Mr. Bishop Robinson
Secretary,
Department of Public Safety and Correctional Services
6776 Reisterstown Road
Baltimore, Maryland 21215
Re: OBSCIS I Reporting System
Dear Mr. Robinson:
I have recently come into possession of the July, 1992 inmate data produced by the OBSCIS I computer reporting system. Prior to this, I was also in possession of the figures for October, 1991 and have been able to secure only partial results from April, 1992.
The data on these reports is summarized on the chart ("Total Prison Inmates in Maryland") attached. The OBSCIS results for each of the five categories (sex, race, age, sentence and offense) are contained in the column for the corresponding month (where available). The difference among columns, along with annualized projections of number and percent change, appear in the remaining columns.
The last three columns merely extrapolate the nine-month differences (between October, 1991 and July, 1992) to arrive at an "annualized" series of values. If the monthly data offered by the OBSCIS system are any indication, the "Annual Projection" and "Annual Percent Growth" columns will be realized as of the end of this month of October, 1992.
Margin of Error: In no category, for any time frame, do the numbers of inmates agree. In fact, for the month of October, 1991 the difference between the lowest and highest number is (for DOC) 137; for the month of July, 1992, the disparity is 285 inmates.
While the difference among the values generated by the OBSCIS I system appear to be on the order of only 0.5% at their greatest variance, the number of inmates obviously miscounted is about equal to an institution the size of Supermax or several of the Pre-Release units combined.
Reliability: Although the data by sentence and offense appear to have the closest correlation (if not agreement), their reliability is tainted by the fact that there is such a range of values during the same time frame in categories less prone to subjective judgment: i.e., sex, race and age. It is not possible to assume therefore that we are dealing with an undercount or, for that matter, an overcount, among any of the values returned by OBSCIS.
Predictive Value: Projecting the October, 1991 and July, 1992 (nine months) OBSCIS figures to predict an annualized rate of growth, it can be seen from the chart that the percent increase varies from a low of 4.5% (the Department as measured by sex) to a high of 6.2% (DOC as measured by race).
While it is well known that the Department and the Division have little control over the number of inmates entering the system or leaving the system in a given time period, it should be well within the purview of the Department's data gathering, computer equipment and statisticians to reconcile the number of persons in custody as of a given date in the recent past.
The OBSCIS output is obviously deficient and should not be publicly disclosed without concomitant analysis of where the data is thought to be lacking in precision (i.e., The OBSCIS report should contain a description of where methodology or other factors outside the ken of the preparers of the report).
The OBSCIS report is further deficient in that it does not contain, over the same period of the report, a measurement of intake (e.g., court commitments, violations of parole or probation) and releases (e.g., expiration of sentence, "mandatory" releases, paroles, deaths, etc.). If for no other reason, the OBSCIS counting from month to month may be tested against input from other data gathering sources (e.g., the courts, the Division of Parole and Probation, etc.) Ultimately, there would be better data on recidivism and the origins thereof.
Finally, the number of inmates in custody in the Department and especially the growth and rate of growth in that number has a great deal of bearing on capital planning and the capital budget. It is in the State's best interests to have an inmate population reporting system that is timely, accurate. reliable and conducive to advance forecasting.
For all of the above reasons, the monthly OBSCIS reports should be reevaluated and revised. Where necessary, data gathering and reporting should be made consistent and subject to quality control. In the event results are still error-prone, the margin of error should be documented and quantified. Under no circumstance should the Department release data that is manifestly uncertain.
Sincerely,
Nancy Moran
nm
Volunteer
nm
enclosure
cc:
Governor William Donald Schaefer
Ms. Nancy Nowak
Mr. Richard A. Lanham, Sr.
Dr. Joseph Henneberry