assess in quantitative terms the consequences on crime of virtually any action taken to control it. But without such data, how can we begin to decide what programs to undertake and how best to expend our limited resources?

Sherlock Holmes, with his usual insight has warned us: "It is a capital mistake to theorize before one has data." It is about time to introduce some capital and talent to reduce the cost in crime of these mistakes.

The fact that we so often think of Holmes or other real or fictional investigators when we think of crime reflects one crucial reason for establishing a statistical data base. Since our journalists and novelists bother to tell us only of the ghastly crimes, the brilliant detective investigations, the dramatic courtroom confrontations and the parolees who go berserk, we come to view these as typical. It takes conscious effort on our part to remind ourselves that the images we get from newspapers, novels, and magazines are very unrepresentative.

In the United States, in 1966, there were reported to the police over 3.2 million "index" offenses of willful homicide, forcible rape, aggravated assault, robbery, burglary, larceny of $50 or over, and auto theft. The newspapers can report on only a small fraction of these.

Murder and forcible rape represent less than 1.2 percent of these "index" crimes. And most of those were probably between relatives or acquaintances.

In California, over three-quarters of the criminal felony convictions are resolved by guilty pleas rather than by a courtroom trial. In some jurisdictions, this rate is even higher.

Arrest, rather than being a rare phenomenon, is probably much more common than is generally recognized. If current trends continue, perhaps half the boys growing up in the United States today can expect to be arrested some time in their lives for a nontraffic offense. Even the officials within criminal justice agencies find it extremely difficult to think of crime and actions in representative rather than anecdotal terms without adequate data.

One of the major functions of a statistics center is the establishment of uniform reporting standards, both horizontally across jurisdictions, and vertically from one part of the criminal justice system to another. Because of the inherent complexity of criminal behavior, and basic inconsistencies of criminal codes, the same act may be reported differently in different jurisdictions. For example, in California, taking valuables from a locked car is burglary, but in Virginia it is classified as larceny. Similarly, in trying to follow the number of people through a criminal justice process, it is very difficult to keep track of the number of individuals involved. For instance, while police and corrections usually tend to count individuals, courts count cases. One case may account for several individuals, or several individuals may be represented in one case. Another important improvement would be the establishment of consistency among the various existing reporting systems, and making these consistent with census data.

The assessment task force report of the National Crime Commission has listed the kinds of roles for a Federal Criminal Justice Statistics Center. These are still completely applicable.

(1) Inform the public and responsible governmental officials as to the nature of the crime problem, its magnitude, and its trend

over time both throughout the Nation and within individual localities.

(2) Measure the effects of prevention and deterrence programs. (3) Find out who commits crimes by age, sex, family status, income, ethnic and residential background, and other social attributes in order to find the proper focus for crime prevention programs.

(4) Measure the workload on the police, the courts, and the other agencies of the criminal justice system, both individually and operating as interacting components of a total system.

(5) Analyze the factors contributing to success and failure of probation, parole, and other correctional alternatives for various kinds of offenders.

(6) Provide criminal justice agencies with comparative norms of performance.

(7) Furnish baseline data for research.

(8) Compute the costs of crime in terms of economic injury inflicted upon communities and individuals, as well as assessing the distribution over crimes and agencies of the direct public expenditures by criminal justice agencies.

(9) Project expected crime rates and their consequences into the future for Government planning.

(10) Assess the societal and other causes of crime and develop and test theories of criminal behavior.

None of these functions can be carried out effectively without the establishment of criminal justice statistics centers both within the States and at the national level by the Federal Government. At the present time, only California has a meaningful statewide criminal justice statistics system; hopefully, other States will be developing such centers in the near future. Presumably, a Federal Center will be established by the new agency in the Department of Justice called for in the President's safe streets and crime control bill. This National Center would build on and use the existing collection programs in the FBI, the Bureau of Prisons, the Children's Bureau, and elsewhere in the Federal Government. It would be concerned with stimulating the development of the State centers, establishing uniform definitions so that State reports are comparable, maintaining information that is of national interest, and publishing appropriate reports to make its data widely available. The National Center would ideally look to State centers as the primary sources of its information, and would probably provide technical and financial support to them. Until State centers are developed, it will have to get the material from cities or counties, wherever it may be available.

The President's safe streets and crime control bill, and all modifications of it that have appeared, place heavy emphasis on planning across the entire criminal justice system. In order for the Federal Government to be able to make wise allocations of the subsidies under that bill, it must have the relevant information. How, for instance, can one plan, intelligently, without good baseline data on how money is currently being spent and on the likely consequences of alternative investments. How will crime be affected, for instance, by using $2 million to hire additional corrections officials to lower the average case

load compared to using that money to buy additional patrol cars so that no police officer has to be limited to foot patrol.

The Statistics Center would probably be built around two basic files-one on agency statistics and one on criminal career records. The agency statistics file would comprise reports from criminal justice agencies on their workloads, costs, dispositional actions, pending cases, and other operational measures. These data would be augmented and calibrated by independent checks such as population and institution

surveys.

These data would provide the basis for regularly issued Center reports designed to provide a maximum of assistance to the operating officials. An effective statistical system requires the active cooperation of the sources of the basic input data, and a flow of complete and reliable data can be assured only if a useful product is returned. The reported information would help the official assess his organization's performance and make various operating decisions, such as deciding how many men he has to recruit the next year.

The Center's records might be maintained on a time-shared computer. A time-shared computer is a central computer with terminals set up at operating agencies around the country so they can make direct inputs into the computer and get a direct response. I hope they will be so maintained in a few years, if not initially. Then, an official would be able to handle many of his routine management information functions from a terminal in his office. The terminal would be connected through a communications network to the computer with its store of operating statistics and canned programs to perform such management computations as projections of workload, cost accounting, scheduling and allocation of resources, and budget analyses.

The second major file to be maintained would contain details on individual criminal records, but with the identifying information carefully removed. Even with that precaution, special care will have to be taken to preserve the security of that file because of the inherently personal information it contains. The National Crime Commission has recommended that the national file be limited to matters of record involving a person's contacts with the criminal justice system, and should include disposition of all arrests.

This file would provide both to the Center's staff and to other qualified researchers an invaluable source of basic data for evaluating the effectiveness of various correctional procedures. It would also permit relating criminal behavior to sociological and economic variables, and could thereby give rise to meaningful hypotheses that might lead to an understanding of the causal factors of crime, about which we are now very much in the dark.

In creating these files, compromise will be necessary between the inevitable desire for completeness of information, on one hand, and problems of tractability in terms of the cost and effort in obtaining and maintaining the records, on the other. In general, I would tend to err on the side of smaller records, simply because no preestablished format can ever be complete, and every new investigator is bound to want information that is either not in the file no matter how complete it is, or is not available in appropriate form. The money and goodwill that could be saved by economizing on the data collected could

then be used to support the individual researchers on their projects directly.

The establishment of the Statistics Center must be coordinated with the creation of a strong research program. There is nothing sadder than to see a room full of punched cards that have never been through a card reader. If it is worth the considerable effort and expense involved in collecting the data, then it is worth at least as much to create a research program that will put the data to use. Only with such interaction-the research program identifies the data to be collected and the data provide the nourishment for the research program-will we have a balanced program. That is an essential part of an effective program for crime control, and at minimum expense to the taxpayers.

To bring this all about, this committee might exert its influence to see that the safe streets and crime control bill is enacted and that funds are appropriated for the National Institute of Law Enforcement and Criminal Justice called for in current versions of that bill. Presumably, the Statistics Center would be a major part of that Institute. In the interim, you might encourage the creation of an interdepartmental task force to coordinate and improve the existing data collection programs. You might also encourage existing statistical series, such as the household census, to be expanded to include victimization questions, thereby permitting a calibration, or an estimate of the reporting error, in the existing crime data which come only from police reports.

Thank you, Mr. Chairman.

Mr. GREEN. Thank you very much, Dr. Blumstein. I appreciate your very fine statement.

You said, "Presumably, the Statistics Center would be a major part of the Institute." Do we have any reason to make such a presumption? Dr. BLUMSTEIN. I would infer that because the authority is implied within the existing legislation. Almost any organization that would create an institute of law enforcement and criminal justice would have to provide the data to support the operation of that institute. This has happened with NIH where we now have a medical statisties center. This has happened in the Office of Education where we have an office of educational statistics, And I would almost consider it irresponsible of any director of such an institute not to include as a major part of his program a center on criminal statistics.

Mr. GREEN. Do you have any estimate as to what it would cost to establish a criminal statistics center which could provide the data needed to make adequate judgments in the area of crime control?

Dr. BLUMSTEIN. It's very difficult to estimate that, because of the question of what part of the use of the statistics one incorporates into the Statistics Center. I think the best estimate would come from data on the other statistics centers in medical statistics and in educational statistics. I don't have the data with me. I believe they are in the report of the Assessment Task Force of the Crime Commission. They run in the neighborhood of a few million dollars per year, I believe.

COUNSEL. The costs are $8.3 million for the National Center for Health Statistics and $5.2 million for the National Center for Education Statistics.

Dr. BLUMSTEIN. In contrast to that I believe the estimate is that about $800,000 today is being spent on criminal statistics in the United

States. It was pointed out yesterday that the Gallup poll has found crime to be the No. 1 problem of national concern. Despite that, we are off by at least a factor of 10 in the kind of commitment to statistics that's required.

Mr. GREEN. I should say that one of the items that caused me personal concern was the report published recently by this committee pointing out that while $178 million a year is spent collecting statistics only $800,000 of this sum is for criminal statistics.

Dr. BLUMSTEIN. I might make the point that the great bulk of that $178 million in statistics is economic statistics. To a large degree, I believe that the significant progress the country has made in stabilizing its economy has resulted from the collection of statistics, enabling us to find out what is going on and to develop means for stabilizing our economy. Hopefully, we might be able to do this in some social areas

like crime.

Mr. GREEN. You mentioned in your statement that the lack of meaningful information about crime and criminals handicapped the work of the Crime Commission. Was there any particular area in which the lack of information was particularly distressing?

Dr. BLUMSTEIN. Let me describe a class of information rather than a single datum. That class of information relates to the effect of all the things that we do in our country to control crime. We cannot estimate what effect those actions will have on crime in quantitative terms, and we and the entire Commission tried.

Mr. GREEN. What would be the value, in your opinion, of the suggested "victim survey"? Do you think the Census Bureau has a role to play? Would you strongly recommend a question like this in a census questionnaire?

Dr. BLUMSTEIN. Yes, I would strongly recommend a question like this.

Mr. GREEN. Of what value would "victim" statistics have been to you, as a member of the Commission, in your report?

Dr. BLUMSTEIN. No. 1, an important question that keeps coming up is estimating really how much crime there is in the United States. The entire assessment task force was created, in part, for the purpose. As Professor Wilkins pointed out yesterday, the reported crimes account for just those crimes that are reported to the police and reported by the police to the "Uniform Crime Reports." These are just a part of the crime that occurs in the United States. There are many crimes that never get reported to the police. And this nonreporting rate differs with type of crime. According to the Crime Commission statements, for instance, rape is reported only about 10-30 percent of the time, whereas auto theft is reported virtually all the time. There have been trends in crime in the United States that have been rather striking and alarming in terms of crimes reported in the "Uniform Crime Reports." There has been a recognition both by the FBI and by the community that studies crime that there is a part of this iceberg that is hidden. It may very well be that there's a trend in the reporting rate-the percentage of crimes that get reported may be increasing. We don't know how much of the increase in reported crime rate is due to an increase in crime and how much is due to an increase in reporting rate, and so it would be very desirable, in order for the Nation to know the nature of the trend in crime, to know the trend in the reporting rate. We now