planned to fill out about 275,000 schedules from the files of probation departments, penal and correctional institutions, and parole authorities. There was to follow the editing, verifying, coding, punching, tabulating, analyzing, and making the results available in useful form. At the beginning the exact objectives of the study were stated as follows:

1. The preparation of a comparative digest of all laws and statutes, and of decisions pertaining to them, relating to the disposition of persons found guilty of crime.

2. The compilation of a census of the facilities available to the courts, probation departments, penal and correctional institutions, parole and pardon authorities, etc., that come in contact with persons found guilty of crime from the time of conviction to the time of final release.

3. The compilation of a census of case history data available in courts, penal and correctional institutions, parole offices, and other departments dealing with persons convicted of crime.

4. An analysis of court cases to determine the relationships between the offense, the characteristics of the convicted persons, and the eventual disposition of the case.

5. An analysis of case history data in probation cases to determine what characteristics are associated with successful termination of probation.

6. An analysis of case history data of parolees to determine what characteristics are associated with successful discharge from parole.

7. The preparation of tentative experience tables (where the findings warrant it) for the guidance of courts and paroling authorities, to aid them not merely in selecting individuals for probation or parole, but also in setting up more individualized forms of supervision.

8. The establishment of tentative standards for probation work, for parole selection and supervision, and particularly for institutional services aimed toward prevention and rehabilitation.

9. A tentative evaluation of the effectiveness of various release practices throughout the country and mapping of

the field to designate areas where more intensive research is likely to be of most immediate value.

To carry out the objectives of this survey the Attorney General appointed Mr. Justin Miller, one of his special assistants to serve as administrative director. Mr. Miller brought to his position a wide background of valuable knowledge and experience. The actual control of the entire project was in his hands, but the direction of the statistical phase of the study was the responsibility of Dr. Barkev S. Sanders who was appointed technical director of the survey. General policies for the study were drafted by an executive committee, appointed by the Attorney General, which originally consisted of Brien McMahon, Assistant Attorney General, Sanford Bates, Director of the Federal Bureau of Prisons, and Justin Miller, director of the survey. Dr. F. Lovell Bixby served on the committee from the time of Mr. Bates' resignation from the Department of Justice until he himself resigned in July 1937.

In January 1937, Mr. Justin Miller was appointed a member of the United States Board of Tax Appeals, and Dean Wayne L. Morse of the University of Oregon Law School was chosen to succeed him as administrative director of the survey. From July 1937, until the completion of the survey, the executive committee was composed of Assistant Attorney General Brien McMahon, chairman; Gordon Dean, Special Executive Assistant to the Attorney General; and James V. Bennett, Director of the Federal Bureau of Prisons.

The Attorney General invited a group of authorities in statistics who had wide experience in the analysis of sociological data to serve as a technical advisory committee. This committee performed many valuable services during the course of the survey. The technical advisory committee consisted of: Prof. Samuel A. Stouffer, chairman, Dr. Morris A. Copeland, Dr. W. Edwards Deming, Dr. Alba M. Edwards, Prof. Merrill M. Flood, Dr. Milton C. Forster, Prof. Thorsten Sellin, Mr. Frederick F. Stephan, Prof. S. S. Wilks.

The field work of collecting factual material about release procedures was carried on in all the States. Eleven geo

graphical regions were established, and the direction of the field work in each of these was turned over to two persons— a director and a statistical field supervisor. Immediately following their appointment, the regional directors and the statistical field supervisors were called to Washington for several days of intensive discussion and instruction in the methods to be followed in making the survey.

Two different approaches were adopted in organizing the field work. In the first place, the staff of regional directors undertook a first-hand investigation of the laws, organization and qualification of personnel, and procedures and practices involved in the administration of the various release procedures as they existed in each of the regions. In addition, the staff of statistical field supervisors set out to establish as many work units as necessary in order to abstract from available records statistical information concerning the personal, social, psychological, and other characteristics of persons convicted of crime. This information was recorded on schedule forms provided for this work and sent to the central office in Washington, there to be subjected to intensive analysis. Because of the nature of the work to be done the work units were established in the offices of prison administrators, prison boards, pardon boards, judges, bureaus of correction and probation, county clerks' offices, and similar places which in many instances were inadequate even for the use of those who generously shared their space with staff representatives of the survey.

The objectives and scope of the study as originally planned would have required several years for completion. Unfortunately, the Works Progress Administration was unable to approve projects for longer than a 1-year period at a time, and when after 1 year funds were not available to continue the research along the original lines, changes in the plans for the remainder of the study had to be made. Sufficient personnel to analyze the great mass of statistical data which had been collected could not be employed; therefore, only such data were analyzed as could be used for illustrative purposes in the volumes on probation and parole.

As originally conceived, the statistical phase of the survey was planned as a "prediction study." Because the data were

collected from this point of view, it has been impossible to obtain from the survey statistics the desired quantitative material, and hence other sources have had to be relied on for much of this type of data. For this purpose reports of the Bureau of the Census and the Bureau of Prisons have been invaluable. The principal difference in the statistical work contained in these volumes and the original plans for this phase of the study is that no attempt has been made to set up prediction tables. The statistical sections of the study are presented to point out some of the present difficulties in the administration of probation and parole. It has been possible to give a general picture of some of the factors that are considered in granting probation and parole and the attendant results. The methods of statistical analysis which have been used indicate that the inaccuracies which might have crept into the data have not invalidated the results. Despite the differences in administrative practices and despite the fact that parole and probation mean so many different things in so many different places, these variations were not sufficiently great to change the general effect of the findings.

Naturally there are many statistical limitations to this study. In the first place, because of limitations in time and in the number of personnel, it has not been possible to analyze all the important factors surrounding the granting of probation and parole. It is entirely possible also that the most important factors to be considered may be the least objective ones and may not even have come to light at all in the course of the survey.

All parole cases within a given period were not studied, but rather a sampling process was used in selecting the cases. Data were obtained regarding only those prisoners who were paroled on or after January 1, 1928, and who were discharged from parole on or before December 31, 1935. This sampling process together with the fact that the schedules were devised primarily for a prediction study limited the value of the statistics in many respects. Then, too, many of the items on the statistical schedules were highly subjective in nature, and the value of the information gathered concerning them was limited not only by its

subjectivity but also by the fact that the interpretations were made by unskilled relief workers. Although these workers were carefully supervised by competent supervisors who followed uniform instructions as to each item, the very subjectivity of the interpretations themselves rendered them of little value. Therefore, the few items which were selected for statistical analysis in the final report were chosen from the group of highly objective items on the schedule, requiring no interpretation by the relief workers.

Since the first steps in the statistical phase of the work were carried on almost exclusively by Works Progress Administration workers, an effort was made to determine the margin of error involved in transcribing information from the original records to the statistical schedule. To measure this, three different samples, each of 500 cases, were recorded and reverified by different individuals and the duplicate recordings analyzed for each item on the schedule. Comparing the results obtained with those of similar studies made by Professors Vold and Tibbits, it was found that the precision with which information was abstracted was about the same despite the fact that this work was done by Works Progress Administration workers while the rerecordings reported by Vold and Tibbits were made by the authors themselves or by capable graduate students.

Perhaps the greatest obstacle to presenting an accurate report on release procedures lies in the inadequacy of existing records. Time and again the application of statistical methods and the drawing of conclusions have been prevented by the faulty character of the data which were available. Many of the shortcomings of this study are traceable not to the methods used in collecting the data or in analyzing it but to the fact that the truth about release procedures is not written in the records. If this survey has disclosed any outstanding need in the administration of criminal justice it is for the installation of a uniform and comprehensive system of record keeping.

The survey has been a valuable instrument for testing the potentialities of large scale research projects carried on with a minimum of trained personnel and a large number of white-collar relief workers. From 1,200 to 1,800 workers