Prisoner release procedures in the Michigan Department of Corrections (DOC) suffer from serious flaws, according to the Intake Processing Unit (IPU). The IPU undertook an audit and review of the way the DOC determined prisoner release dates, and issued its findings in an October, 2005 report. The IPU tested the process by examining release records, and concluded that five serious problems prevented proper release calculations and resulted in some prisoners being released too early while others were held too long.

The Michigan DOC currently houses approximately 50,900 prisoners, and their sentencing information is stored in two databases. The Offender Management Network Information System (OMNI) contains all initial data with regard to a prisoner at intake, such as the length of sentence, type of sentence, etc. This data is transmitted electronically to the Correction Management Information System (CMIS), where central records office staff are responsible for calculating prisoners release dates.

The CMIS database includes comprehensive data on prisoners incarcerated in the DOC, community placement, boot camp, and on parole. The IPU's investigation of the integration and implementation of OMNI and CMIS by the central records office staff revealed the following problems.

First, the IPU found that CMIS miscalculated release dates for prisoners with consecutive sentences. It determined that between October 2003 and August 2004, seven prisoners were released from 39 to 147 days early. Subsequent to the IPU's findings, a double-check by DOC uncovered 15 more prisoners who had been erroneously released. The release discrepancies ranged from almost three years early to over a year late. The IPU used four methods available to the central office staff for computing release dates; all four methods yielded different results even when used on the same subjects. The audit concluded that CMIS was not correctly programmed to process release dates with respect to complex sentences.

Second, the IPU discovered that CMIS had insufficient data-edit control. That is, invalid data could be entered and accepted by the CMIS database. Neither did CMIS accurately reflect corrections. The IPU revealed 813 sentences in which the sentencing date preceded the offense date. For 29,400 sentences the corrected date preceded the offense date. The IPU concluded that the CMIS data editing process and its data dictionary were seriously flawed.

Third, data provided by the courts was not always accurately entered into CMIS. Nor did any form of information verification exist within the computer system or within the central records office procedures. For six of 12 cases identified by the IPU as needing verification, no attempt was made by the staff to contact the courts for accuracy. This negligence, coupled with the inaccurate input of release date adjustments, resulted in at least two errors in the 30 records checked.

Fourth, DOC had no method of conducting audit trails. Consequently, sources and accuracy of information, computations and adjustments of prisoners sentences could not be verified.

Fifth, the IPU concluded that DOC had not completely ensured the integrity of OMNI or CMIS for security purposes. This allowed access to confidential information by unauthorized persons, and changes in the system could not be tracked back to the person who had made them.

For the sake of accuracy it should be noted that only raw figures were used in the audit report. Because these numbers cannot be compared to a total figure, the actual percentage of error cant be determined from the report. It is also noteworthy that DOC agreed with all five of the IPU's findings, stating that the problems identified by the auditors were primarily due to an aging computer system and multiple sentencing law changes. The corrections department said it intends to fully implement the auditors recommendations by June 2007.

In another procedural failure related to early releases, a separate investigation revealed that 40 Michigan prisoners were prematurely freed due to improper parole revocation hearing policies, including one who subsequently committed three murders.

Patrick Alan Selepak, 27, was released on parole in June 2005; he returned to prison five months later on a parole violation for assaulting his girlfriend. Prison officials failed to hold a hearing within 45 days, and Selepak was freed on Jan. 10, 2006 despite having pending charges. The Michigan Supreme Court had ruled in 2003 that the DOC was not required to release prisoners who weren't provided hearings within 45 days, and Selepak's release violated departmental policy. A month after he was let out of prison Selepak and his fiancé, Samantha Bachynski, 19, allegedly murdered Scott Berels and his pregnant wife, Melissa, and later shot and strangled Winfield Fred Johnson. Had Selepak not been erroneously released he would have still been incarcerated at the time of the killings.

State Sen. Alan Sanborn sharply criticized the DOC, saying that "while parole officials didn't pull the trigger, they provided the access so that these animals could get the gun." He told DOC director Patricia L. Caruso that her employees were culpable and should be terminated. Following a three-month investigation Caruso obliged, firing parole supervisor Larry Baran. Baran had reportedly lied about his knowledge of the 45-day policy and had urged parole officials to conceal the Supreme Court's ruling from parole officers.

Another parole supervisor, Daryl Cantine, was forced to retire early under pressure. Cantine's supervisor, Carol Duncan-Smith, the administrator of Field Programs for the parole department, was demoted due to inadequate oversight and management of her subordinates. Lastly, Selepak's first parole officer, Martin Awe, received a ten-day unpaid suspension; Awe had failed to test Selepak for alcohol use, which was a condition of Selepak's parole since he had a known alcohol problem.

Our mission is to protect the public and I know that you feel we have failed in that, DOC Director Caruso told lawmakers. And I would be hard-pressed to argue with you. To prevent another failure by the parole supervision unit that had been responsible for monitoring Selepak, Caruso abolished the unit in May 2006. "There was an apparent lack of leadership, rampant misuse of discretion and very little supervision by the individuals responsible for overseeing that office", she said.

Such actions did not appease everyone, however, including House Speaker Craig DeRoche and Republican Party Chairman Saul Anuzis. We must remember that innocent human beings died because of this foul-up, stated Anuzis. Simply letting people off for early retirement is not the proper course of action. I reiterate my call for Gov. Granholm to ask for the resignation of Director Patricia Caruso. Governor Granholm declined to fire Caruso.

Selepak's serial murder spree generated widespread media attention that resulted in a larger investigation into the DOCs practices related to parole revocation hearings. It was discovered that within the previous year, 40 other prisoners were improperly released after they didn't receive timely revocation hearings. All have since been accounted for 23 were returned to custody, one died, seven were returned to parole, eight are no longer under parole supervision and one is a fugitive. Besides Selepak, several others had committed crimes following their improvident releases, ranging from robbery to sexual abuse of a child; one remained at large for almost a year. In May 2006 Selepak pled guilty to first degree murder and other charges. He faces a life sentence.

Sources: Lansing State Journal, Detroit News, Detroit Free Press.

Subscribe today

Already a subscriber? Login