critical income data, on which so much analysis and activity ride, may be as much as 2 or 3 years old when released. These complaints are, of course, a compliment to as well as a criticism of the census, for they reflect the central role that the census of population and housing has come to play in the social programs of the country and our intense dependence on it. Also, the difficulties encountered by STRENGTHENING THE FEDERAL ROLE the 1970 census reflect the effort of census officials to expand greatly the subject matter and geographic detail of census output and in a multitude of ways make this tremendous survey more accessible and useful. This effort has constituted the Bureau's recognition and response to the vital need for information on the microlevel: Census tracts, ED's (enumeration districts), and blocks. We salute this effort and deeply hope that it will not be sacrificed or diluted as a result of the processing travail of the 1970 census. One might argue that the 1970 census difficulties are ad hoc, hence primarily internal problems of the Census Bureau. But the general problem of timeliness of Federal statistics for small areas is a continuing and major issue to which the President's Commission is urged to give considerable attention. The replies in themselves did not, of course, constitute a coherent inquiry into the issue of timeliness, but they most emphatically raised it. An incomplete inquiry of regular Federal series for SMSA's (standard metropolitan statistical areas) or below, made for the purpose of this paper, indicated that the gap between the period to which the data refer and publication ranged from a few months to 3 years. The issue is real and important and merits investigation by the Commission. With respect to the second dimension of timeliness, the infrequency of some important Federal statistical surveys or other data programs, the plea for a quinquennial census is a special case, albeit the largest one, of this dimension. Numerous replies pleaded for Federal help in obtaining small-area demographic, social and economic information not only quinquennially, but even annually. One method of producing more fre- quent data that was mentioned in several replies is a systematic Federal grant program to State and local governments for the express purpose of enabling them to carry out their own periodic "little census" or similar community survey, tailored to the community's particular needs and dynamics. This proposal is not as far-fetched as it may seem at first. There are numerous precedents for it. There are many locally designed and controlled surveys going on right now with Federal money all over the country, for example in urban renewal, (continual ad hoc housing and land use surveys), education (e.g. the yearly surveys to determine the allocation of Federal impacted area funds), transportation (O&D and other types of surveys), health (e.g. community health needs surveys); the local data bank .movement has had Federal grant input, and the Federal Urban Information Systems Interagency Com- mittee (US AC) program is a large-scale Federally financed but locally controlled demonstration effort in several cities to improve their information systems; the Federal so-called 701 planning grants help support regional and information systems in many places. Most, but not all, of these examples refer to Federal grants that were funneled through Federal agencies with specific subject interest to local bodies with corresponding specific subject interest. What is new in the present proposal is the explicit recognition of the need for timely multipurpose statistics, and the systematic provision of Federal financial support for local programs to fill this need. This support could perhaps be buttressed by roving teams of Federal technicians moving around the country to provide professional guidance and as- sistance to the local efforts. Such teams might be drawn from the professional corps of the Census Bureau, Bureau of Labor Statistics and other Federal agencies. A very specific, valuable, and easily implemented improvement in the Federal statistical program with respect to timeliness of small-area data is the expansion both of the publication program and the basic sample of the Current Population Survey (CPS). This has been widely recommended. In 1968 for the first time annual yearly average estimates of unemployment, by color, for particular central cities were published from the CPS. These were restricted to the central cities in the twenty largest SMSA's. Subsequently a few additional yearly average labor force items for the largest States and SMSA's began to be published. This fairly new pro- gram of exploiting the CPS down to the central city level was a significant breakthrough. It was warmly endorsed and supported in our replies. It also raised the following question. There are numerous other characteristics in the CPS gathered with the same frequency as items already published and having equally good or better statistical reliability. The expansion of the CPS pub- lication program to these items would be most welcome to the communities affected. But, of course, such publication expansion would be restricted to the largest local areas, given the present CPS size. The replies urged the substantial expansion of the CPS sample in order to permit such publication for many more areas. There is certainly nothing un- precedented in this proposal. The CPS has been sig- nificantly expanded twice in the past 20 years- -from 21,000 households to 35,000 to 50,000. If, say, this sample size were doubled to 100,000 and the sample appropriately designed, it may enable the publication of badly needed benchmark figures for the central cities that blanket most of the urban problems and Federal urban programs. This proposal is serious and important given today's circumstances, even if a middecade census materializes. If a middecade census regrettably does not materialize, then this proposal takes on great urgency. It is in no sense an adequate substitute for a census. It could not update below- city estimates, a matter of great im- portance; and its subject matter scope would remain very limited. But it would provide timely and continuing estimates on a city level of some of the most important characteristics that guide Federal small-area programs. This is especially important with respect to the allocation of Federal funds between cities or metropolitan areas which is a continuing source of conflict and disharmony once data become obsolete. ADEQUATE ONGOING HOUSING DATA One of the major ongoing efforts in most cities in our country is the improvement of housing. Public housing, urban renewal, and various forms of subsidized housing are vast Federal programs related directly to this problem. Many other Federal programs are vitally affected by it. For example, most public works programs such as transit, highways, school construction, public buildings and others that use Federal funds are required by law or regulation to find suitable relocation for dis- placed households. In order to qualify for Federal funds most local programs that have an interface with housing must supply housing quality, housing need, relocatability, and other housing data. IMPROVEMENTS FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS Yet aside from the snapshot decennial census of population and housing there is almost no program, Federal, State or local, which provides in a systematic, coherent way even reasonably adequate information about our housing stock. This was pointed out in reply after reply. A primary reason for this paucity of information is the absence of a generally accepted model of housing need that can be carried out with reasonable accuracy and dispatch. Even the Census Bureau has had great trouble with this subject. The history of "housing con- dition" in the census is well known and need not be repeated here. The replacement of the unstable "housing condition" items of the 1950 and 1960 censuses with more elaborate "plumbing facilities" items in the 1970 census has not made a significant inroad on the problem. The relationship of housing quality, as determined by local specialists, to plumbing facilities is at best a gross one. It is not refined enough to permit plumbing facilities to be a substitute to delineate various levels of housing conditions or to indicate housing condition in very small areas such as blocks. This remark holds throughout the housing stock, including single-unit structures. Another aspect of the problem is the focus of present census statistics on housing units rather than structures. This of course is deeply embedded in census procedures and serves well for population and most housing items. But it serves poorly for determining housing quality in multiunit structures. Another closely related aspect is the disregard of environmental factors in present census procedures, e.g., trash littering the property, location in wash gullies or river banks that flood with every rain, etc. The replies indicate that what is clearly required is, first, a sustained, thorough effort to arrive at a rea- sonably useful standardized model of housing stock, especially housing quality. A start of this was made some years ago by the American Public Health Association. It is the responsibility of such Federal agencies as the Office of Management and Budget, Department of Housing and Urban Development, Department of Health, Education, and Welfare, and the Census Bureau to provide the financial and professional leadership to develop an acceptable conceptual and methodological framework for housing information. This work should provide differential weighting of various elements to reflect different climatic and other circumstances, and in other ways produce a standardized conceptual survey and estimation model that will yield housing statistics comparable from one geographic area to another and usable down to the microlevel such as block. What is required, secondly, is a regular, periodic national housing survey that will, in fact, provide the required information, at least down to the city level in the early years. There is nothing visionary about these proposals. The President has issued two reports on National Housing Goals, in both of which he called for improvements in current statistics on housing. A Cabinet Committee on Construction, through a Subcommittee on Construction Statistics, issued a report in 1970 calling for an annual housing survey. Both the Census Bureau and the Depart- ment of Housing and Urban Development have requested funds to initiate such a survey. One might initially consider such a survey as part of an expanded Current Population Survey. On the other hand, the CPS is so heavily labor-force oriented in statistical design, field training, field procedures, and sponsorship that an entirely different survey, oriented in all aspects-- sponsorship, design, field work, etc.-- toward housing, may be more practical. What is being urged here is first that housing needs are of such magnitude that they deserve a statistical program parallel to the national labor force statistical program of which the CPS is a part; and second, that in constructing such a program, the solution to the conceptual and methodological problems involved in determining housing quality should play a prominent role. IMPROVED SMALL-AREA STATISTICS OF CRIME AND THE CRIMINAL JUSTICE SYSTEM The tremendous increase in citizen concern over crime and the criminal justice process is so well known that no words of hortatory introduction are necessary here. The dismal condition of public statistical in- formation on this subject as a data base for effective modern research, planning, and programing has been thoroughly documented in the recent reports of the President's Commission on Law Enforcement and Ad- ministration of Justice. Parallel documentation in detail for a single city, fairly typical of cities over the country, is found in the reports of the President's Commission on Crime in the District of Columbia. This issue came up in letter after letter. The subject of small-area statistics on crime and criminal justice is clearly one of great concern to which the Commission is urged to turn its attention. This is a matter of direct Federal interest. First, there are, of course, Federal crimes and a Federal criminal justice system. At the very minimum the Federal statistical system has a responsiblity at this level. Second, enormous amounts of Federal money are being spent on crime control in its totality, at every jurisdictional level, and as with other vast Federal grant-in-aid programs, it is imperative that the Federal Government have effective statistical information if it is to have any understanding of or control over what is going on. In- deed, with the use of computer technology the Federal Government has vastily increased its interchange of information with State and local authorities. The National Crime Information Center (NCIC) of the FBI is a computerized clearinghouse for crimes outstanding. Its current files include information on stolen vehicles, license plates, guns, boats, securities and articles, as well as wanted persons. This is an operating intelligence file, not a statistical file. An estimated 4,000 law enforcement agencies are able to enter their own records into NCIC and make inquiries directly from terminals on their own premises. The inquiries relate to individual records and are for enforcement, not statistics. In addition to this existing system, the NCIC is now developing cooperatively with State and local law enforcement agencies a Computerized Criminal History file, each record of which will consist of the recent criminal history of every individual the cooperating agencies choose to enter into the file. The file will be used primarily for direct law enforcement intelligence, not statistics. Third, the only reasonably comprehensive statistical series now available on criminal offenses, at all levels, STRENGTHENING THE FEDERAL ROLE is a Federal program, the Uniform Crime Reports (UCR) of the FBI. Thus, although law enforcement in the United States is highly decentralized and more State and local than Federal, Federal and non-Federal activities are inextricably intertwined. Federal sta- tistics covering non- Federal jurisdictions have long precedent, and the entire field is one not only of legitimate but, we urge, of high priority concern to the Commission. As stated above, the UCR has long been the principal Federal source of crime statistics. It provides statistics on crimes and arrests by age and also by color (not both together). It also includes a very limited effort to provide disposition statistics. It has established a manual of definitions in order to make reporting as uniform as possible. The UCR has erected a substantial edifice over the years which merits recognition and admiration. Law enforcement in the United States is primarily a State and local operation. There are thousands of separate and independent police agencies. They are not required to report to the Federal Government, although there are strong inducements to do so. Furthermore, there are inumerable local differences in legal codes, definition of criminal behavior and arrest, offense classifications and terminology. In this jungle the UCR has managed to erect a national crime reporting system, over the years patiently and persistently obtaining the cooperation of thousands of local law enforcement agencies to report in a fairly uniform way. (Of course its work has been considerably facilitated by the hierarchal assistance of State agencies.) Thus, it in no way detracts from the accomplishment of the UCR to attempt here to summarize briefly the pressing needs expressed in the replies to our survey (and also expressed much more professionally and thoroughly in the pro- fessional literature and in the President's Crime Com- missions). Furthermore, it is clearly understood that the UCR deals almost solely with offense and arrest data from police agencies. It does not deal with the courts or with correctional institutions. The present UCR is very limited in its offense classifi- cation, and hence does not lend itself to in-depth studies which are critical for sophisticated planning and budgeting. A single example must suffice: the "assault" codes do not distinguish between a stranger-to-stranger confrontation and a marital fight- -entirely different kinds of assaults that require entirely different kinds and magnitudes of police action. There are dozens of such examples. There appears to be relatively little offense- specific police planning. What appears to be needed is a substantially more sophisticated offense classification scheme than is presently available. Present UCR statistics contain very little cross- classification of characteristics of arrestees, for example age, sex, race, economic status, education. Rates, in addition to incidence, play a major role in measuring trends and comparing communities. UCR does not in general attempt to define populations-at-risk in calculating rates. Rather, it uses total population as a general denominator for all statistics. It has experimented with other bases. In the narrative part of its reports it presents for the entire country auto thefts as a proportion of registered autos, and rapes as a proportion of the female population. These are not carried into the regular tabulations or extended to other offenses. In fact, this is a field of professional controversy--just what is the best way to construct valid and comparable statistical measures. It is, of course, difficult to obtain valid denominators of subgroups of the population in many communities, or other objects or quantities that would seem to be more appropriate denominators. A possible substitute might be intercity comparable statistics for subareas of the city as comparable ghetto areas, slum areas, business districts. What seems to be needed is a sophisticated analysis of ways of getting at populations- at-risk and of obtaining thereby better rates and other derived statistics for more effective planning. The police blotter is, of course, only the start of the criminal justice process. Let us turn from the police agencies and the UCR to the courts and corrections. Here we are almost in prehistory. There is not a single Federal series on persons going through the courts that includes the non- Federal judiciary and is down to any local area level. Yet everyone recognizes that logjams in the judicial process are a major factor in crime today. What are the trends on lengths of time and delays in the various stages of the judicial process? on plea bargainings? on costs? employment? What are the statistics on the various ways persons enter and leave the system? Disposition statistics are abysmally deficient. These are merely examples of significant questions. Dozens more could be asked, all of them im- portant for improved crime control and criminal justice planning. There is today scarcely the semblance of a compre- hensive program of statistics on the judicial process. Even integrated Federal source data on employment and costs of the judicial system in the census of governments, the only recurrent program that attempts this, tend to be buried in "central administration," with a minimum of data separately for States and a few very large cities. Both the National Crime Commission and the District of Columbia Crime Commission were frustrated by the inability to match statistics and account for people from one step in the total process to the next--from arrest to arraignment to prosecution to disposition to appeal to corrections. At each step thousands of people simply disappeared--poof. The D.C. Commission was so struck by this that one of its principal recommendations was the creation of an Office of Criminal Statistics for the city of Washington, D.C. This recommendation was accepted by the Congress and such an office established. One of its principal activities has been the construc- tion of a mechanized recordkeeping system that will keep track of an individual from the point of entry into the process to the point of exit, something which is not now possible either in the District of Columbia or, as a broad generalization, anywhere else in the country. The situation is only slightly better in corrections statistics. The decennial census exhibits persons in institutions by age, sex, and color for counties. But institutions here include not only correctional but also mental and chronic disease hospitals, homes for the aged, the handicapped, and dependent and neglected children. This is hardly a data base for any correctional planning. The quinquennial census of governments has an extremely limited amount of information on em- ployment and expenditures for correctional institutions- - and that is all there is, on a recurrent basis. Statistics of the Federal Bureau of Prisons are restricted to IMPROVEMENTS FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS Federal institutions. There are no Federal series at all on probation and parole. Yet data on the correctional process are critical for meaningful planning in the area of criminal rehabilitation. What are the statistical trends of recidivism'.' They don't exist. Who are the recidivists? Who knows? No meaningful analysis of rehabilitation programs or comparisons of programs are possible without meaningful data on recidivism. There have been many special studies, but no regular statistics. The forthcoming Computerized Criminal History file may wall up this gap considerably. What are the trends of kinds of people reaching the end of the line--prison? The trends in length of stay? No re- curring data exist, although there have been several special studies. Yet such information is basic for efficient correctional planning. What are the living con- ditions in prisons and jails? the food? the sanitary facilities? the crowding? the rehabilitative programs? What is the offense distribution of the prison population? There are no regular data on these or dozens of other important questions. A special jail survey recently funded by the Law Enforcement Assistance Administration revealed, among other things, the astonishing estimate that 50 percent of the jail population at the time of the survey were not being held for conviction of a crime. They were under preventive detention! The relatively new Federal Law Enforcement As- sistance Administration is well aware of all of these matters. Its statistical unit is attempting to generate useful information in many of these areas. It is funding several research programs and pilot studies. In addition, it is supporting the design of a large-scale national household sample (about 60,000 units, plus an additional 10,000 in the central cities of the 10 largest SMSA's). This sample will concentrate on characteristics and histories of victims of crime. An extension will cover a sample of businesses and other nonresidential es- tablishments. The results of this survey will be national, plus a few of the largest States, SMSA's, and cities, but nothing for local areas. The effort that may prove to be the most valuable in the long run for small-area data is the encouragement and assistance to States to establish central State agencies to develop criminal justice information. An adequate statistical program on crime and the criminal justice process is clearly one of the most im- portant fields for attention by the Commission, not only for small-area statistics but for State, regional, and national statistics. The Commission is strongly urged to explore the initiation of a "crime registration area" system analogous to the present highly successful birth and death registration area systems. The situations are structurally similar, and the hoped-for emergence of State information agencies being encouraged by LEAA may make this entirely feasible. The Commission is also urged to make a serious investigation into the pos- sible establishment of a large-scale, recurrent, master sample of police agencies, courts, and correctional in- stitutions, or any part of these in order to develop a meaningful criminal and criminal justice statistics program on a small-area basis. What is visualized is a program similar to the Current Employment Statistics Program of the BLS. The area of effective law en- forcement and criminal justice is of great importance to the entire Nation. The moral, social, and economic stakes are very high. The Nation deserves a modern, sophisticated, meaningful, Federal statistical program. IMPROVED LABOR FORCE STATISTICS The emergence of large new local programs in recent years has revealed serious information gaps in small-area manpower statistics. Two major pro- grams are particularly handicapped by the present situation, as pointed out in the replies. Manpower training is a large-scale and growing activity. It involves substantial Federal funds. It needs small-area data, for in general it is carried out locally and is oriented toward separate metroplitan areas or labor market areas. Clearly a critical component in planning a useful manpower training program is detailed knowledge of the local labor market--in particular, occupational trends of both the employed and unemployed. Unfortunately, there is today no ongoing, systematic, statistical information on employment by occupation for small areas. The only detailed information available is the decennial census. As valuable as the census is, its infrequency makes it of minimal value for developing trend data and analyses upon which a successful daily ongoing operating program must by based. In the absence of regular data series, numerous sporadic "area skills surveys" have been undertaken by local com- munities. A methodology to assist such surveys has been developed by the Department of Labor. But these surveys are locally expensive, depend on very chancey local, Federal or foundation funding, are ad hoc and very irregular. At best they are undertaken every several years. There have also been experimental efforts to exploit other records, such as files of the Equal Em- ployment Opportunity Commission. These various efforts reflect the serious need for such data and the thrashing about of the professional community to try to fill the gap. The situation is deeply unsatisfactory and an almost classic example of a clearly perceived and defined informational gap which the Federal statistical system should fill. We urgently recommend to the Commission the development of a regular work-force occupational series for local areas. One possibility is deeper use of the Current Employment Statistics Program of the Bureau of Labor Statistics. This program provides annual in- dustry data for over 200 labor market areas. The Bureau is well aware of the need for occupational distributions and is attempting to develop such a program. We urge upon the Commission the support of this effort, as well as a thorough study of the problem and possible solutions over the whole range of Federal statistical programs in order to assure an optimum development. Burgeoning programs spawned by the urban revolution, with its increasing concentration of daytime populations in various loci of a metropolitan area, have created a large-scale need for detailed information on where people work and their work- residence linkages. Urban renewal, so long dominated by housing programs, has increasingly included the renewal of commercial, industrial, and other work areas of the city. Renewal of the city as well as the increasing development of suburban employment centers require intricate planning and substantial public work commitments-- water demand, sewage, solid waste, civil defense shelters, economic planning, and the behemoth, transportation planning. These programs involve enormous Federal funds, and the Federal Government is deeply affected by the effective- ness and efficiency of this area of community planning. 8 STRENGTHENING THE FEDERAL ROLE Thus, it has a powerful stake in the adequacy of the data base on which such planning must depend. The decennial census for the first time in 1960 sought place-of-work information, and was thereby able to produce such data by county, linking the data as far down as census tract of residence. The Census Bureau and Department of Transportation are cooperating on an elaborate cross-tabulation of work- residence based on the 1970 census down to the block level as a statistical building block. This constitutes a sub- stantial breakthrough in comprehensive work- residence statistics. However, as with all decennial census data it is a once-in-10-years snapshot. The letters in our survey do not indicate whether fine-grained place-of work and work-residence tabulations decennially will be adequate to the various programs they will serve. Since these are multipurpose statistics, their adequacy will probably vary from use to use. This area must be closely monitored. If such infrequent data are inadequate, then an intercensal program should be devised. A middecade census, if it includes the appropriate questions, would of course be such a program. Otherwise, other sources must be tapped. At this point it is necessary to mention for the first time in this paper the statistical use of files of Federal operating programs. Daytime populations, work force, and work- residence linkages, are required for areas below the city level-- neighborhoods, traffic zones, and other below- city levels. It probably is not feasible to develop a periodic national sample survey so large as to permit reliable estimates at these levels. Potentially, however, such statistics may be possible by the exploitation of files of the Internal Revenue Service, the Social Security Administration, or the Unemployment Compensation Boards. For example, most individual income tax returns today are accompanied by employer W-2 wage forms. These could possibly form the basis for recurring place-of-work and work-residence tab- ulations, using, say, the Census Bureau ADMATCH program for computerized geocoding. The quarterly em- ployer social security and unemployment compensation returns are a potential source of below- city place-of- work tabulations. It is beyond the resources of this paper to discuss these possibilities in detail. We are cognizant of major stumbling blocks--the less-than-total coverage of these files; the problem of establishment reporting, especially below the county level; the large-scale new mechanization that would be required; the interface of a statistical program with the primary operating mission of an agency which of course must take precedence. Mod- ifications of present reporting forms might be con- sidered; also occasional special surveys to acquire adjustment factors. There may be alternatives, for example, the proposed Federal Industrial Directory for at least place-of-work data. The principal point intended here is to convey to the Commission the need for place- of-work and work-residence data for microareas. We urge that the Commission develop pilot projects to establish clearly the objectives these needs reflect, the time frames required, the geographic and subject- matter detail, the various ways that the data can be obtained or the objectives met, the relationship to operating and other programs, costs, and all related matters. CURRENT SMALL-AREA STATISTICS ON INCOME It seems almost unnecessary to say that of all statistical series measuring the overall success of the almost innumerable social and economic efforts going on in our cities and towns to reduce poverty and elevate the economic status of our citizens, the most basic and obvious one is income. Yet, as many replies emphatically noted, beyond the decennial census there is no statistical program, Federal, State, or local, that provides a small- area time series on person, family, or household income. There are small-area income series, for example, the County Business Patterns (Census Bureau), Employment and Earnings (Bureau of Labor Statistics), personal income for SMSA's (Office of Business Economics). Each one of these is important, needed, and used. But they are not effective measures of income trends of persons or families in cities and neighborhoods. The closest series that exists today is the 5-digit ZIP data on ad- justed gross income made available on paper copy, tape, and microfilm triennially by IRS. It is worth noting that the first IRS printing of 3-digit ZIP data on income in 1966 was exhausted in a month. It has continued to be a steady best seller. Many local data users believe that there is much still to be mined in the individual income tax files. Over and over this was mentioned in the replies—income ranges, special categories of expense such as medical, microarea tabulations such as census tracts. There are, of course, substantial limitations to IRS records- -for example, they cannot provide family and household data beyond husband-wife joint returns; there are thousands of poor persons who do not file tax returns; many large taxpayers use distant mail addresses. Nevertheless, many local program administrators and researchers feel that generally valid and useful time- series comparisons can be generated from IRS records. Some ask for county data, some for census tract. County was an item on the 1040 form for many years. It could be restored, or, as indicated above, the techniques developed by the Census Bureau for the 1970 census could be used for computerized geocoding. The Commission is urged to explore the possibility of a meaningful microarea income series. The IRS has been singled out here because it dominated the letters that discussed this subject. But, of course, the entire spectrum of Federal possibilities should be explored. IMPROVED GEOGRAPHIC IDENTIFICATION Many of the replies referred to the difficulty of obtaining data where geographic identification is not clear. The context is particularly the absence of house numbers in areas beyond the city delivery areas of the postal service. While this problem has been with us for a long time, its high visibility and increased urgency is a direct result of the computer revolution. Let us describe briefly the whole fabric of which the required improvement is a single but very significant strand. The tremendous advances in computer technology have enormously widened the horizon of data availability on a small-area basis, even on a microarea basis, both in Federal and non- Federal programs- -and indeed have very greatly strengthened the possibilities of cooperation, interchange and intertwining of Federal and non- Federal data files. The largest and most obvious IMPROVEMENTS FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS 9 example of all of these changes in the context of this paper is the geography program of the Census Bureau. It is not necessary to go into detail here on how the 1970 census was taken, of the computerized geocoding tools and procedures that were developed for the census, of the substantial strengthening of State and local data access capability by use of such Census-developed tools as ADMATCH, ACG, and DIME. These de- velopments are well known to the Commission. It is important to emphasize here that, in general, computer technology has opened information vistas from geo- graphic base data files, and improvements in file organization, control and management at all levels of government that were utterly beyond our reach only a few years ago. The development of computerized data banks and the emergence of computerized man- agement information systems are without any doubt the wave of the future. New specialties and skills have appeared, new professional societies, a new language. This technology and its consequences are developing so rapidly that it seems quite safe to predict that, with the typical absence of a sense of history of the young, the next generation will look back at those primitive days BC--Before Computers- -and wonder how anybody was able to know anything about anything. It is a sentimental archaism that appeals to a bucolic nostalgia in the American psyche, while in fact being a roadblock to efficient delivery of services and to efficient social, economic, and physical planning. Its elimination is part of the price we must pay for extension of the benefits of modern technology. The Commission is urged to explore how best to mount a systematic, national program to assign a house number and street name to every existing house at least in present SMSA's. Of course in general, the assignment of house numbers is a local responsibility. The most logical and effective leadership channel to gain local cooperation probably is the U.S. Post Office, which itself would benefit substantially from such identification. Some replies even urged a rural postal-addressing pro- cedure. The success of such a program even if re- stricted to metropolitan areas will substantially exepedite rational planning at all levels of government. NEED FOR WIDER KNOWLEDGE OF FEDERAL STATISTICS FOR SMALL AREAS But, with all technological revolutions, we pay a price. (We are paying a price in many dimensions of life, and if we were to digress into a philosophical discussion, a plausible case could be made for the proposition that the expected gains are not worth the price. The proposition, true or false, is irrelevant because the revolution is upon us, we are in the midst of it, it is inevitable. All we can hope to do is try to guide it toward its most productive potentials and try to minimize its destructive aspects.) Part of the price we must pay is the increasing need for standardization and precise definition of data elements. This need appears in many ways and at many points. In particular, we must be able to establish a unique, codeable identification of a unit record in a file. For the present purpose we narrow down to one specific needed improve- ment, one small but crucial linchpin to keep the magnificent monster moving, one single but significant strand in the panoramic tapestry. (Whew! ) In order to apply the technology to any geographic area we must have a unique identification for every house in the area. A place known only as "Rural Route 7, Hutchinson, Kansas", or "General Delivery, Garrett Park, Maryland" cannot enter the system as a unit record because this is not a unique identification. Thus it cannot be cross- tabulated with other records, separated, redirected, pawed over, manipulated. It is an outsider, an orphan. If there are many such orphans, then the system is incomplete and its usefulness and maneuverability cor- respondingly reduced. It is a fact that a great many such orphans exist inside the boundaries of numerous SMSA's. Whole large areas considerably built up have no unique house identifications. This substantially hampered the 1970census. Itcontinues to plague not only the geographic programs of the Census Bureau but other Federal programs such as the Census- BLS Current Population Survey. It is a baleful im- pediment to any sample survey design that involves such areas. It hampers the mechanization of local files and is an obstruction to the progress of computerized management information systems in suburban areas. There clearly emerged from the replies the need for a concerted effort by the Federal Government to dis- seminate more widely the knowledge of existing Federal statistics for small areas, of the characteristics and limitations of existing Federal data files, and of the nature of the Federal statistical system. This need emerged in several ways. Some of the replies showed lack of knowledge about the present availability of Federal statistics, making recommendations or requests for data that, in fact, are available. Other replies explicitly articulated the sense of being lost in a bowl of chop suey- -overwhelmed with Federal data, but in a chaotic and unordered way that actually left them wondering where to turn for specific information. These puzzlements did not pre- dominate in the replies, but they were there. There are, indeed, important Federal statistical series for small areas that are not well known even to reasonably well- informed professionals. Some examples: (1) The various component series that constitute inputs for the estimates of SMSA total personal income prepared by the Office of Business Economics (they are not pub- lished, but are available on request); (2) the 5-digit Zip, triennial, individual income statistics of IRS; (3) the Federal outlays by State, county, and large cities pre- pared by the Office of Economic Opportunity; (4) the historical Community Profile series by county prepared by the OEO. All but the first of these examples are availabile from the National Technical Information Service. It is safe to say that NTIS itself is probably unknown to large numbers of local data users. In 1^62, the Advisory Commission on Inter- governmental Relations published a Directory of Federal Statistics for Metropolitan Areas. Building on this base, the Census Bureau in 1966 published a com- prehensive Directory of Federal Statistics for Local Areas. This was a highly successful guide. Several replies urged that it be updated. None of the examples in the last paragraph are included in the guide. An update should include data available in unpublished form. 10 STRENGTHENING THE FEDERAL ROLE As valuable as this guide is, it deals with only one dimension of the dilemmas expressed in many of the letters, the dimension of "what data are where." In another dimension, some letters made suggestions that showed misunderstanding of Federal data files (for example, requests that "family income" statistics be extracted from IRS individual income tax records) or that were out of touch with reality, either now or in the foreseeable future (for example, some letters urged an annual census). A more serious fantasy appeared in some letters that urged widespread, adequate, and regular Federal funding to local communities. It would be similar to the USAC effort. It would be possible to integrate reg- ular operating records of local government with some local survey supplementation, sufficiently to do away with Federal surveys such as the census and obtain essentially the same information regularly as byproduct of integrated local operating records. Several replies urged increased Federal-local inter- change that would enable local users both to understand the Federal system better and to air their problems, dissatisfactions, and recommendations. Others urged Federally- sponsored training conferences in elementary statistical methods or in the uses of Federal statistics. All of these seemingly diverse recommendations have been brought together in the final section of this paper because they all point to a single general conclusion— that the Federal Government should establish a broad program to publicize the Federal small-area statistical program; to educate local users in how data are generated, the nature and limitations of Federal data files, and the realities of statistical life; and to develop a regular interchange program with users of small-area statistics (who of course include other Federal agencies, and large nongovernmental organizations as well as States and local users) in order to articulate changing and emerging data needs. And this brings us logically to the final point that this paper extracts and elaborates from the replies to our survey. A large step forward toward a modern Federal statistical program would be made by the establishment of a Federal Office of Small- Area Statistics. The need for a comprehensive, aggressive Federal unit whose sole purpose is the improvement of small-area statistics was implicit in many of the replies and was explicitly stated and discussed in several. Here are some of the things that such a unit would do. The unit could articulate in detail the Federal small- area statistical program to its users. This is essentially an extension of the remarks made above. As part of the activities mentioned there, the unit should develop an intensive course (or manual or handbook or textbook) that would explain inconsiderable detail every federally gene/ated series on small-area data, including its methodological generation, its limitations, and examples of its use; and that would explain in considerable detail the characteristics of all Federal operating programs that gather microdata. ("Considerable detail" would include, among other things, what can and cannot be constructed from the file; the present state of mechanization of geocoding, of pub- lication; confidentiality restrictions; the relationship of statistical output to operating program, and a wide range of other points essential to understanding and use of the records.) The unit could also carry out a program of pro- fessional research in small-area data problems and develop packages for use by local agencies. Some examples of this are: (1) Develop the meth- odology for effective social indicators on the local level. There is a large literature on "social indicators," very little of which is of practical use to local officials seeking an effective monitoring system of the condition of their community by means of social indicators. Most of the literature is philosophical, or national in orientation, or suggesting new variables which are nowhere collected. There has been little input by pro- fessional statisticians, and even less by those working in a real, live, local context. Hence, there has been little application of statistical science to the construction of composite indicators and delineation of trends in the quality of life in various functional areas using local files and data records actually available. A critical element here is to develop models enabling administrators to use the indicators to help tell them what they should do. If the social indicator movement fails to provide guidance to local administrators, it is not likely to receive their support. Only now is input from professional statisticians beginning tentatively to emerge. (2) Develop practical packages for rapid and inexpensive con- structions of statistical tabulations and statistical ma- nipulations from computerized, unit-record, operating files, packages operable by local data users who are not trained computer programers. There has been con- siderable activity in this field, but reasonably satifactory performance has not yet been reached. (3) Develop a model of housing quality that is workable and widely acceptable. This was discussed above. Very closely related to the point made previously for a program of professional research the unit could develop standards and aggressively press for their adoption by local government units. Standards elevate the accuracy and usefulness of data and make data comparable over time and from area to area. One of the great virtues of current Federal programs that provide small-area data is their standardization, so that words and statistics have the same meaning everywhere. Examples are the programs of the Census Bureau and labor force, price index and other programs of BLS. A prime example of work that should be undertaken by the proposed Federal unit is a standard land, structure, and establishment code that is practical for use by local operating programs such as assessment and licensing. Another example is an im- proved criminal offense code and a national crime registration area as discussed above. A third example is a standard code for housing and structure quality if a practical model of housing quality, as discussed above, can be developed. A fourth example is the development of standard confidentiality and protection-of-privacy pro- cedures as local files become increasingly mechanized and increasingly integrated- -two aspects of the computer revolution that are historically inevitable and that pose serious issues as mentioned earlier. A fifth example is a greatly stepped-up program to complete the national marriage and divorce registration areas that have been progressing slowly for many years. IMPROVEMENTS FEDERAL STATISTICAL PROGRAMS FOR SMALL AREAS 11 Develop an appropriate funding program to stimulate improvement of small-area statistics, and become a focal point of encouragement and knowledge about small-area statistics programs. This may be considered in two general dimensions- - local programs and Federal programs. Federal grants- in-aid contingent on local improvements and conformance to Federal standards have long been in use. What is recommended here is that a Federal Office of Small-Area Statistics develop such a program in its subject area. Some Federal grant funds to localities for information activities have of course been available for many years for ad hoc demonstration or research projects (one of the largest is the current USAC effort); and indeed in a very limited way there have been small, semicontinuous Federal funds available for information- gathering activities as part of the funding of larger pro- grams such as continuous planning programs. What is visualized here is different from either of these and also different from the classical continuing grants-in-aid for substantive programs. There are numerous improve- ments in local programs where "start-up" or "conversion" costs are high but, once started or constructed, are not unreasonably expensive to maintain. Examples: me- chanization of important management or planning data elements that are collected but not mechanized because they are out-ranked in priority claim on scarce local funds for the primary operating mission of collecting agency; improved coding schemes whose use would re- quire conversion of collection and data-processing tools. In the past, Federal agencies have been reluctant to fund such terminal start-up or conversion costs because they "don't demonstrate anything new" and were con- sidered costs associated with regular operations that are a part of a local community's own responsibility. This view should be subjected to a searching reappraisal with respect to the ability of local governments to initiate improvements and the interest of the Federal Government in improvement. Such a reappraisal is particularly appropriate with respect to use of Federal funding leverage for adoption of standardized models and schemata and standards of local practice. Federal level, the proposed office would become in the improvement of Federal small-area actively assisting other agencies in obtaining would seek to develop fully those recom- s in this paper and in the many letters we that it would consider feasible and worth It would exert initiative in monitoring Federal programs, perceive gaps and trends, and new data needs. It would make policy rec- ions to agencies, to the President, and to the On the involved programs funds. It mendation received pursuing, and local anticipate ommendat Congress. One special aspect of this was pointed out in a few replies and is worth capturing in this summary. Most Federal grant programs require some data submissions by local bodies as part of their applications for funds. These data requirements sometimes are so intensive or extensive or both that they place almost intolerable burdens of application preparation on the local bodies. Sometimes the justification data are literally impossible to acquire. Indeed their acquisition is sometimes part of what the grant itself would do. Under these circumstances only very sophisticated local staff can hope to put to- gether the data required to make successful applications, or the data are simply manufactured; most often, many local bodies simply don't apply because they haven't the knowledge or manpower to spend on lengthy applications that have little chance of success. The proposed Federal unit could monitor Federal grant programs, determining cooperatively with the grant agency the reasonableness of application requirements, what can and cannot realistically be expected, what data are and are not available. Conversely, the proposed Federal unit could provide advice and guidance to local bodies upon request, suggesting where information is available or how it might be developed. Thus the proposed unit, if it acquires the extensive know-how and knowledge of all small-area source data, Federal, local, and non- governmental, that is expected of it, could perform a very useful middleman role between grantors and grantees, improving the performance of both. Develop teams of professional statisticians, systems analysts, classification specialists, and whatever other skills are relevant whose primary purpose is to be available to local communities for professional help in statistical and data problems. One of the most serious weaknesses of local government is the scarcity of pro- fessional know-how in statistics and related fields and the consequent inability of most local governments to take advantage of modern professional techniques. Adequate technical knowledge is today restricted to a very few of the largest cities and States. Many bits and pieces of the work of the proposed unit are being carried out now by various Federal agencies. Some examples: the Federal interagency USAC program is trying to develop on a demonstration basis modern, local, total management information systems, not oriented toward statistics but primarily toward the most effective modes of mechanization of local government operations. Statistics would be a byproduct of this mechanization; there have been numerous Federal grant projects sponsored by HUD, HEW, and DOT dealing with various aspects of matters mentioned here such as land-use codes, work-residence data, data banks, health surveys, and many other activities; in almost every recommendation made in this report, comment is made of some work already being done. Probably the most comprehensive efforts have been and are being made by the Census Bureau- - historically, its long standing census tract program and ever- broadening small-area statistics output, its inter- governmental seminars on Federal statistics, its continuing efforts to develop and improve geocoding techniques, the numerous activities stimulated and guided by the Census Bureau with other Federal agencies as well as State and local sponsorship embodied in the New Haven Census Use Study and in the Southern Cal- ifornia Regional Information Study, its efforts to in- tegrate local and State population estimates into a national network. Hence, is the proposed new Federal unit necessary? What could it contribute that is not being done now in a decentralized way? In our view the creation of a Federal Office of Small- Area Statistics would be of substantial benefit to the Nation if it is successful. There are indeed sub- stantial advantages in program decentralization, some of them being that it spreads the work and the budgets and reduces bureaucratic impediments to creative activity. In our view, the proposed Federal Office of 12 STRENGTHENING THE FEDERAL ROLE Small- Area Statistics would not diminish the initiative of other Federal programs in the same field, but rather would stimulate them by perceiving their relationship to other agency programs and encouraging symbiotic and integrative arrangements. The principal characteristics of most current Federal agency, small-area programs is that they are conceived and perceived primarily as assistance to the primary mission of the Federal agency, which is a Federal mission. While most of the things discussed in this paper have been perceived by Federal agencies, the priority and support they are given is proportional to their usefulness to the primary Federal interests of the agency, and these are by no means necessarily the same as the primary interests of local users. The primary mission of a Federal Office of Small- Area Statistics would be to represent the interests of small areas- -the metropolitan regional bodies, the counties, the cities, the neighborhood associations-- unconstrained by other priorities or loyalties. Such a mission does not exist today in the Federal Government. The entire raison d'etre of this office would be to serve the needs of small areas. It would, if successful, become their spokesman, their advocate. It may be noted that many of the activities proposed for this office here do not involve Federal statistics at all, but rather the improvement of local statistical capability. The office would require statistical, computer system, and planning expertise; intimate familiarity with the workings of local government including its point of view, its atmosphere, and context; and comprehensive knowledge of the Federal statistical system. Are small area statistical needs that important? Do they require such special recognition in the Federal statistical system? The answer is yes, and the rationale of the answer is concisely stated in the introductory paragraphs of this paper. Such a special recognition, if successful, would constitute a breakthrough in the ability of American towns and cities to control and guide their destinies. INSTITUTIONAL ASPECTS OF SUBNATIONAL DATA COLLECTION AND USAGE Wilbur A. Steger, President CONSAD Research Corporation INTRODUCTION Some people have vivid imaginations. I once imagined a full-fledged undertaking of the study and analysis of our national production and use of subnational statistics. All existing and potential users and suppliers of small-area data- -public and private, national and local level, perodic and ad hoc- -were included within the scope of this imaginary study. It took several years to perform, as you might expect, the exerting of a lot of expert effort, and the involvement of many people and organizations. The outline of this study I imagined to be comprehensive in describing current problems and exhaustive in analyz- ing alternative solutions. As such, these opinions are strictly mine and not the Commission's. Neither I nor you can afford the luxury of such an ambitious and thorough undertaking. While this subject, in our combined opinions, is complex and important enough to justify something akin to a "Royal Commission Inquiry," it does not have the glamorous or the political sensitivity to warrant the expenditure of, say, 30 or more man-years of effort and extensive survey and involve- ment of all concerned parties. Certainly, this was not-- could not- -have been the Commission's purpose. I am afraid that what we are reduced to, instead, is a con- siderably more limited set of findings and conclusions, based more on approximations, conjectures, and as- sertion. Absent the needed surveys, we can only best guess at certain key magnitudes; missing the required analyses, we must hazard at preferred solutions. Left to our own devices this way, I shall restrict my observations to those of an institutional, not a narrowly "technical" nature. I shall discuss the nature of the guidance required by our subnational data efforts of the 70' s and criteria by which these guidance efforts might be evaluated and existing institutions appropriately modified. This is not to say that no one has thought or written about these matters before. Indeed, a number of interesting recent papers about the introduction of computer and associated technology have briefly con- sidered the roles which institutions play in the processes of innovation, invention, and diffusion 1 in subnational J W. L. Garrison, "Computer Technology, Urban Processor and Problems, New Futures for Urban Areas," presented at The Association for Computing Machinery, N.Y. , Sept. 1970, pp. 18-21. area management, decisionmaking, and information sys- tems. Nowhere, however, has there been a thorough documentation and analysis of the various public and private institutions which have already played a role in the development of the production and use of sub- national statistics. Nor has there been a discussion of the meaning which various changing social and political contexts (e.g., decentralization, program evaluation, increasing nondefense expenditures, subnational com- petition, etc.) have for the existing institutions, in- stitutional arrangements, and the possible need for new institutions and arrangements. As we shall see below, the President's Commission on Federal Statistics, which has commissioned several individual efforts primarily to review the state of the art and to inform the general public on possible future directions, has "started the ball rolling." Given this impetus, several directions for future study now exist. MAGNITUDE OF THE PROBLEM The magnitude of the subnational statistics problem cannot be restricted to dollars and resources, alone. But the resources involved are by no means insignificant. While no breakdown of the $336 million annual Federal statistics budget (this was for fiscal 1969) exists for purely national as opposed to subnational reporting, the latter is undoubtedly a substantial proportion. In ad- dition, there is a very significant amount of nonfederally produced statistics for subnational areas which, by conservative estimate, must be provided at a significant fraction of $1 billion during each decade. (See the 1,800 statistical publications reported by the U.S. Bureau of the Census, Directory of Nonfederal Statistics for States and Local Areas: 1969.) Fifty to $100 million annually would not be an overly liberal estimate for the primarily subnational component of the Federal statistics system. Undoubtedly, tens of thousands of individuals are occupied in the production and hundreds of thousands of individuals in the use of subnational statistics, both at Federal and at subnational levels. Other statistics are, perhaps, even more indicative of the magnitude of the resources involved and required. At the State level, for example, a recent report of the National Association for State Information Systems re- ports that the States are spending more than $300 million annually for the development and operation of management information systems for government de- cisionmakers. A similar total for all U.S. SMSA's would entail only a little more than $1 million annually per/SMSA--not a difficult amount to reach if all items appropriate to data collection and usage of small-area 13 14 STRENGTHENING THE FEDERAL ROLE data were included. (Cities like New York, alone, undoubtedly spend tens of millions of dollars on such programs. Rarely, however, are they considered and itemized separately.) Nor is it only major governmental units that collect and use small-area data: of the 81,248 units of government in 1968 (approximately 18,000 municipalities, 17,100 townships, 3,050 counties, 21,800 school districts, and 21,300 other special districts) data collection and use clearly must average several thousands of dollars apiece, annually. 2 Properly calculated, a total closer to three-quarters of a billion dollars annually for subnational data collection and usage--for all levels of U.S. Government, and excluding the private sector's production and use--would not be surprising. One might speculate that the trends in more efficient computer technology, greater effectiveness in data re- duction and sampling procedures, and more efficient intergovernmental coordination will serve to decrease what otherwise is rapidly approaching a $1 billion annual expenditure. These trends are, fortunately, with us; however, I believe they will more likely than not act, primarily, to (a) increase the effectiveness with which rising budget levels for subnational statistics are utilized, and (b) substantially reduce what otherwise would be more rapidly rising total costs of using and producing subnational statistics. So much for the dollar magnitudes. Exaggerated? Perhaps; but I feel, in the right ballpark. This is one reason I feel the state of small-area statistics is in such a critical turmoil: not so much the absolute magnitudes, but more so that we do not know what they are, and have no method for using all the resources in a coordinated fashion. Given these resources, what about new constraints, trends in political philosophy, and the desire for cost- consciousness? SOME TRENDS, CONSEQUENCES, AND INSTITUTIONAL REQUIREMENTS Trends in our "new" and/or "creative federalism" will influence the costs of delivering and using sub- national statistics; however, they will affect other fundamental dimensions of subnational statistics per- haps even more radically. Four examples of such trends are presented here. Trend 1 Decisions about the use and incidence of public sector resources are being made, or at least participated in, at lower public sector levels than in recent decades. This will be no less true and, in fact, will be somewhat accelerated if revenue- sharing becomes a reality in a significant way. created metropolitan planning agencies (in the early 60' s), regional economic development agencies (in the late 60' s), and recent developments in decentralizing HUD and HEW functions to district offices. Further- more, these agencies--along with the traditional city and State organizations- -are providing types of services not formerly supplied by public agencies, such as family planning, Federal grantsmanship, and manpower training assistance. In addition to both of these factors, there is a substantial increase- -in dollars and services provided- -in the traditional urban public services, such as welfare, public health, education, and public housing. A final factor, in this regard, is the increasing use of information in performing these public sector roles: increased training of the subnational bureaucracy has altered the mix of qualitative and quantitative, favoring the latter. Consequence Thus, it is not surprising to find the requirements for economic, demographic, social and physical sta- tistics at subnational levels to have increased: policy makers, planners, managers, and operators of these public programs need statistics for and about their relevant geographical areas to be able to satisfy their respective public roles. Institutional Requirements There are several possible alternative roles for the Federal Government, given this context: (1) Information Roles Planning in the aggregate to coordinate the pro- vision of subnational data, identifying gaps where the private sector would not be able or willing to provide these statistics. Providing of methodology, e.g., the translation of sophisticated statistical and decision models into practical terms for widespread public and private usage. (2) Regulatory Roles Public utility- like regulation of rates and con- ditions of service where ownership is private, but competition is not feasible. Transfer agent, where third parties receive social benefits or bear social costs. (3) Economic and Financial Roles Discussion There is an increasing number of subnational in- stitutions providing public services, e.g., newly Other large consumers of data collection and usage resources include regional development commissions, ad- ministrative regions, Model Cities and community action programs, special impact programs, special environmental control regions, etc. Public investment, where public goods "external- ities" are present. Subsidies, where marginal costs are substantially below average costs in the provision of these statistics. I have examined, elsewhere, the model of the Federal Government providing guidelines and local grants, within INSTITUTIONAL ASPECTS OF SUBNATIONAL DATA 15 the context of substantial local option, 3 and I have found it substantially attractive. The pressure, of course, is put on Federal Government guidelines with regard to data quality, giving local option as to scale and scope of local collection and processing efforts. Trend 2 Discussion The situation now is very much out of control. So much is going on in so many directions that one can only feel that the situation is chaotic, burgeoning in several uncoordinated directions, and duplicative and wasteful. There will be increasing concern, at subnational levels, with noneconomic "quality of life" measuring data. Discussion Subnational levels of government have not, for the most part, attempted to measure the gross or net economic well-being of their area (e.g., the "gross regional product"). But given the increased attention being paid to the social environmental factors which make the quality of life differ from one area to the next, subnational areas are increasing their efforts in social and environmental planning as well as in physical planning. (Economic planning and develop- ment, while an important subnational function, is felt to be so subject to external pressures and economies so open, that effort to measure a real economic product has not increased.) Consequence Local and regional agencies are finding it increasingly necessary to define and measure how well they are meeting the "softer," quality-of-life needs of their residents. Measures of the results of a real en- vironmental and social public sector efforts are crucially needed. Institutional Requirement(s) Implications for Federal action and roles seem fairly obvious with regard to this trend: (1) A substantial amount of Federal research, de- velopment, and demonstration in the total and areawide measurement of quality-of-life in- dicators is desirable, with a gooddeal of emphasis on local experimentation within the context of substantial Federal overview and guiding effort. (2) Federal agencies involved in promoting or in- fluencing quality of life and its indicators should develop programs for local option in data ex- perimentation, measurement, and processing of quality-of-life indicators most important to each subnational area, again within the context of strong Federal guidelines. Trend 3 While more resources, both relatively and absolutely, will be devoted to subnational statistics production and use, there will be an equivalently strong effort to rationalize and make more effective that which is being done. 3 W. A. Steger, "inventing Economic Institutions for Urban Survival," prepared for Carnegie Mellon University Conference on Advanced Urban Transportation Systems, Pittsburgh, Pa., May 25-27, 1970. Consequence While it is possible, even likely, that substantial net benefit continues to exist for most subnational sta- tistics programs, it is time for the exercise of strong budgetary controls and to place the entire subnational system onto a long-range, well-planned programmatic basis. Institutional Requirements Since it is governmentwide, interagency planning, pro- graming, and budgeting functions that are involved, a decision is crucially needed about the role of existing, or perhaps newly created, planning and operational agencies in this substantive area. This is no longer merely a staff or advisory function and, while extreme care is required (given the trends already discussed above) to avoid central control, centralized management- -planning, programing, and budgeting- -is crucially needed. Trend 4 Continuing into the 70' s will be the schizoid political reality of (a) the need and desire for local-level citizen participation and involvement and (b) fear of confidentiality invasion of privacy. Discussion Data possibly leading to "individual dossiers" is one issue on which the political right and the political left coalesce. During the 70 s s is is likely that this political "agreement" will dissolve and will be re- placed by a more realistic approach to the various substantive issues involved. Consequences Subnational organizations will have to reach agree- ment about a balanced approach to aggregating data about individual needs, resources, and uses of public services, and to what degree invasion-of-privacy problems will restrict the types of data otherwise useful for subnational planning and operations. Institutional Requirements Interagency requirements, particularly at subnational levels, require a good deal of Federal guidance in the matter of confidentiality. Certain minimum require- ments could be established, for example, and model sets of guidelines, technical methods for computer control, etc., developed by the Federal Government. Research, development and demonstration of these methods could be funded and promoted by one or more Federal agencies, and Federal- subnational level agreements reached. 16 STRENGTHENING THE FEDERAL ROLE SUBSTANTIAL STUDY OF SUBNATIONAL STATISTICS AND ITS INSTITUTIONS IS NEEDED The first half of the 70' s is a good juncture during which several public policy questions regarding sub- national statistics should be raised, issues formulated, and alternatives discussed and evaluated. One reason for this, pure and simple, is the existence of the President's Commission on Federal Statistics. Previous Commissions have not looked in depth at the many sub- national data issues so obvious to us now in 1970. Ideally, it is now appropriate to raise three basic questions about subnational statistics and, while the Commission can only begin to investigate these issues, it will leave as a legacy these types of questions. First, there is the matter of resources, touched on above. While no one can say, authoritatively, what fraction of the total Federal statistics budget is con- sumed in making data available for subnational areas, it is undoubtedly a large and increasing amount. Related to this concern are the unknown, but also substantial resources devoted by subnational areas to increasing the supply of statistics not available through Federal means. Are these quantities of resources too much, inadequate, or "just right"? Furthermore, are there ways to change the allocation of resources to increase the effectiveness of the resources now devoted to subnational statistics? A second question one would ideally wish to see analyzed is that of the Federal public sector role itself. Why, in each instance, is not the private sector (or perhaps some quasi-public agency) more involved in providing for subnational statistics, or other levels of government? Could and should the subnational areas be doing more or less? In what way could different sub- national levels be more effective in producing and using subnational statistics? A third question might be addressed to the matter of what Federal statistics coverage should be. For any given Federal statistics budget, what demographic, economic, social and physical statistics should be made available by the Federal Government? These questions are obviously interrelated: how much in toto (the first question); how much and what role for the Federal sector (the second question); and what sta- tistics (the third question)? To complete the dimensions of such an ideal study of subnational statistics, a full systems analysis approach might proceed by defining all actual and potential suppliers (and producers) and demanders (requirements for) of subnational data; deriving the benefit and cost functions attendant to meeting present and future gaps between supply and demand, through altered institutional arrangements, changing budget levels, and technological advances; evaluating these alternatives, using qualitative and quantitative factors for selecting between competing alternatives (i.e., competing institutional arrangements, budget levels, and technologies); and recommending favorable alternatives to interested parties and decision- makers. So much for a glimpse of the magnitude of the subject area and an ideal approach to formulating and evaluating alternative solutions. Given the context of the Presidential Commission's program and the fact that no previous Commission on or about Federal statistics had even briefly explored the vast ranges of the subject matter, it is not sur- prising that the Commission selected the role vis-a-vis subnational statistics which it evolved to deal with the many-faceted problems and subproblems. No one, in fact, can fault the Commission for the role it selected, that primarily of description or "pulling together" what was known about the overall topic, as well as making partial analytic and factfinding explorations, one probing a crucial institutionalization question, the other an equally important technological problem area. The forbidding level of effort involved in "starting from scratch" is only one of the numerous reasons for the Commission's decisions to pursue a specific rather than an all-embracing course of action. Several calendar years and certainly many man years of effort would have been required to perform the total study outlined above, with an enormous amount of intergovernmental cooperation required and a large body of newly created data amassed. While I do not look forward to the making of such a study, I look forward even less to the consequences of not making it. That the benefits will exceed the costs I have no doubt. Meanwhile, it is not too early to begin the development of the institutional changes required for the environment of the 70' s. INTRASTATE STATISTICAL COORDINATION Herbert Alfasso Director, Statistical Coordination New York State Division of the Budget INTRODUCTION The President's Commission on Federal Statistics, in exploring ways to improve Federal-State-local cooper- ation in the development of small-area statistics, com- missioned a study on intrastate statistical coordination. The study was concerned with improving Federal- sub- national statistical relations through intergovernmental cooperation and coordination, focusing specifically on the role of States as each State finds ways to make its own intrastate statistical practices more effective and efficient. The specific objective of the study was to analyze and evaluate intrastate statistical coordination activities with specific attention paid to those which affect, and/or are affected by, Federal statistical activities. Emphasis was given to seeking practical and effective Federal policies, stategies or guidelines for strengthening intra- state statistical functions. No precise definition of "statistical coordination" was developed in the study. Rather, all practices of coordinating statistical activities both between the State and its local government and among State agencies to rationalize and make more effective use of data through joint production, through cooperation with Federal agencies and other such means are embraced within the concept "statistical coordination." Moreover, there was no attempt to assess the precise benefits and costs of such programs except in a rough, qualitative manner; perhaps the only way possible given the current state of understanding about use of statistical information. The study involved two major segments. One was a questionnaire survey of all States to learn about ongoing intrastate coordination activities (if any). The second part consisted of an in-depth study of three States operating successful statistical coordination functions. RATIONALE FOR FEDERAL INTEREST IN INTRASTATE COORDINATION Several facts have served to increase the President's Commission's interest in this subject: Despite efforts by all Federal administrations during the 1%0's to improve Federal-State re- lations in statistics, progress has been disap- pointingly slow. There has been an increased awareness at both the Federal and local government levels that data on the subnational level (States, counties, cities and even small areas within cities) are increasingly essential as all units of government assume greater responsibilities for dealing with the Nation's social and economic problems. Many of these data might be mined from ongoing State and local programs. While a clear logic for the Federal Government to provide leadership in this area can be adduced, the priority that the development of intrastate coordination should have and the best ways to implement this admittedly difficult and vaguely defined goal are not clear. There is a strong case for Federal promotion of intrastate statistical coordination which is consistent with the current national policy to delegate handling of many of the Nation's problems to the State/local level. The major areas in which an interstate coordination program in each State would facilitate the efforts of individual Federal Agencies and the Federal Govern- ment as a whole are: (1) Large and increasing amounts of moneys trans- ferred from the Federal Government to State and local governments, require considerable local area data for operational purposes. In fiscal year 1971, over $25 billion was transferred from the Federal Government to State and local govern- ments as grants-in-aid. (2) Nearly every Federal program has to be justified to members of Congress in terms of impact on the local constituencies they represent. Consequently, uniform measurements of the local impact of each program are a requirement in furnishing infor- mation to Congress. (3) The logistics of Federal program operations alone require considerable local data. Ninety percent of Federal employees are based away from Washington, D.C. Cost-benefit analyses of their activities call for comprehensive uniform State-local data. (4) It can be argued that where a single office in each State is responsible for establishing statistical 17 18 STRENGTHENING THE FEDERAL ROLE standards, there is greater likelihood for develop- ment of interstate comparability of statistical series. The prospect of achieving interstate comparability of statistics in the recent past was generally considered a major benefit to be achieved by creating State statistical coordi- nation offices. (5) Much data which can be used by Federal depart- ments is actually accumulated as a byproduct of State operations. However, these statistics often become available only under the impetus and guidance of a State statistical coordination office. (6) Having a single contact point which represents the entire State government in itself can be a major benefit to many Federal agencies. (7) A State statistical coordination office can help expedite certain Federal statistical programs, especially for those Federal agencies which have no direct State counterparts. The considerations listed above are those that would be of current major interest to the Federal Government. In addition, central statistical offices can perform many other functions which would benefit their respective States, including the whole gamut of activities which tend to increase the proficiency of statisticians and the effectiveness of State statistical activities. Some of these "other functions," moreover, may, in the long run, prove to be of greater value to the Federal Govern- ment. RECENT EFFORTS AT FACILITATING FEDERAL- STATE RELATIONS AND INTRASTATE COORDINATION In February 1966, at the recommendation of the Ex- ecutive Committee of the National Governors' Conference, the First National Conference on Comparative Statistics was held in Washington, D.C. It was a well-attended affair sponsored jointly by a host of organizations representing Governors, mayors, cities, city managers, and States, and organized by the Council of State Governments. One of the recommendations- -probably the most significant- -was that each State develop an agency to coordinate statistical activities within the State and to serve as a channel to the Federal Govern- ment and to other States. Until then (1966), only a handful of States, including New York and California, had developed statistical co- ordination programs. Although agreement for holding a National Conference on Comparative Statistics on an annual basis was universal, there was some problem in deciding which organization should bear responsibility for guiding the efforts. There was much feeling that the Advisory Commission on Intergovernmental Relations (ACIR) would be the most appropriate agency, but it declined. The Office of Statistical Standards (OSS), U.S. Bureau of the Budget, agreed to undertake this function, but it was reluctant at that time to have the Federal Government assume leadership. The final agreement was that OSS would serve as secretariat, by organizing the annual conferences and generally providing the staff work, but the policymaking would evolve from the newly created National Conference on Comparative Statistics comprised of State and local officials. 1 As a result of the 1966 conference, efforts within States to establish statistical coordination offices in- creased. According to information furnished to the National Governors' Conference in August 1967, 27 States had established or were in the process of establishing State statistical coordination functions. But a survey conducted by the Office of Statistical Standards in 1968, just after the Second National Conference, revealed that only 13 of these offices were in actual operation. (Based on the surveys made in both 1968 and 1971, the number of functioning statistical coordination offices- -frequently reported at 30--appears to have been highly exaggerated.) Shortly after the Second National Conference in 1968, activity by OSS in this area waned. Despite lack of leadership, however, new State offices continued to develop. At the Second National Conference, the program included sessions on both successes and problems of newly formed State statistical coordinating offices and some Federal actions since the First Conference. The latter included mention of the Census Bureau seminars for State and local government officials (these 1-week seminars, started shortly after the First National Con- ference, proved highly successful, and have been con- tinued to this day); the OEO Federal-State Information Exchange System; the Census-State Cooperative Pro- gram for Developing County Population Estimates; and "some new services from the Bureau of Labor Statistics." The Census Bureau especially plays an invaluable role in making conscientious effort to fulfill the recom- mendations made at the First National Conference on Comparative Statistics (later reinforced by the Inter- governmental Cooperative Act of 1968). The recom- mendations pertaining to Federal agencies included exploring the needs for consistent and comparable sta- tistics at the State and local level, fostering intergovern- mental coordination to meet these needs, and communica- ting statistical information useful for planning and evaluating many of the Federal-State-local programs. The list of Census Bureau activities along this line is too long to repeat here, but it includes many other services in addition to its seminars, such as publications to acquaint users with statistical services available from the Federal Government and the Census-State Cooperative Program for developing Local Population Estimates. Of equal importance, is the Bureau's helpful attitude towards State and local governments which encourages such officials to look to the Bureau first for many kinds of assistance including speakers for State programs, technical advice in carrying out surveys and general guidance on almost any problem in the statistical area. 'The unwillingness to have Federal leadership (or even the appearance of it) at the time, in this writer's opin- ion, is unfortunate. The avoidance of Federal "control" is largely forgotten now, but it probably had a detrimental effect on efforts to establish State statistical coordi- nation units because there was no continuing pressure to initiate action while enthusiasm for intrastate coordina- tion was still intense. INTRASTATE STATISTICAL COORDINATION 19 In carrying out the spirit of the suggestions made by the National Conference on Comparative Statistics, the Census Bureau has partially filled a large vacuum in Federal-State cooperation. Unfortunately, the Bureau does not have the authority to bring the entire Federal statistical establishment together to foster improved Federal-local relations. MAJOR FINDINGS A. Findings based on 37 responses to a mail question- naire survey of all States conducted for this report: 1. Sixteen States have formal intrastate statistical coordination functions in operation and ten more have plans for creating such functions. Intra- state coordination is growing slowly and in a haphazard fashion. 2. Fourteen of the 16 "operating" States use per- manent staffs attached generally to a budget or planning office; the remaining two operate solely through committees (without special secretariats). 3. With one exception, States operate the statistical coordination function with very small staffs of between one and four full-time professionals. 4. Nearly all State statistical coordination offices with permanent staffs have some operating sta- tistical responsibility. Most often, this consists of publishing an annual statistical abstract and/or a periodical "Statistical Reporter." In addition, many offices also have been assigned functions with little evident relevance to intrastate coordination. 5. There is an apparent trend to develop State "management information systems"; most (but not all) appear to be associated with the intrastate coordination function. Although there is some tendency for statistical coordination activities to relate to computerized data systems, the pattern of these relationships was not clearly revealed by the survey. 6. Most existing statistical coordination offices ap- pear to be performing only minimally in most of the areas identified earlier in this paper as being of major benefit to the Federal government. Notable exceptions exist in those fields where Federal guidance or leadership has been manifested. 7. Despite the fact that one of the reasons given at the First National Conference on Comparative Statistics for creation of a central State sta- tistical agency was to improve Federal-State relations, four of the 16 States indicated that this was not part of their coordination activity. B. Findings based on an in-depth analysis of intra- state coordination operations in three States (California, New York, and South Carolina) 1. Intrastate statistical coordination in each of these States has produced significant benefits, parti- cularly when provided with top-level support and competent leadership; the benefits under those conditions were high in comparison to the cost. 2. Although each State operates differently and under different conditions, all currently have serious problems which stem from a number of different events and situations. California, which operates solely through an interdepartmental committee, currently views its role narrowly ("providing limited coordination efforts"). It is the oldest of the intrastate coordination efforts (established in 1945) and its operations appear to have become static. New York State is encountering a fiscal crisis which may limit further growth in (or even narrow) the role of intrastate coordination. South Carolina's agencies generally lack research- statistical units. This lack of statistical pro- ficiency is forcing the State statistical coordinator to focus heavily on promoting such proficiency, in contrast to the coordinating emphasis possible in California and New York. 3. None of the States has focused on developing strong State-local government statistical relationships. 4. Fragmentation of Federal-State-local statistical policies by Federal agencies, plus the lack of leadership by a central Federal agency in provid- ing recognition and guidance, are seen as detri- mental to the development and effectiveness of intrastate coordination in New York and South Carolina. 5. Computerized data systems of various types are being developed in all three States. However, the degrees and types of relationships between intra- state coordination and such "systems" vary considerably. 6. The staffs in New York and South Carolina carry out a number of operational functions, some of which are not related to intrastate coordination. The California committee also carries out opera- tional functions, such as preparing analytical re- ports for the Governor regarding major policy issues on poverty or unemployment, but these are done sporadically. RECOMMENDATIONS The recommendations made to the Commission on the basis of this study are: 1. The Federal Government should take a leadership role in encouraging and promoting intrastate statistical coordination by helping create such functions where they do not now exist, and by providing recognition, technical guidance, and other assistance for improving intrastate coordi- nation activities, where such functions do exist. 2 2 The logical Federal agency,' and the one with a "man- date" for assuming leadership, is the Office of Statisti- cal Policy, Office of Management and Budget (formerly Office of Statistical Standards, Bureau of the Budget). It has the requisite authority to (a) coordinate the many existing Federal-State-local statistical cooperative ef- forts; (b) encourage increased Federal-State cooperative efforts; and (c) to develop a Federal statistical policy with regard to local areas. Such a policy should stand- ardize or provide guidelines not only on what information and detail should be made available to local governments, but also help standardize information requested by Federal agencies from local governments. Further, the policy should include not only what is requested by Federal sta- tistical agencies, but also by regulatory and administra- tive agencies. 20 STRENGTHENING THE FEDERAL ROLE As part of a Federal policy with regard to local governments, the data needs of State governments should be given greater recognition than at present. Meeting State data needs might require some adjustment of priorities of data production at the Federal level. in all States, it should be located in a central staff agency or, perhaps, the legislature. Since the location factor is crucial to the ultimate effective- ness of the intrastate coordination function, guidance should be offered to States on an in- dividual basis as needed or desired. 3. Federal promotion of intrastate coordination functions should include the following activities: (a) Maintain continuous contact with central State statistical agencies to provide information both on what other States are doing, and on the status of Federal activities and policies. (b) Organize regular meetings of State statistical coordinators, at Federal expense, to facilitate sharing of experiences and to obtain feedback from States on what the Federal Government might do to improve Federal-State cooperative efforts. (c) Secure continuing support and involvement by the Advisory Commission on Intergovernmental Re- lations, the Council of State Governments, the National Governors' Conference and other public and private organizations directly involved in current problems. The development and improve- ment of State central statistical coordinating agencies will be maximized by promoting activities directly related to the problems perceived by these organizations. (d) Route Federal requests to States for information (not involving a specific Federal-State program) to (or through) the central statistical coordination office. (e) Establish a training program, on perhaps a 50-50 Federal- State cost basis, to help States with weak statistical/research capabilities, as, has beendone by the National Institutes of Health for training biostatisticians in the public health field. Of course, budgetary limitations, as well as analysis of the costs and benefits of this form of training as opposed to other training programs, merit further attention. 4. In helping to establish or improve intrastate co- ordination activities, the Federal Government might be guided by the following considerations: (a) Intrastate coordination activities should include permanent staffs with the highest caliber leader- ship available, since quality of both staff and leadership are critical to a successful operation. The role of intrastate committees of directors of research and/or statistics should be limited to advice not direction. (b) Intrastate coordination offices should be en- couraged to conduct related operational activities, such as publication of a statistical abstract, preparation of population estimates and pro- jections, and development of a central State data system. (c) While there is no single best agency to which the intrastate coordination function should be attached (d) Central State coordinating agencies should be encouraged to foster access to information de- veloped as byproducts of ongoing operations in State agencies. These statistics, which are use- ful immediately for intrastate purposes, should eventually be standardized, with Federal guidance, for interstate comparisons, where desirable and feasible. INTRASTATE COORDINATION IN ALL STATES- SUMMARY OF SURVEY To obtain data for this report, a mail survey of all States was conducted, employing a questionnaire, to determine whether a formal intrastate coordination function exists, how and when it was established, how it operates, and its activities. For States with no current coordination activity, a question was asked as to whether plans exist for establishing formal procedures in the near future, and when such plans could be expected to be implemented. Thirty seven States responded to the survey. Of these about 16 had some formal intrastate statistical coordina- tion function. (The number having formal intrastate coordination functions is approximate because of defini- tional problems. At least one State considered OEO's S/FIX System a "formal" statistical coordination function (which it is not) and one or two States indicated that there was no such formal function, although descriptions of their activities indicated the contrary.) Eleven of these States were also engaged in setting standards for either municipal codes or statistical standards. In addition to the 16 offices "in being" in May, 1971, 10 are planned for the future. The time-table of estab- lishment is as follows: Year Before 1966. 1966.1968. 1969-1970 Subtotal-offices existing in May 197 1 June-December, 1971 1972-73 After 1973 Date uncertain Subtotal-potential future offices Total offices-in operation or being planned. . Number of States 4 5 7 16 10 26 INTRASTATE STATISTICAL COORDINATION 21 Thus, 26 of the 37 States responding to this survey have organized or are planning to create formal State coordinating functions. Three of the States said they are currently in the process of creating such an activity; others have plans for as late as 1974-75. Most of the offices were established by either Gover- nor's executive order or administrative action. Only four have some authority derived from legislation and two of these four had their start through some sort of executive action. Fourteen States operate through a permanent staff. Two operate solely through committees (a permanent committee in California, and an ad hoc committee in New Jersey). Eight States having offices with permanent staffs are attached to the State planning agency, another five are located in the State budget or fiscal office. The placement of an office does not seem to be associated with State size or geography. Since location of the statistical coordination office within a State structure may be critical, this is an area for additional study. Materials reviewed for this study and discussions held with State officials indicate that there was, in some States, a good deal of concern over the very important location factor. Most States operate the statistical coordination function with very few people. By far the largest is the State of Washington with 10 professionals; all the others have from one to four full-time professionals working on statistical coordination. (Iowa's office was still unstaffed at the time of the survey.) In general, the total number of professionals as- sociated with statistical coordination and other related functions is not much greater than for the statistical coordination function alone. (There are some outstand- ing exceptions, however.) Although reliability of some of these figures may be questionable because of the definitional difficulties confronted by those filling in the questionnaires, the trend is unmistakable- -intrastate coordination is carried out with very small staffs. Responses were solicited for 10 major functions which were believed to be assigned to offices of this type: (a) Clearing questionnaires. Only three of the 16 offices reviewed and cleared questionnaires; two more either do it "on oc- casion" or "it is under consideration." (b) Publishing statistical or statistics-related materials. Nearly all offices issued publications, the most popular publication being a State statistical ab- stract (10 offices) with a newsletter or "statistical reporter" a close second (8 offices). Four offices publish inventories of statistical series or other publications, including statistical guidelines, an economic report, and "special" reports. (c) Coordinating EDP. No office indicated that it combines EDP coordi- nation with statistical coordination. Where co- ordination of computers is accomplished (as in New York and Washington), that function is carried out by another unit. (d) Developing management information systems. This survey did not specifically ask whether statistical coordination is associated with manage- ment information systems. However, comments added to some questionnaires indicated that sta- tistical coordination is frequently associated with developing State information systems. (e) Preparing population estimates. At least 10 offices prepare population estimates. (f ) Issuing or promulgating standards. Eleven offices issue standards for municipal codes, including four which also issue statistical stan- dards. Another three offices issue statistical standards but not standards for municipal codes. (Statistical standards typically include guidelines for conducting surveys and formats for presenting statistical materials.) (g) Reviewing statistical budgets. Only three of the 16 offices review budgets of other State statistical agencies. (h) Maintaining Federal liaison. Despite the fact that one of the reasons given at the First National Conference on Comparative Statistics for creation of a central State statis- tical agency was to improve Federal- State re- lations, four of the 16 offices indicated that this was not part of their coordinating activity; another response was "only partially." (However, two of these four states are California and New Jersey, which operate solely through committies, and do not have permanent staff necessary for this activity.) (i) Helping to secure Federal data for State use. Eleven of the 16 offices do some cooperative work with the Federal government. Much of this was connected with obtaining summary tapes from the 1970 census. (j) Providing services to local governments and to nongovernmental organizations. Half of the 16 offices indicated they did some work with local governments or nongovernmental organizations. Both "local governments" and "nongovernmental organizations" were combined in one question. Judging from the comments elicited by this question, very few States have strong ties to local governments. Except for Georgia and Hawaii, where there was a definite indication that the State central coordinating office works closely with local governments, the other offices answering "yes" to this question 22 STRENGTHENING THE FEDERAL, ROLE appear to deal only minimally with local govern- ments (answering occasional inquiries, distri- buting publications, etc.) (k) Performing other functions. Other functions performed by statistical coordi- nating offices include: conducting socioeconomic research, developing an A-95/A-98 (Federal circulars), acting as a clearinghouse , performing EDP systems work, reviewing legislation for the Budget Director, or conducting reports manage- ment. In some cases, the functions listed above were cited as the major activity of the intrastate statistical co- ordination office. As elicited by questionnaire responses, the most popular functions performed by intrastate coordinating office are: Issuing publications (statistical abstracts, news- letters); maintaining Federal liaison for general informational purposes and to obtain Federal data for State use (in many States, "obtaining Federal data" is limited to securing 1970 census tapes and thus ignores the broad spectrum of other data which could be used by the State); and promulgating standards, especially for municipal codes. This would imply that fostering computer- ized data systems is often related to the intra- state coordination function. CONCLUSIONS RESULTING FROM THE SURVEY The purpose of the survey was to provide some indication of the state of development of statistical coordination among States. The responses indicated that, since 1968, there has been progress in terms of numbers of States instituting formal State statistical coordinating activities. And more such progress can be expected in the future. But the writer cannot help being impressed at the small staffs involved, and the apparent subordination of statistical coordination to other functions, such as making population projections or establishing and operating a census summary tape center. Although it has been demonstrated that such operational functions can strengthen the office and should be included with the intrastate coordination function, it appears that most States do not perform in many of those areas which would be of most benefit to the Federal Government (as cited earlier in this paper). Managing coordination functions solely through com- mittees and does not seem to promise vigorous develop- ment. While many of the intrastate coordination offices show considerable vitality, few have substantial relations with the Federal Government other than the Census Bureau. The other major Federal statistical agencies which work with States tend to deal with their State agency counterparts directly, thus bypassing the central State statistical office. In summary, the intrastate coordination picture as viewed from this survey is growing, but it is growing slowly and haphazardly. Without Federal guidance, there appears to be little chance that intrastate coordination offices will (for example) either make estimates for small geographic levels or help accumulate data as a byproduct of State operations which can reasonably be compared from State to State. Even minimal benefit of having a single State statistical contact point which represents the entire State government may not be assured unless these offices are recognized, dealt with and helped by that Federal agency in the best position to accomplish this--the Office of Statistical Policy. What appears most incongruous and unfortunate is that so much could be accomplished in this area with so little manpower since the mechanisms, (the statis- tical coordination offices) already exist in many States. Without Federal help, in the form of continuing guidance to States, there can be no standards for intrastate co- ordination, just as without Federal intervention there is little likelihood of achieving comparability of statistics. DESCRIPTION OF INTRASTATE COORDINATION IN THREE STATES The States to be selected for in-depth analysis were those which met the following criteria: (a) Had been operating a statistical coordination pro- gram for several years, (b) were different from each other in the manner in which they functioned and in their activies, (c) represented different size States, (d) were considered to have generally successful operations and, as a practical matter, could be dealt with within the relatively short period of time available for doing the study. The three States ultimately selected were those which met the above criteria and with which the writer had had some prior familiarity- -New York (in which he heads the statistical coordination functions); California (which operates solely through a committee); and South Carolina (a small State in which the head of statistical coordination is also responsible directly for other functions). For each of the cases examined, the origin of the coordination unit, its functions and methods of operations, and current status were described and reviewed. THE NEW YORK STATE EXPERIENCE New York's statistical coordination program started in 1964 with a permanent staff of two persona in the Division of the Budget. It was created by an adminis- trative action of the Budget Director. The specific impetus for establishing the office came when the State Budget Division instituted an integrated INTRASTATE STATISTICAL COORDINATION 23 planning-programing-budgeting-system (PPBS). How- ever, support for such an office had come from the influential Albany Chapter of the American Statistical Association as early as 1956. Although there was a host of statistical gaps, poor reporting of statistics, duplication and other such problems which created the demand for statistical coordination, research and statistics activities in almost all State departments were mature functions which reached back for decades in many of the major depart- ments. The quality of the State's professional re- searchers compared favorably with Federal professi- onals. Consequently, the research fraternity, while willing to have the Statistical Coordination Office work with them, was a formidable group to lead. Further- more, the Office had no built-in controls other than implied powers over statistical budgets by being located in the Budget Division. In broad terms the four principal functions of the central statistical office were: ( a ) To improve statistical output necessary for decisionmaking, budgeting, forecasting and planning. (b) To promote the exchange of information within State service and between the State and non- State organizations. ( c ) To act as a clearing house. (d) To provide technical assistance. Among its specific functions were: (a) Reviewing and analyzing statistical output of agencies, encouraging improvement in quality of data, initiating new series, and eliminating duplica- tion and outmoded series. (b) Establishing standards and uniformity in report- ing (tables and charts), in terminology, bases, geographical areas, etc. •(c) Promoting more current reporting of statistical data. (d) Encouraging production of basic data needed for agency forecasting. (e) Acting as secretariat to the Interdepartmental Committee on Research (an informal group of research directors) to facilitate exchange of re- search ideas and data, encourage greater inter- personal relationships among the research directors, and promote training and other such programs. (f) Holding periodic conferences of State statistical and research personnel to discuss topics and resolve problems of mutual interest, including economic, financial, and population forecasts, and new statistical techniques. (g) Publishing the bimonthly New York State Statistical Reporter containing information about research and statistical studies and reports, and other pertinent information. (h) Conducting an inventory and publishing a list of statistical series emanating from State agencies. (i) Preparing and publishing an annual statistical abstract. (j) Acting as liaison with the Bureau of the Census and other Federal agencies to facilitate inter- governmental coordination of statistical activities. (k) Responding to inquiries concerning sources of data. (1) Assisting research and statistical offices in State departments in locating qualified employees and in facilitating interagency transfers (an in- formal and effective employment service function which helped optimize development and use of sta- tistical personnel). (m) Assisting budget examiners in determining needs of statistical units, quality and importance of out- put and related aspects. (n) Counselling agencies in their research and sta- tistical projects and activities. (o) Recommending to the Budget Director and Gov- ernor policies and positions relative to statistical issues and activities of the Federal Government and other organizations. (p) Representing the State on intergovernmental and professional bodies dealing with statistics and research developments. Developments Since 1964 The New York State Office of Statistical Coordination soon became the model for many States starting a similar function. Its New York State Statistical Reporter was widely emulated and many of its functions were incor- porated by newly formed statistical coordination units. Probably the major reason for its popularity as a model State statistical coordination office was the fact that the First National Conference on Comparative Statistics in 1966 featured the New York experience which had started only 2 years before. The unit's national reputation for effectiveness was based in part on: (a) Being among the first States to undertake sta- tistical coordination with full-time permanent professionals. (b) Initiating several publications which were later widely emulated. (c) Drawing its personnel from experienced State statistical- research personnel with wide acquaint- anceships among the professional fraternity. (d) Establishing acceptance by Federal officials and participating in initiation and formulation of some 24 STRENGTHENING THE FEDERAL ROLE of the Federal-State cooperative projects listed earlier in this report. and unprecedented cutbacks in many State pro- grams. Major changes since the organization of the unit in 1964 have included the following: Method of Operation Employed by the Statistical Coordi- nation (a) The publication of the Statistical Yearbook in 1968 provided a "handle" on the statistical output of all State agencies. The Yearbook also proved useful as a device to identify and fill statistical gaps. (b) The abandonment of the State's planning-pro- graming-budgeting system PPBS program, which proved impractibable, freed the unit to concen- trate on basic data for other purposes (some of which were analogous to PPBS). (c) The Statistical Coordination Office began to add new functions which were only peripherally related to its original functions. Many of these were more closely related to the main functions of the Budget Division, such as reviewing pro- posed legislation. (d) The Office developed closer ties with Federal agencies. For example: the State joined the Federal Statistics Users' Conference with the Director of the Office as its official represen- tative and currently one of its trustees; the Director also became a member of one of the Census Bureau's advisory committees. (e) An effort was made to develop user "focal points" for the large amounts of underutilized statistics developed by State agencies. Such focal points could include economic and social indicators or an econometric model. (f) Through the Interdepartmental Committee on Research, the Statistical Coordination Office in- creased its efforts to improve the quality of State research and statistical personnel by such means as increasing inservice training opportunities; improving recruitment methods to permit hiring of qualified people from outside State government (the State's civil service system tends to freeze out professionals who are not prepared to enter as trainees); and improving examinations for re- searchers. (In New York, a competitive written or oral examination is required for every pro- motion and, because of the large volume of tests to be given, assistance to test-makers is useful in focusing on qualities most important to the job. (g) A Statewide Information System Steering Com- mittee (SWIS) was created with a secretariat provided by the Statistical Coordination Office. SWIS was designed to help coordinate the infor- mation systems of State agencies and to help meet the information needs of the new planning com- missions created by the Governor in the late 1960's, as well as the older agencies. The current status of SWIS is vague since most of the planning commissions have been combined into a single State planning agency whose policies have not yet been formulated. These actions were a result of the current budget crisis which resulted in severe The effectiveness of the statistical coordination function in New York depends on its staff keeping abreast of statistical needs and problems in the State and devising means to meet the needs. Relationships between the various agency research and statistical bureaus and the Office vary from close to minimal, based on the nature of policy problems. When a new statistical function develops, the Office participates in assigning responsibility to an appropriate agency. Thus, the Office was instrumental in getting the State Labor Department to deal with the Bureau of Labor Statistics in purchasing additional area wage survey data for additional SMSA'sfor use in negotiations between local and State governments and their employees. The Office also assisted in developing a Census Tape Summary Center in the State. In addition to resolving specific problems, the sta- tistical coordination staff meets periodically, on an informal basis, with research directors to discuss their research and statistical activities. The Office makes suggestions on how the agency research office can better serve its own agency and the central staff agencies such as the Budget Division. Despite the close relationships, the statistical coordi- nation staff is required to maintain distance and objectivity and does not consider itself a friend in court to the research directors. When reviews of statistical budgets are made, they are made from the viewpoint of the Budget Director. One of the most persistent efforts made by the sta- tistical coordination staff is the encouragment for greater analysis and use of the State's vast statistical output so that the data might be better used in the decision making process. There is a general dearth of such analysis, resulting in failures to adequately evaluate existing State programs. Current Status of the Office of Statistical Coordination The current status of the office is obscured by a number of new developments, the most important of which is the budgetary crisis facing New York State. The budget crisis has affected the Statistical Coordi- nation Office through termination of publication of the Statistical Reporter (although the Yearbook continues) and curtailment of funds for coordinatingwith the Federal Government. Thus, plans for securing unpublished data on New York State from the Current Population Survey has been postponed. For the next year, it appears as if the Office will have to focus on functions which will have potential budget- savings implications. INTRASTATE STATISTICAL COORDINATION 25 THE CALIFORNIA EXPERIENCE California's statistical coordination program is pro- bably the oldest such formalized State activity in the country. It was established in September 1945 by an Executive Order issued by Governor Earl Warren when he created the State Interdepartmental Research Coordinating Committee (SIRCC). Each Governor of California since then has con- firmed the appointments of the members of SIRCC by personal letter and has made successors and new appoint- ments as the situation required. There is no permanent staff assigned specifically for intrastate coordination. But in 1968, at the Second National Conference on Comparative Statistics, the chairman of SIRCC said that California was trying to establish full-time positions to serve as secretariat and to carry out assignments proposed by SIRCC. How- ever, no staff was ever allocated for this function. The creation of SIRCC as early as 1945 demonstrates both the strong sense of professionalism among the research directors in California and the recognition of the value of research and statistics by the State government. Functions and Method of Operation Governor Warren, when he established the State Interdepartmental Research Coordinating Committee, cited the following functions: To serve as clearing- house to promote interchanges of information among the statistical and research divisions of State agencies; to suggest procedures for coorelating the statistical and research work of State agencies for eliminating duplica- tion; and to recommend the further processing and analysis of data assembled by State agencies that would provide information to State officials and to the public. The Committee is comprised of 20 members who meet monthly (except August) for about 2-1/2 hours. The post of chairman is rotated annually. Since being on SIRCC is considered an important status activity, attendance at these meetings is obligatory. The main activity of the Committee is (according to the reply received for this survey) "to acquaint members with reports of general interest and research projects being undertaken by various agencies (rather) than to coordinate statistical efforts." But from information available, it is evident that SIRCC has accomplished a great deal in many other areas. For example, the Committee initiated and published the California Statistical Abstract in the late 1950's. ' Other accomplishments of the Committee include establishing standards for ethnicity, confidentiality, county codes, and areal definitions. In addition, the Committee has organized ad hoc committees to carry out special projects such as formulating resolutions opposing Congressional proposals to limit the 1970 census; developing reports and programs for the Governor, such as one called "Picking Pockets of Poverty," after the Watts riots and assessing the impact of post-Vietnam on the State; and training statisticians, research analysts, and economists. While each California Governor has supported the Committee with enthusiasm, the effectiveness and power of SIRCC has come from other sources as well. Evidence of its strength is indicated by the proportion of active people on the Committee, which was reported to always be high. (This contrasts with the low level of participation of New York's Interdepartmental Committee on Research.) Until recently, the real strength of the Committee came from the senior budget man who was a member and leading sponsor of SIRCC. He provided the muscle. The Committee also drew much of its strength from its cohesiveness--the research directors tended to pull together. California, like New York, had the makings of a superior intergrated statistics and research center. The basic ingredients were all there: high quality of staff and of research and statistical activities, functions which are well regarded by succeeding admin- istrations and by the government in general. But no independent statistical coordination office was created to help focus these activities or to motivate the re- searchers to utilize new technologies to meet the State's emerging problems. It appears that the Committee's apparent strengths of professional pride and cohesive- ness helped keep the organization static. And that the long-standing tenure of many members of SIRCC gave it a conservative orientation. Of significance about the Committee setup in Cali- fornia is that SIRCC would remain primary even if a staff were created. Any assigned staff was to serve mainly as secretariat and was not intended to be inde- pendent, as are the coordinators of New York, South Carolina, or the Federal Government. THE SOUTH CAROLINA EXPERIENCE The South Carolina statistical coordination and stand- ardization function was created shortly after the First National Conference on Comparative Statistics, held in February 1966, but not as a direct result of the conference. The need for a central statistical unit was discussed by various State officials in the summer of 1965. The Division of Statistical Research in the South Carolina Budget and Control Board was given the responsibility to serve as "central statistical unit" in April 1966. In May 1967, Governor Robert E. McNair formally es- tablished the statistical coordination function. The research and statistical environment in South Carolina is markedly different from that of the other two States. Whereas California and New York are both large, well-developed States from the point of view of professional research and statistical personnel and activities, South Carolina is not. Of course, the State is much smaller (population 2.6 million) and its problems, accordingly, are lesser in scale. However, South Carolina, in common with other Southern States, has 26 STRENGTHENING THE FEDERAL ROLE become increasingly industrialized, and statistical infor- mation has taken on much more significance for planning and decisionmaking than in the past. Nevertheless, with the exception of a few functional areas such as employ- ment security and welfare where Federal involvement has promoted establishment of standards, statistical competence in the State is generally poor. Of considerable importance to the intrastate statistical coordination function is the relationship between the executive and legislative branches. In South Carolina, as in most Southern States, the executive branch is relatively weak and the legislature (General Assembly) relatively powerful. Although the Budget and Control Board is the . most powerful State agency, it does not have the power of the New York State Division of the Budget or the U.S. Office of Management and Budget. The Budget and Control Board essentially carries out legislative intent with few of the broad or day-to-day controls that are in the hands of the aforementioned agencies. Consequently, because of the relatively low level of research and statistical professionalism and the lack of centralized executive control of State agencies, the sta- tistical coordination function created in 1967 was set in a much different milieu than that of New York or Califor- nia. South Carolina, however, did resemble New York in important aspects. Both States had permanent staffs assigned to the activity and both were able to enlist top level leadership to get the statistical coordination function started and operating effectively. The current program of intrastate coordination is being conducted with the following functions and ac- tivities: (a) Establishing a group of statistical representa- tives which represent the State agencies. Regular meetings of these representatives are held to provide a forum for exchange of information among agencies. The Statistical Coordinator is chairman of the committee. As needed, ad hoc subcommittees for specific purposes are established. (b) Publishing an inventory of State statistical series which lists the major resources emanating from State agencies. The publication is similar in scope and format to Statistical Series of New York State. (c) Publishing the South Carolina Statistical Reporter which provides research information to all State agencies. This publication also closely resembles the one issued by New York State. (d) Promulgating statistical standards for the State, such as base periods for indexes, definitions of payroll periods, and standards for statistical surveys. (e) Reviewing and clearing questionnaires to be sent out by State agencies for collection of data. Currently, only new questionnaires are reviewed, which are few in number. Although an adminis- trative order by authority of the Governor was issued making clearance mandatory, all new questionnaires are not necessarily sent to the Division for review. Further, the review function does not have the force behind it as does that of the Federal Government. The Division, however, does perform a useful service in this area, ex- ercising its leadership to achieve essentially voluntary compliance. (f) Maintaining liaison with Federal agencies and representating the State on Federal statistical policy issues such as SMSA definitions, support for a mid-decade population census, and repre- senting the State on the Federal Statistics User's Conference. (g) Coordinating and helping State agencies make greater use of Federal data for State use such as using the Office of Business Economics' county data on personal income and cooperating in the Census Bureau-State-County Population Estima- tion Program. (h) Promoting research activities ineach State agency. Most State agencies do not have research staffs. (i) Organizing seminars and training programs to develop the research potential in the State. Federal assistance, particularly from the Census Bureau and OBE, is often solicited and obtained. Closely allied with these functions are several others, discussed below, which are also the responsibi- lities of the nine-member staff of the Division of Research and Statistical Services. One function is economic analysis and projections. The most important aspect of this function is preparing projections of major State tax revenues to be used by the Governor, General Assembly, and the Budget and Control Board as the basis for developing the State's fiscal plan and budget. As more county information becomes available from State and national sources, personal income projections will be made for the economic areas of the State. Although the Division does not prepare a statistical abstract (a small one is prepared by another State agency), it does prepare an annual economic report containing many tables of economic data on the State. Population estimates for counties is a responsibility of a unit within the Division and it maintains close liaison with the Census Bureau as part of the county population estimation program established by the Census Bureau several years ago. Intercensal estimates, by county, are used as the official figures for computation of various rates such as birth and death rates, crime rates, and accident rates. The statistical information system has not yet been implemented because of lack of funds. It is anticipated that selected statistical data of broad interest will be placed in a data bank including the tapes of the 1970 census. Summary of Method of Operation South Carolina has closely tied intrastate coordi- nation to several other functions which tend to be INTRASTATE STATISTICAL COORDINATION 27 supportive of the statistical coordination program. The most important of these other functions, and the most visible, is making revenue projections for the State. Starting with this base, the Division of Research and Statistical Services undertook preparation of popu- lation estimates and projections as part of the State- Census Bureau Cooperative Program. Although inter- censal population estimates were made in the past by the State's Health Department, it was not deemed an important function by that Department. But it was viewed as important and valuable to the operation of the Division of Research and Statistical Services, which brought a higher degree of competence to this activity. In turn, exercise of this function strengthened the Division's position, as well as its role in statistical coordination. To these functions were added clearance of question- naires, publication of a statistical reporter, an inventory of State statistical sources, and an economic report. The Division also provides technical statistical assistance to the other divisions in the Budget and Control Board, and to other State agencies. Despite these substantive functions, the method of operation for interstate coordination is mainly by per- suasion and not by control. The greatest problem facing South Carolina's sta- tistical coordination program is the lack of emphasis on research and statistical activities in the State govern- ment. There are so many vacuums that, in order to coordinate an effective system, there first must be developed a group of professional statistical offices to work with. It is also clear that South Carolina desires Federal leadership or guidance. Too much time is being spent in seeking such help from the Federal statistical agencies wherever it can be found. Furthermore, fragmentation of Federal statistical policies relative to local govern- ments make intrastate coordination more difficult than it need be. FEDERAL-STATE-LOCAL COOPERATION IN STATISTICAL DATA PRODUCTION Morris B. Ullman, Consultant President's Commission on Federal Statistics In the spring of this year, a report was prepared for the President's Commission on Federal Statistics entitled "Federal Government Involvement in Data Col- lection for Subnational Areas." Since this report will appear in volume II of the Commission's report, its contents will only be outlined briefly here and only one or two points that appear to warrant further thought will be discussed. First, let me say that the term "subnational" rather than "small-area" is used, since the original intention was to include State and regional distributions as well as small areas. Small-area connotes counties and cities and their subdivisions. Also, in considering subnational data needs, we found that the whole subject was completely intertwined with the topic of Federal-State relations in data collection. The two topics involve issues so integrated that, in this report, we did not try to separate them. Separating the question of Federal-State relations from the question of small- area data needs confused, rather than clarified, the problem of developing subnational data. After extensive review of various Federal activities producing subnational data and extensive discussion with persons involved in these programs, the report was written in three parts. In the first part, the extent of the Federal interest in subnational data was reviewed and the overlapping needs of State and local governments were pointed out. In 1970, 14 percent of the Federal budget was transferred to States and local governments to aid in implementing Federal programs. Additional related activities were financed directly by the Federal budget, such as Federal-State cooperative data collection or training programs for State and local officials. It was also pointed out that, because of the extensive overlapping between Federal and State and local data needs, there is an extensive potential for further develop- ment of Federal-State cooperative programs. In addition to the ongoing cooperative programs in labor, agricul- ture, and vital statistics, significant developments are taking place in crime statistics, in education, in housing and transportation, and in other areas. The second part of the paper discusses related problems and proposals under three main headings: (1) Federal- State- Local Cooperation in Data Collec- tion, (2) Producer-User Communication, and (3) The Federal Leadership Role. Suggestions included the establishment of a task force to explore and develop further areas of Federal- State-local cooperation; the 28 further development of State central bodies for coordi- nating data collection in the States and to act as a channel for further coordination with the Federal Government and with other States; the establishment of a Federal Data Center with State and regional branches to handle the distribution of available data more efficiently; and the recognition of a Federal leadership role in terms of the assignment of specific responsibility and appropriate resources for implementation. The third part of the report is an appendix of 13 sections of which four describe the background for subnational data activities and nine describe ongoing Federal-State-local cooperative efforts. These state- ments supplement, rather than replace, the 1966 report on "Federal-State- Local Cooperative Arrangements in Statistics," Working Paper No. 1 of the Office of Statistical Standards, Bureau of the Budget (now the Office of Statistical Policy in the Office of Management and Budget). It should also be pointed out that the listing in this appendix is illustrative rather than exhaustive, and omission should not reflect on programs not included. Since this report was prepared, there has been addi- tional material to be examined in the reports by Alfasso and by Barraclough, in the survey of small-area data needs conducted jointly by the Committee on Small- Area Statistics of the American Statistical Association and the Census Committee of the American Institute of Planners, and in a survey of the small-area data needs of its members by the Federal Statistics Users' Conference. As usual in such a variety of contributions, one can find instances illustrating almost any position. There is, however, a strong undertone for more Federal-State- local cooperation and much evidence in support of a more definite Federal leadership role. In the ASA-AIP report mentioned above, the bulk of the replies are from persons ".. .immersed in the daily practical problems of the urban revolution and that are operating State, regional, county, city, and neighborhood programs that rely heavily on data." In summarizing these replies, the report states ". . .the overriding issue mentioned far more often and with more vigor (one might almost say more agony), than any other, is the need for a quinquennial census of population and housing. " Generally the demands are for more frequent, more accurate, and more detailed local data than are now available, especially in the fields of population and housing. In addition to a mid-decade census, there is also mention of an expanded CPS (Current Population Survey) and even for annual population and housing counts. A national house-numbering system, more FEDERAL-STATE-LOCAL DATA PRODUCTION 29 local participation in the collection and editing of the returns to improve accuracy, and more timely publication of the results are also mentioned in these demands (or rather, expressions of need). The demands for more detail are sometimes tempered by requests for more summarization at higher levels of aggregation than the block or the tract. Underlying these requests is the assumption that the tools exist which can meet all these demands simultaneously. It is interesting to note that most of the requests for timeliness are directed at the census operation which operates with probably the most sophisticated equipment available both in terms of hardware and software. It is my contention, stated briefly and in highly oversimplified terms, that the best way to attain the needs expressed would be through combined Federal, State, and local action. The data collection job which is to meet the needs for both detail and timeliness must be broken down into manageable segments. The casualty of such a procedure would be comparability between areas and quality, since it is doubtful that there are both resources and trained personnel to staff a data collection organization in each locality. To overcome these difficulties, the overall picture calls for strong Federal leadership to prepare specifications and procedures for each operating unit to follow; an extensive training program to train staff for this type of operation; and, hopefully, a research and evaluation program to apply quality control procedures and keep track of the problems of coordination. There is considerable comment from local respondents on the delays in makingthe 1970census results available. If we were to have a mid-decade census, there would still have to be a similar time span after collection before data would become available, namely, at least 9 months and probably a year or more. This timing apparently does not meet the request for more current data, although it may meet other important needs. Where the task is too big to achieve the objective, as this seems to be in terms of speed of data delivery, we can follow two paths--simplify the task or subdivide it into more manageable segments. Simplification of a national census operation, either by shortening the questionnaire or by operating through a complete mail-out/mail-back procedure- -which as- sumes the availability of a national address register- - will not result in any major savings of time. In the data collection phase we still must go through collection, control, and editing procedures. A short questionnaire would result in a higher rate of response and less follow- up and editing, but the complete procedure itself would be shorter than the 1970 operation by only a month or two. There would be savings in programing and tabulating time; but these are relatively short on the time scale. If everything were ideal, we might get our data in 9 months. From a more practical standpoint, it would be a year or more before we get the local feedback which seems to be needed. The second alternative is to subdivide the national task into smaller, more manageable units. From a national point of view, this would probably mean States and large cities, which, in some instances, would have to subdivide their efforts further into local units or even into subdivisions of the larger urban areas. There is nothing at present to stop any local area which feels an urgent need for planning data from taking its own census or developing procedures for getting its data from local records. In fact, many localities contract for special censuses and have other data projects on their own. The additional factor we are considering here is the possibility of maintaining some national comparability and attaining minimum quality standards over the wider areas. We have already developed, over time, Federal-State systems for labor statistics, for agriculture statistics, and for other data. There is considerable activity pointed in the same direction for education statistics, for crime statistics, for highway data, and in other fields. Each of these programs has worked with its counterpart State agencies independently of each other. While the need for coordination into a semblance of an overall integrated system has been expressed at the National Governors Conference and elsewhere, only the first faltering steps have been taken. The needs for such a system involve: (1) A set of national specifications and procedures which can probably best be drawn up at the Federal level, but with extensive potential-user participation. States and local areas could supplement these national requirements with additional items needed for specific local situations; (2) The development of State and local staffs adequate to carry out these tasks. At present such organizations are small, where they exist at all. In addition to building up the personnel for these activities, there will need to be an extensive training program. In this area again, there is need for Federal leadership; and (3) A research program both to develop and improve procedures and to evaluate the results, so that quality standards can be applied and used to improve the final product. Not only do we have the experience of the existing Federal-State data systems to draw upon, but the Federal Government, cooperating with AID and UN programs, has already had international impact with this type of program. The training material for foreign statisticians developed by the Census Bureau could easily be adapted for training State and local personnel. The Housing Survey Manual published by the Office of Management and Budget and material prepared for specific purposes --such as the procedures for making housing vacancy studies--could also serve as training material. In developing such a program there could be provision for quick payoff techniques, such as use of local adminis- trative records, as well as long-range preparation for participation in national efforts. A program such as the one sketched above will not be quick or easy, yet it seems to be the direction of the future. Coupled with a better data delivery system, it seems to me the only way to achieve most of the objectives needed to meet local demands. We have already taken 30 STRENGTHENING THE FEDERAL ROLE the first steps in this direction through a number of Federal-State systems and through the organization of 16 State coordinating bodies mentioned in the Alfasso report. Our next steps would be: (1) more specific recognition by the Federal Gov- ernment of its leadership role in the development of a Federal-State-local data collection system including the assignment of specific responsibility for this function and the allocation of appropriate staff and other resources; (2) the extension of the system of State sta- tistical coordinating bodies and their development to function properly as key elements in the system; and (3) more activity at the State and local level, both in the development of local sources of data and in more extensive use of statistics in planning the solution of local problems. Basic to the success of any such program is the question of executive acceptance. At all levels, we must demonstrate the value and need for a statistical approach to the administrators who must be willing to invest the resources necessary to produce a useful product. In conclusion, one of the letters received from a local official in the ASA-AIP survey contained the following suggestion: "A definite proposal by a lead agency, such as the Federal Government, to send teams to small areas and establish, in conjunction with local help, continuing programs as presented in possibilities disclosed by the trial census run conducted in New Haven. The Census Use Study could be immensely useful to local govern- ment, if established on a continuing historical basis, and although local government probably would not buy every- thing contained in these studies, it would be a tremendous step forward to establish a geographical base file with information from local government records. .... "the possibilities and ramifications disclosed by the New Haven Study present a very significant break- through. . . .but it is intuitively recognized that this valuable tool will never be used unless the first steps could be physically initiated by the Federal Government. I would think that inasmuch as this work would be pretty much repetitive from small area to small area, that Federal teams, complete from people doing desk or record audits to programers, would develop quite a bit of waste- saving expertise." This paper leaves many questions unanswered. It merely presents a broad outline and a long-term objective. The planning, development of staff, and marshalling of resources for such a program will require dedicated leadership and time. We have experience with our on- going Federal-State systems in specific areas and we know that we cannot coordinate 50 sovereign jurisdictions easily or quickly. But if we are practical in our objectives as we work on the ongoing and developing projects, we can gradually build up the needed components. But first, we need Federal Government recognition of its leader- ship role and, along with that, more State and local acceptance of the importance of statistics in their planning and administrative operations. Without such executive acceptance, the achievement of a Federal-State- local cooperative network for data collection will be very difficult to achieve. The development of such executive acceptance is an activity to which statisticians interested in small-area data must dedicate themselves. COMMENTS ON PAPERS ON SMALL-AREA STATISTICS Daniel B. Rathbun, Executive Director, President's Commission on Federal Statistics I want to acknowledge the indebtedness of the Com- mission to Messrs. Alfasso, Ullman, Steger, and Mindlin. At the same time, I must point out that there is no necessary connection between the positions that have been taken and the positions to be taken by the Com- mission, just as there is no necessary relationship between my remarks and the Commission's position. The Commission recognizes that the small-area data problem is one of the most important problems it will address. I suspect that the Commission will make five broad recommendations, one of which will deal with small-area data problems. Because of the Commission's keen interest in these problems and the desire of all members to improve upon existing arrangements, the preceding papers were commissioned and the views of the American Public Health Association, the American Statistical Association, the American Institute of Planners, the Urban Institute, the Council of State Governments, and the Advisory Com- mission on Intergovernmental Relations were solicited. Because of a feeling that the major problem is to make more effective use of existing data, Robert Barraclough was asked to write a paper on the potential of geocoding. In addition, officials in Federal agencies with special knowledge with these problems were asked to present their views. Included were Conrad Taeuber, Robert Voight, Robert Klove, Caby Smith, George Farnsworth, and John Deshaies of the Bureau of the Census; Walter Keim of the Bureau of Labor Statistics; Theodore Woolsey of the National Center for Health Statistics; and Paul Krueger and Lawrence Bloomberg of the Office of Management and Budget. The preceding speakers have suggested that a major change in the role of the Federal Government in providing small-area data is called for. They, particularly Mr. Mindlin, offer what they describe as gaps. Claims that these gaps exist have been made so frequently, have been based on such desirable objectives (e.g., to solve the troublesome problems of urban life), and the claimants have been so persuasive that many influential members of Congress and the Executive branches have been convinced of the existence of gaps and have devoted energy to the task of trying to persuade the Federal Government to fill them. Many additional statistical projects have been instituted in recent years (e.g., USAC and SCRIS); but spokesmen for local statistics are convinced that the surface of the problem has not been scratched. Some of the assertions seem unwarranted. The first is that the data needed are standardized statistics- standardized in the sense that they are derived from Nationwide surveys or report systems using standardized definitions and procedures taken at uniform dates or for uniform periods and disseminated in single publications or series of publications. A second assertion open to question is that such data can be provided by the Federal Government and should be a part of the Federal statistical system. This assertion is often accompanied by the observation that many of the data at issue are required to be reported by State and local officials in support of applications for Federal program assistance and grants. These assertions and associated recommendations have an air of reasonableness about them but one need only start listing the measures asked for to experience misgivings. Two things stand out in these assertions. First, the Federal Government should not only provide desired data, but should bear all or a large part of the cost. Second, there is reason to believe that the data that would be demanded if problems were thought through would not embody nationally standardized definitions and procedures and would not be wanted for nationally uniform dates or periods. In a large proportion of cases the relevant data would be tailored to specific conditions in specific States or other small areas with very special awareness of local singularities such as sources of municipal revenues, skills of the labor force, and journey to work. The proposal that the Federal statistical system provide the data seems to derive in part from desperation to "find a number," rather than from a clear understanding of the relationship that should exist between the analytical questions to be asked and measures needed to answer them. Public health workers appear to have firm analytic models in mind. However, the data they ask for would require the expenditure of vast sums. The public health demands seem to be basically for popu- lation numbers to provide denominators for indices of a great variety of health ratios. Superficially, this may appear to be only a demand for frequent population counts for small areas, but this appearance is deceptive. The advanced state of epidemiology and related health analysis is such that a veritable multitude of detailed character- istics of the population, and of the environment, is wanted- - including such things as socioeconomic status, housing conditions, employment history, food resources, water and sewer services. One may question whether generalized statistical vehicles should be looked to for such data or whether ad hoc vehicles designed for the job would not be both cheaper and more effective. It would seem that the experimental design underlying the demand for data is serendipitous- -the hope that if enough numbers can just be examined, some fortunate discoveries surely will be made. The bulk of the demands that gaps in small-area statistics be filled falls between these two. In large measure, they rest on incomplete analytical bases and imply or state a need for a multiplicity of highly refined data, usually with great frequency. An example of a 31 32 STRENGTHENING THE FEDERAL ROLE stated need is for demographic data keyed to place of work in addition to residence in order to give us an accurate picture of the demands for public services in urban centers. What kinds of public services? What demographic characteristics are relevant? What do we know of the relationship between these two? And above all, is this a Federal responsibility? Considering the great differences among areas in which suburban resi- dence and central city place of work is important, is this set of issues not one in which local understanding of the situation and the problems calls for local analysis and tapping of local sources of information? A gap frequently mentioned is timeliness, interpreted both as speed of publication and frequency of surveying or reporting. Clearly, where uses are time-dependent, data should be published as promptly as possible. At the same time, it is necessary to recognize that the more complex the data collected, the greater the delay will be. In the little time I have remaining I will simply comment on some uses and ask a few questions. We know too little about the real demand for small- area data. I am told that both the Decennial Census Review Committee and the National Research Council Advisory Committee on Census Enumeration were unable to obtain from the Census Bureau the examples they sought of local uces to which such data are put. The argument that each and every locality needs data in order to evaluate the Federal programs is not convincing. After all, the Federal Government can try out a program like Head Start in a controlled field study and, if it works, simply launch it in many communities. In fact, local users know very little about the non- sampling errors of census tract and block data. Without knowledge of this component of error I doubt that they know that census data are superior to local data. Apart from this, the significance of timeliness, in both senses, is not well understood. For example, apart from the public health uses, the need for timeliness is usually adduced with respect to expensive, time- consuming projects such as urban renewal or location of medical service complexes. This seems unreasonable. If changes that occur between surveys (or in the surveyed characteristics within the period that publication may be delayed) are critical to the location or magnitude of such ventures, it seems unlikely that such projects can ever be located or dimensioned rationally. More likely, perhaps, is that the illustrations are spurious--used because their cost is impressive rather than because the data are relevant. In addition to failure to state problems clearly and to specify pertinent data wanted, claims of gaps appear to rest on a belief that local agencies must, perforce, be users of data, and that the only data they can use are generalized Federal data. Examples of gaps cited in submissions to the Commission suggest, in a large proportion of cases, both that the local users must become producers of data they need, and that in a great many cases, the data they need are not generalized data but data designed to their specifications. I submit that the evidence suggests that much more attention should be paid to exploiting local sources, as opposed to efforts to make the Federal Government do what it is poorly equipped to do. Of course, there is need for more effective co- ordination of the activities of Federal agencies and the Commission can point this out. Mr. Mindlin asked for employment data by occupation. I am under the impression that OBE uses a 1-percent sample to develop such data for counties and that BLS has plans to develop such data for over 200 areas. (It was pointed out that the OBE data is by industry, not by occupation.) Mr. Mindlin asks for data by place of work and place of residence. The census provides these data for 1970. Is it appropriate to provide them on an annual basis? When we add the requests for data on incomes, on occupation, on place of work, the total is very great. And when we consider that the Federal Government is being asked to educate people so they will understand the realities of statistical life, and the Federal Government is asked to develop methodology for social indicators at the local level, and the Federal Government is asked to develop a model of housing quality, one could conclude that the Federal Government is being asked to solve problems that have not been solved at any level. In a semiserious vein, I want to observe that such requests cause me to think that people like Mr. Mindlin have, in the unwillingness of Congress and Federal agencies to respond to these kinds of requests, the greatest protection in the world. If Congress and the Federal agencies were to respond and this avalanche of data were to be delivered, then Mr. Mindlin would have to do something with it. Small-Area Statistics in Criminal Justice -- Activities of the Law Enforcement Assistance Administration STATISTICS DIVISION SUPPORT FOR CRIMINAL JUSTICE PROGRAMS Patti Rowan Law Enforcement Assistance Administration The Statistics Division of the Law Enforcement Assistance Administration (LEAA) awards grants to State and local governments, institutions of higher learning, and nonprofit organizations to collect, evaluate, and disseminate data on the condition and progress of criminal justice in the United States. Because LEAA is mainly concerned with assisting State Governments, the majority of the grant programs of the Statistics Division are devoted to the coordination of criminal justice statistics at the State level. Through a program of discretionary grants, the Statistics Division encourages each State to establish a central office for the reporting of uniform crime report data. New developments in the collection of these data are also being encouraged and funded. These include the gathering of sample data on nonreporting jurisdictions to supplement the data from reporting police agencies within the State. Another State-oriented program has been established to develop information about the criminal justice process as it relates to the criminal suspect or offender. This is being done through statewide transaction statistics systems. Since the basic unit common to all justice processes is the offender, the method of transaction statistics tracks the offender as he passes through the criminal justice system and records the pertinent data for each transaction. By this method it is possible to examine the entire process of the criminal justice system of a State and gain extensive information on time delays, recirculation rates, and fallout rates. The transaction statistics method will also allow the collection of traditional criminal justice agency summary statistics. In addition to the State programs, the Division is investigating methods of supplying statistical data on various aspects of crime and criminal justice as they apply nationally. One aspect of special interest, because of its recently recognized impact on the criminal justice system, is probation. The trend toward increasing use of probation as an alternative to incarceration has led to the need for information on probation programs. Through a grant awarded to the National Council on Crime and Delinquency, a project using the San Francisco Bay area probation departments as a model will be undertaken to assess the data needs of probation officials and to develop the methodology to collect these data. In addition to State and national data collection programs, the Division provides statistical support for various agency activities. For example, we collect both primary and secondary data to provide planning or evaluative insight into various agency programs. It might be enlightening to examine one of these programs in more detail. A year or two ago, a number of demonstration projects were established in medium- size cities. The purpose of these so-called "pilot city" projects has been to institute a broad array of procedures to improve the effectiveness and the quality of justice in the cities. We were asked to provide the data to permit an objective program evaluation. Since this effort was not a controlled experiment in the usual sense, and since the goals could not be quantified because of the program mix, we felt that we could test the utility of the programs if we could measure both changes in the crime level and changes in the attitudes of the citizenry. A large sample survey was designed for both of these cities. Each survey consisted of about 5,000 households and a smaller number of commercial establishments. The first of these surveys was conducted in January in both San Jose, Calif., and Dayton, Ohio. Two years hence, these surveys will again be conducted to measure changes in the crime level, crime patterns, and in the public perception of crime and attitudes toward criminal justice agencies. Though the results of these surveys are not yet available, a discussion of possibilities for analysis follows. The President's Commission on Law Enforcement and the Administration of Justice in 1967 commissioned a number of surveys to determine the incidence of crime and public attitudes toward it, both at the national level and in a few selected cities. One of the task forces of the Commission drew the following conclusions from the findings of the 1967 surveys on attitudes toward crime: The first is that the public fears most the crimes that occur least. Second, the fear of crimes of violence is not a simple fear of injury or death or even of all crimes of violence, but, at bottom, a fear of strangers. Third, this fear of strangers has greatly impoverished the lives of many Americans, especially those who live in high-crime neigh- borhoods in large cities. 37 38 CRIMINAL JUSTICE STATISTICS Fourth, the fear of crime may not be as strongly influenced by the actual incidence of crime as it is by other experiences with the crime problem generally. 1 Other observations and theories on the fear of crime will be tested against the data in these surveys. The Bureau of Social Science Research, after conducting its surveys on crime for the President's Commission, wondered, with good reason; "Is it possible that victimi- zation is kept at the level at which it is only through constrictions of life activities of respondents — staying home at night, not venturing into parks, installing stouter locks, moving to a 'better' neighborhood and so forth." 2 A vociferous advocate of city life, Jane Jacobs, believes; "It does not take many incidents of violence to make people fear the streets. And as they fear them, they use them less, which makes the streets still more unsafe." 3 By means of the series of attitude questions included in the surveys, we hope to probe and test these con- clusions and theories. To gather data on a general attitude toward crime, the fear of crime, and a change in life style due to crime, we have posed questions such as: Why did you select this particular neighborhood in which to live? If you have changed residences recently, what were the factors you considered in making this move? Is there anything about your present neighborhood which you dislike? Do you go out in the evening for entertainment? Do you go out more or less than a year or two ago? If more or less, why? How safe do you feel in your own neighborhood? How do you think your neighborhood compares to others in the area in regard to crime? Are there some parts of the metropolitan area where you feel unsafe? With these and other like questions we hope to determine the effect crime, or the fear of it, has on people's daily lives. Does crime or fear of it influence the decision of where a person shops? Will the resulting data show that a person feels safest in his own neigh- borhood even when that neighborhood is, in fact, a high-crime neighborhood? We also hope to gain some insight into claims that segments of the population have ceased to attend social and cultural events because of widespread fear of venturing forth into the city in the evening. Are fewer people attending these kinds of events? Have the suburbs supplanted the city in being the preferred place of entertainment? Has this fear of crime precipitated a downturn in the business activity of the city? Is there actually a lessening of sociability and mutual trust among 'President's Commission on Law Enforcement and Admin- istration of Justice, Task Force Report: Crime and Its Impact — An Assessment . Washington, D.C : U.S. Govt. Print. Off., 1967, pp» 88, 89. Hereinafter referred to as Assessment Task Force. 2 Albert D. Biderman, Louise A. Johnson, Jennie Mcln- tyre, and Adrianne W. Weir, Report on a Pilot Study in the District of Columbia on Victimization and Attitudes Toward Law Enforcement . Field Survey I, President's Com- mission on Law Enforcement and Administration of Justice, Washington, D.C: U.S. Govt. Print. Off., 1967, p. 159. 3 Jane Jacobs, Death and Life of Great American Cities . New York: Random House, 1961, p. 40. members of the community? How does the individual reconcile his desire for freedom and his perceived need for security? If it is true that people feel safe in their own neigh- borhoods but experience great fear of crime, why does this fear exist? Possibly the results of these surveys may indicate that this fear is based on media accounts of crime, acquaintances' experiences with crime, or simply a lack of familiarity with other areas of the city or metropolitan area. In addition, with the data gathered in the San Jose and Dayton surveys, an index of anxiety might be developed, as was constructed from the President's Commission surveys, combining concepts of security against crime as a criterion for a place to live, perception of crime in one's own neighborhood, desire to move to a safer neighborhood or out of the city altogether, and curtailment of activities because of the fear of crime. Does anxiety about crime vary widely by sex, by race? Does the presence of children in the household seem to have an effect on the respondent's perception of crime and anxiety about it? Does having been a victim of crime increase one's anxiety about crime or does it seem to have the opposite effect? Does this differ by socioeconomic characteristics of the respondent, of the neighborhood? Will we find a significant change in people's anxiety about crime? In addition to assessing a change in life style due to crime, an attempt is being made to assess the public's view toward the police and compare that attitude after a 2-year interval. The respondents were asked if they felt the police were doing a good job; also if they could suggest ways in which the police might improve. These attitudes toward the police will be correlated with the respondent's experience with crime and his attitude toward it, as well as his socioeconomic char- acteristics. Do attitudes toward police vary by sex, race, and/or income of the respondent? The data may show that a person's attitude toward the police has a close relationship with his anxiety about crime. We may find that anxiety about crime indeed correlates with a favorable attitude toward the police while, on the other hand, we may find that experience as a victim correlates highly with an unfavorable attitude. Will these attitudes change over time? We hope to reach beyond these surface data to attempt to determine if people who feel great anxiety about crime tend to be more prone to accept, or even advocate, stronger measures of law enforcement administered by the police. Will people willingly relinquish some part of their freedom in return for security? How much? Other survey questions were designed to arrive at the respondent's opinion of the problem of crime and his knowledge of its patterns. Does he believe that crimes of violence usually occur between strangers or between people who are acquainted? Does he feel that the media is portraying a fair picture of crime? Finally, does the respondent think that people limit or change their activities because of crime? Has he changed or limited his? This last question is of particular interest in light of a statement made in a report of the President's Commission: "A further indication of the importance of vicarious impressions in forming the public's perception of crime is that a majority of citizens, almost every- where, think that the situation where they live is not so SUPPORT CRIMINAL JUSTICE PROGRAMS 39 bad. While the predominant opinion is that the crime The attitude and crime incidence data gathered in these situation is terrible and getting worse, most people tend surveys will be used not only for evaluation of programs, to think of the situation as one that characterizes places but also as base data for research. other than their own immediate neighborhood." 4 As LEAA's program expands in cities and other small areas, we expect that the Statistics Division will be called upon more and more frequently to provide the 4 Assessment Task Force, op. cit., p. 87. data of the type I have jUSt discussed. SAN JOSE PILOT CITY CRIMINAL JUSTICE STATISTICAL INFORMATION SYSTEM Paul K. Wormeli, Vice President Public Systems, Inc. and Robert L. Marx, Vice President Public Systems, Inc. One of the major problems in all of the programs undertaken by the Law Enforcement Assistance Adminis- tration is a lack of data with which to describe or analyze the criminal justice systems in the country and to evaluate the impact of changes made as a result of LEAA support. This problem is particularly critical in the pilot cities program. In general, the pilot cities program of the National Institute of Law Enforcement and Criminal Justice in LEAA seeks to work with individual juris- dictions with a strategy of (1) identifying the weaknesses of the criminal justice system in each jurisdiction; (2) instituting the substantive changes to improve the opera- tion of the criminal justice system; (3) evaluating the results of these changes so that the successful changes can be transferred to other jurisdictions. Obviously, there is a basic need for quantitative information to assist in accomplishing these tasks. The two basic data requirements must support the diagnostic aspect of the program (to identify weaknesses of the system) and the evaluation component. In general, the state of the data in criminal justice is so poor that we are unable to rely on existing data collection for information systems to assist in either of these tasks. Even the basic data on the number of crimes that have occurred are quite unreliable. There are so many inconsistencies in reporting and gaps in data collection, coupled with ambiguities in classification of events, such that accurate measures of the occurrence of the various crime categories are not available. Sub- sequent actions taken by the criminal justice system may be better documented in individual case files, but most States are just beginning to develop statistical systems which record the transactions occurring between the criminal justice systerr and the offender. The work of the statistics division of LEAA is primarily con- centrating on this important irea. The development of improved statistical systems is proceeding slowly, and the pilot cities program in general cannot wait for the final development of idealized systems. If there is to be a way of determining the effects of changes made, there must be a data system in being now so that "before and after" statements can be made about the operation cf the criminal justice system. As a result of this current requirement, many pilot cities have created a baseline data system which can be used to describe and analyze the existing criminal justice 40 system and which will support additional measurements of the system performance as the various programs under the pilot cities projects are implemented. A baseline data system consists of a number of component data bases, where the components are ordered along the processes and events carried out by the criminal justice system. For discussion purposes, we can define the components of the baseline data system in four major areas. These are (1) crime occurrences, (2) police response, (3) judicial processing, and (4) cor- rectional processing. In each of these major categories, a data system must be created that allows the collection of information in the kind of an input-output conceptual- ization. That is, the events occurring as workload in the various processing categories must be counted, the output (in terms of system performance) must be measured, and the processing times must be recorded. In creating a baseline system, data modules must be established for each of these areas, providing ways to link the processes together so that the total criminal justice system output can be described. It turns out that several modules are required in each of these areas. In measuring the incidence of crime, it is necessary to account for both the unreported crime and that which is reported to the police department. LEAA is presently developing a series of victimization surveys with the assistance of the Bureau of the Census to develop statistical systems which describe the unreported crime. Ideally, the results of these surveys could be used to calibrate the reported crime data available to the police. In the area of police response, the data base describing the handling of events is required, and a separate module is necessary to describe case disposition. Similarly, in the area of judicial processing, systems are required to describe the intake in pretrial processing as well as the trial and sentencing operations. In the correctional area, both institutional and community supervision pro- grams must be treated. The overall intent is to obtain data to evaluate each agency in the criminal justice system with respect to the goals of the total system as well as the specific agency goals. Since much of the LEAA program and the pilot cities program must eventually deal with the problem of crime reduction, a great deal of emphasis in the baseline data SAN JOSE PILOT CITY PROGRAM 41 systems is on reliable statistical systems for measuring the occurrence of crime. In the remainder of this discussion we will describe some of the work that has been carried out in San Jose, California to produce the reported crime module of the baseline data system. The unreported crime in San Jose has not been ignored, in that the pilot testing of the Bureau of the Census victimization series was carried out in three of the pilot cities, including San Jose. Some preliminary data will be available from the survey to calibrate police data collection and analysis. The original goal of the San Jose baseline experiment was rather broad; to obtain a precise quantitative description of the operations of the patrol function with- in the San Jose Police Department, together with its command control and recordkeeping support functions. In achieving this goal, the baseline team was to have complete access to the records and forms presently kept by the police department, but was not allowed to institute any new data collection procedures, or to redesign report forms, or to engage in personal inter- views with police officers involved in the patrol or supporting functions. The first chart (fig. 1) shows in simplified form the sequence of events beginning at the reception of a call for service from a citizen at the telephone complaint console, the delivery of required information to the dispatch function at which the appropriate mobile police unit is chosen and dispatched, the rendering of service by the patrol division, and the return of the entire function to the "ready" state for another call. Two major data sources are available to collect the types of information needed for a description of the patrol function at its supporting functions. First, a complaint card (fig. 2) is generated each time a call is received from a citizen. At the complaint desk, this card is time-stamped with the time of reception of the call. The type of service requested by a citizen is recorded in a coded form corresponding to radio brevity codes in use, and the location of the call- -either a street address or intersection- -is recorded. Finally, through a table lookup procedure, the police needs for patrol area in which the service is requested are recorded on the complaint card. Next the complaint card is passed via conveyer belt to the dispatch console, an appropriate available mobile unit is chosen and dispatched, and the time of dispatch is time-stamped on the complaint card. The mobile unit proceeds to the scene and at point of arrival, reports arrival by radio to the dispatcher, who time- stamps the complaint card accordingly. At the completion of the required service, the mobile unit reports completion of service to the dispatcher who time-stamps the complaint card with the time of completion of service. The second major source of data is the offense re- port (fig. 3) which is completed by the officer responding to the call in those cases in which followup investigation may be required or where there is reason to believe that an actual offense occurred. This occurs in approximately one-tenth of all calls for service. The offense report form consists essentially of a cuing or coaching form for the preparation of a narrative report. In addition to some repetition of information collected earlier on the complaint card, several new elements of data appropriate for the description of patrol division operations becomes available through the offense report. Figure 1: DISPATCH LOGIC FLOW QUE Figure 4 shows the types of information available and collected during the baseline experiment being described here. Details of the offense itself, the premises at which the offense occurred, of the prior realtionship, if any, between the offender and victim are available as well as the more common data concerning age, sex, race, and alleged penal code violation reported. During October and November of 1970, a pretest of the baseline data collection scheme was conducted in San Jose. The goal of this pretest was threefold. First, we wanted to determine that linkages could be established between complaint cards and offense reports arising from those complaints without requiring the assignment of a case number at the time of complaint reception; this would have been a deviation from our overall con- straint of not requiring new reporting elements from the department. We established that by using time of call receipt, patrol area in which the call originated, and type of service requested we were able to make this linkage effectively. Second, we wanted to determine the adequacy of the coding structure we had developed to cope with the 42 CRIMINAL JUSTICE STATISTICS occurrences actually reported within the department. The basic coding structure followed the lines of previous work performed at the California Bureau of Criminal Statistics, modified to meet the special needs of the baseline experiment in San Jose. With the exception of a few minor modifications in the initial coding structure which were accomplished about halfway through the pretest, we found that we had both sufficient precision and sufficient scope to cope with the real data encountered on the reports. average hourly labor rate of $3 and a workload of ap- proximately 25,000 complaint cards and 2,500 offense reports per month from the department, the cost of data collection and conversion, including keypunching were determined to be approximately $1,300. The results and conclusions of this pretest experi- ment were published in December 1970, and we prepared for operational ongoing conduct of the baseline data collection plan. At this point, for reasons totally ex- ternal to the project itself, funds were severely curtailed Figure 2: COMPLAINT CARD CAR DETAILED NATURE OF COMPLAINT (USE RAttO CODE NUMBERS) REGARDING LOCATION OF COMPLAINT D AMBULANCE NEEDED □ AMBULANCE OfSPATCHED DATE- TIME RECEIVEO DATE — TIME BROADCAST OATE- TIME CLEARED COMPLAINANT^ NAME Dc LZJ OO NOT CONTACT VEHICLE COLOR DIRECTION TRAVEL N S E W DETAILS OF COMPLAINT NATURE OF COMPLAINT liu iii : siouxasi H COM PL. OPERATOR BEAT CAR i ) « i i ) i i h ii ii i) u a ti n ii it a ji n n u a it v a a a n n n m s s » » a « ii u u h u u « a u h h u u « a a u s* u u ti o a u a a n a a >i n n n n n it n n n a DETECTIVES NOTiriED 1 .JUVENILE NOTIFIED 1 DEPARTMENT NOTIFIED DESCRIPTIO N OF SUSPECTS SEX RACE AGE SEX RACE AGE HOT WGT. BUILD HGT WGT BUILD COMPL HAIR EYES COMPL HAIR EYES GLASSES MUSTACHE GLASSES MUSTACHE HAT COAT HAT COAT ADDITIONAL DETAILS: PANTS SWEATER PANTS SWEATER SHOES SHOES CSCM»t-7»-UM Finally, we wanted to find out the level of skill required for the coding of information from the complaint cards on offense reports and the costs per report processed. Beginning with trained analysts with signif- icant law enforcement experience, we gradually lowered the skill level of the coders, finally utilizing clerical personnel from within the records and communications sections of the police department. We found that clerical personnel could process approximately 125 complaint cards or 25 offense reports per hour. At an and we were forced to reduce our goals accordingly. We dropped all activity in the area of complaint card data collection and changed our goal to the quantitative description and profile of reported crime in San Jose. The data collected from offense reports remained es- sentially unchanged from the pretest form. Since January 1971, we have collected, coded, analyzed, and displayed data from every offense report generated within the San Jose Police Department, a total of approximately 15,000 reports. SAN JOSE PILOT CITY PROGRAM 43 Up to this point, I have discussed only the collection of data for the baseline experiment. The first, and by far the most expensive and time-consuming portion of the processing and analysis of the collected data related to the determination of incident location of a form suitable for computer manipulation and analysis. There is avail- able in San Jose a computer compatible file, known as the property location index or P LI, on which are recorded, among other things, the address and map coordinates (to within 1 foot) of each tax assessment parcel within the Figure 3: OFFENSE REPORT BEA1 NO RECORDING IimE TYPE OF CRIME: OFFENSE REPORT SAN JOSt POLICE DEPA«TMENT RECORD NO 1RANSCRIBING Utrt OATE OF REPORT: Dole ond lime occurred do, ol -eel locOtion ol Occurrence Oole ond time reported to depo. men, v,cl,ms nome IFirm nome ,1 business! Residence oddress (busin ess Oddress .1 l.rm) Residence Phone Business Phone or Address Person reporting oHense Residence Address Residence Phone Business Phone or Address erson - o discovered crime Residence Address Residence Phone Business Phone or Address Witnesses Nome Residence Address Residence Phone Business Phone or Address Victim's Occupolion Roce Se. Age ype premises or oco ion - ere o ense -os commi e Point -here enl.once -os mode CHIMES AGAINST PtOPHTY Weopon Force or moons used CIIMES AGAINST THE PEtSON E.od locohon ol prope.tr -hen stolen E.od locotion ol .iclim ot lime ol o«ense Instrument used (describe! Victims oclivily ol lime ol often, e Method used to goin entronce E.od words used by suspect Where -os occuponl ol lime ol oHense Hos detendont been od.ised ol h,s rights' Properly booked"' Apporenl mot>*« type of pfoperty token Of obtained Irodemorlt of suspect|s) Actions ehide used by suspect(s) oke, body lype, color, license number, ond on, other identity. ng marks SUSPECTS »MD CtftSONS ARRESTED NAME ADDRESS COMPL. SE. „.,» lr E! ' WEIGHT ACE ID.O.B. OR l.O. NO. ARRIS TED YE S Ml DETAILS Describe evidence Summonre delo-ls not given abov ond describe ony property obtained, including serial numbers ond volue Invesl.aoling OH.c Luf obov», cos* numbers ot ony ofher offenses cleared or connected with this reporl Use Form 200-3 il necessor fo«m 212 oo. 2) ,/7, RECORDS CUPY 44 CRIMINAL JUSTICE STATISTICS city. This was used as the basic data file for the conversion of address and intersection locations recorded on reports to map coordinates for further analysis. Since not every street address constitutes a tax parcel, and since tax parcels are recorded only as addresses and not as intersections, methods for interpolation and extrapolation of map coordinates as a function of street address had to be provided. In addition, a software method for determining the map coordinates of inter- sections by mathematical manipulation of address infor- mation contained in the PLI was developed. Extensive editing software also had to be developed to cope with the large number of misspelled street names, trans- position of characters in addresses, and "impossible" intersections recorded on the forms. Manual editing has been required for final cleanup, although this re- quirement has been diminishing as report writers have become aware that someone is really looking at the information in detail. interpretation of "hot spots" within a city but tends to lump in greatly dissimilar areas, if, for example, a major freeway divides an individual 1-mile cell. The third major display format developed responds to the need to "zoom in" on portions of the city to determine the incidence of offenses with the greatest possible geographic resolution. In this third display format, a computer-driven plotter places points at the geographical coordinates specified by the cards being run. By way of illustration of these three formats and the way we are using them, let me briefly walk through a hypothetical example, using actual data from April 1971. The type 1 format (fig. 5) is generally run with the offense breakdown shown here; as usual we find that half of the offenses reported are burglaries. Also, as usual in San Jose, we find that the beat 17 area is a major contrib- utor to this burglary incidence. Beginning here, the Figure 4: INFORMATION CATEGORIES, OFFENSE REPORTS CASE NUMBER BEAT OFFENSE date/time OCCURRED DATE REPORTED TIME REPORTED DISCOVERER STREET SUFFIX DIRECTION HOUSE NUMBER PREMISE WHERE OFFENSE OCCURRED LEVEL OF VIOLENCE — PERSON TYPE OF WEAPON USED LEVEL OF FORCE CRIME TARGET — PROPERTY SERIAL-NUMBERED ITEMS VALUE OF PROPERTY STOLEN NUMBER OF OFFENDERS INVOLVED SEX-RACE (OFFENDER) SUSPECT PRESENT POLICE DISPOSITION VEHICLE (OFFENDER) AGE (VICTIM) SEX-RACE (VICTIM) ROLE OF VICTIM COMPLIANCE OF VICTIM VICTIM-OFFENDER RELATIONSHIP ALCOHOL INVOLVEMENT DRUG INVOLVEMENT The result of this processing is a deck of cards containing the data elements described earlier taken from the offense reports with the exception that incident location is in the form of XY coordinates. The primary analysis technique used to date has been to provide three alternative display formats and to allow the in- dividual users to specify the input parameters for the data to be displayed. Briefly, the first output format consists of a tabular array of incident types by patrol area (fig. 5). Row and column totals are provided, and either percentages or absolute numbers can be displayed. Since patrol areas or beats are highly irregular in size, and in fact are ordinarily designed to provide roughly equivalent workloads per beat, the second display was provided in which incidents are "blocked" into cells of equal size (fig. 6). The size most often used in our displays to date has been a 1-mile by 1- mile square grid. This has the advantage of easier analyst switches to format 2 (fig. 6) in order to begin getting a clearer picture of the problem. First he specifies that only residential burglaries be considered, setting aside burglaries from automobiles, from business establishments, phone booths, etc. He finds that resi- dential burglaries comprise approximately 50 percent of the total, and that again they are concentrated in the general area of beat 17. He may make several more inquiries of this type in sequence, gradually narrowing the parameters to gain more insight into the problem. For example, he may look at only burglaries which occur on weekends or only those which occur during the day (during which professional burglary rings are most active), or only those which occurred in multifamily residences. Having gained as much insight as he can from the data presented in format 2, he might switch to format 3 to get an even clearer picture of what is going on. For SAN JOSE PILOT CITY PROGRAM 45 O O •-• =T in "O if) Ot-3" i_ cm -n ■-< a- a- * Z -I UJ < UlLh __ or o O o ^ in — < cm HjHrtonfl •-'o oo Jin^ o. 10 O O CM CO r~ CM —i o> ?K\ or ui «h * ;T mfl UI CO -H OQ U. 2 s: O ui o co vo o «h -h > a' r~ o to cm voo mooo>io a - '-' -h co r- 3U. in o — < a' a - r~ cm >-< cm r- -h -h to o o -^ r\l r\i o^a- \o Z u. to co to — i a- o^ o O omfti H 03 vd-ho-0 a- 0000003-^10000 tO-n^OOOOOOOO-H ifl ON(MO 3" ^ into CT> CM to eg i\i to CM OOOOOOC0O3-OOO Ot<"\O'-'OC\Jh cm a- UJ 0._ r- OOOOOO'-'OOOOO vO^OOOOOOOOOO (\IOhho 'HO l> »H CM -~i C/1 <\i -H oooHHono'iMON a-ooooooooooo ? on^ ■-< co cm a- ^ o -* •— I y cm cm to in -_ m CM to ooo^-'oO'-'Cmiocmocm ocm— 'ooocmocmO'^o h im a- in o ^oa- to a- Z ,_■ i cm -n a- cm -< :*■ < rt .-H -H t/i _7 O Ol OOOCMOOa-*c\I-HOO JHHOO-n^flOOlMo OOOino r~CM >k> a- IjXcm cm «-< cMin uj oo o -i- w_ O >h ooo(MoHco H 3'flo(M r-a-cM-nocMa-oa-o-H-H oo^ao in vo vo in a- in .■?"!.• CM rH H tO «"l CM ST 0* o ooooooaocMoocM voa-o— 'ootootooo— ' co o a- to — i CM CM o in o rO sO — < — i to r- to cm ^ vo cm r-- «h •h ^ CM I— CC t- co ooocM-Hor-'O v O' H oo iMrtooniHHHooflo coor-too oin o in co a' ~i-i-<' H • H * cm a- co OQ ^h ? uj _. i>_ Q_ cot~ oooincMocMa-o^^oto — iinoootOvoa-— ioo*-t — < o a- vo -< >-• o cm-- a-- r- i>_.|jJ-H (M nO CM IOvOIO O zl r\ "* ~L '~ i ■— ■ vo ooooooao-ioocM vOin^ooooooo«-Hrtoo^ r~ooooovOin— io -h OC^O^ w rt CMCO QC — > Du cj ooo^o— •— iN>oooh oHrtoooJiMHoiMH toot-t-^ toto >-ito a' a- 3 UI -H ^-(CM «-t to t-- r* 3- ooovoooc- H r--iO'-io h^woohiimmhoho oohoo oa- t>in o^ a- LU — < — i a' <-i sf O s *5 o ooo-- 1 — lotooinoo*-' r- , -''- < ooooooo'-io i^— itocM-n ina - ocm cj^ cm CO — -h rt-n ^H^ia- UJ •" ,-, o^ oooooo^— 'inoO'-< — i--h j jo a- to o to o a- LO >H CM CM -H CM CO Q^ in ooocmooco— 'a-oo— i i-cohoocm(\|hhoho inoa-ino >-iio coto vo to ^ «h -h cm in 3 6J) ST 000- , OOr--3 a-a - to a- -i to «h a- co CM OOO-^OOCOIOfO-HOO Oa-CMO-'O'-'OOOOO CMOCMCMO OCM iO ? fO IO cm — i to in —i oootototO"-h a-'-ia - --' <-H K U CO UJh-> (-UJLJS UI H-h-OIQT c/)>UOS>-" c/)t- co > u z dqi- ui i u u.i-i m u > o u Oco< u. ui aui_i>-ujct:i- ui co o; r>- >-i -j ocol. t- zcq us cosct: i-SQ: uiui i-Ox u. mqioo: to _i i- i- < (-or > mo u 2 o£i-i 2 d u qcdui oicom z ddih: ozo. uizq. ujz ui_i< O «D