Vermont State Auditor Sex Offender Registry Report on Reliability 2010
Download original document:
This text is machine-read, and may contain errors. Check the original document to verify accuracy.
Report of the Vermont State Auditor June 25, 2010 SEX OFFENDER REGISTRY Reliability Could Be Significantly Improved Thomas M. Salmon, CPA Vermont State Auditor Rpt. No. 10-05 Mission Statement The mission of the Auditor's Office is to be a catalyst for good government by promoting professional audits, financial training, efficiency, and economy in government, and service to cities and towns. This report is a work of the Office of the State Auditor, State of Vermont, and is not subject to copyright protection in the United States. It may be reproduced and distributed in its entirety without further permission from the State of Vermont or the Office of the State Auditor. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately. Please contact the Office of the State Auditor if you have questions about reproducing this report. THOMAS M. SALMON, CPA STATE AUDITOR STATE OF VERMONT OFFICE OF THE STATE AUDITOR June 25, 2010 Addressees (see next page) Dear Colleagues, As required by Act 58 (2009), this report provides the results of our performance audit of the State’s sex offender registry (SOR). The SOR contains identification, location, conviction, risk, and treatment information on offenders who have been convicted of sex offenses listed in 13 VSA §5401(10). For those offenders that meet the requirements of 13 VSA §5411a, a subset of the SOR data is also posted to a public website, http://www.dps.state.vt.us/cjs/s_registry.htm. We performed our work at the Department of Public Safety’s Vermont Criminal Information Center, which manages the SOR, and the Department of Corrections and the Courts, which provide critical information to the registry. Our audit was primarily focused on the reliability of data in the SOR as well as controls to prevent errors. With respect to the reliability of the SOR data, we found a sizeable number of serious errors. In addition, the processes used to submit and enter data into the SOR were largely manual and controls were not always documented or consistently applied. During the course of the audit, each of the organizations made changes that are expected to improve the SOR’s reliability. Moreover, the Vermont Criminal Information Center corrected errors in the SOR as they were brought to its attention. This report includes recommendations designed to improve the SOR’s data reliability and controls. In particular, we recommend that the organizations that we reviewed form a working group to reassess and possibly redesign the processes related to the Vermont sex offender registry to include possible system solutions to more effectively and efficiently submit information to the SOR. I would like to thank the management and staff at the Vermont Criminal Information Center, Department of Corrections, and the Courts for their cooperation and professionalism during the course of this audit. If you would like to discuss any of the issues raised by this audit, I can be reached at (802) 828-2281 or at firstname.lastname@example.org. Sincerely, Thomas M. Salmon, CPA Vermont State Auditor 132 State Street • Montpelier, Vermont 05633-5101 Auditor: (802) 828-2281 • Toll-Free (in VT only): 1-877-290-1400 • Fax: (802) 828-2198 email: email@example.com • website: www.auditor.vermont.gov ADDRESSEES The Honorable Shap Smith Speaker of the House of Representatives The Honorable Peter Shumlin President Pro Tempore of the Senate The Honorable Richard Sears Chair Committee on Judiciary Vermont Senate The Honorable William Lippert Chair Committee on Judiciary Vermont House of Representatives The Honorable Alice Emmons Chair Committee on Corrections and Institutions Vermont House of Representatives and Chair Joint Committee on Corrections Oversight The Honorable James Douglas Governor Mr. Robert D. Hofmann Secretary Agency of Human Services Mr. Thomas R. Tremblay Commissioner Department of Public Safety Mr. Andrew A. Pallito Commissioner Department of Corrections Mr. Robert G. Greemore Court Administrator Office of the Court Administrator Contents Report Introduction Page 1 Highlights 3 Background 4 Errors in SOR Were Pervasive and Timeliness Generally Could Not Be Gauged 11 Significant Control and System Limitations Warrant Urgent Attention 24 Observations Pertaining to the Posting of Sex Offender Addresses on the Internet 36 Conclusions 38 Recommendations 38 Management Comments and Our Evaluation 40 Appendix I: Scope and Methodology 43 Appendix II: Statutory Definition of a Sex Offender 49 Appendix III: Example of the Information Posted on the Internet SOR 51 Appendix IV: Profile of Sample of Community-Based Offenders 52 Appendix V: Summary of Errors in the Community-Based Statistical Sample 54 Appendix VI: Response of the Commissioner of the Department of Public Safety 56 Appendix VII: Response of the Commissioner of the Department of Corrections 59 Appendix VIII: Response of the Court Administrator 61 Contents Tables Page Table 1: Major Forms Submitted to VCIC Related to the SOR 8 Table 2: Offender SOR Status Categories 10 Table 3: Number and Types of Errors Found in Sample of 57 Community-based Offenders 54 Figure 1: Effect of the Errors in the Community-Based Sample (57 records) 3 Figure 2: Simplified Diagram of the Sources and Types of Data Sent to the SOR 7 Figure 3: Effect of the Errors in the Community-Based Sample (57 records) 12 Figure 4: Timeframes for VCIC Receipt of Offender Annual Verification Letters (sent between February 18, 2009 and January 18, 2010) 24 Figures Abbreviations CPI DOC DPS GAO PO SOR VCHIP VCIC VSA Computer Projects of Illinois Department of Corrections Department of Public Safety U.S. Government Accountability Office probation and parole officer Sex Offender Registry Vermont Criminal History Information Program Vermont Criminal Information Center Vermont Statutes Annotated Introduction In 1996, with the passage of 13 VSA, Chapter 167, Subchapter 3, the Vermont Sex Offender Registry (SOR) was established at the Department of Public Safety’s (DPS) Vermont Criminal Information Center (VCIC). Vermont’s SOR has undergone substantial changes since the passage of the original act. In particular, Act 58, approved on June 1, 2009, greatly expanded the number of offenses that require that the offender be posted to the Internet.1 Section 14 of Act 58 also requires that addresses of offenders be posted to their Internet records if the offender (1) has been designated as high-risk by the Department of Corrections (DOC), (2) has not complied with sex offender treatment, (3) has an outstanding warrant for his or her arrest, (4) has a conviction for a sex offense against a child under 13 years of age, or (5) has been electronically posted for an offense committed in another jurisdiction that required the person’s address to be electronically posted in that jurisdiction. This section of the Act takes effect July 1, 2010, after the issuance of a performance audit report.2 (The remainder of the report will refer to the primary registry as the SOR and the Internet portion as the Internet SOR.) The objectives of our audit were to assess (1) the extent to which the data in the State’s SOR is reliable and current, and (2) whether the State’s controls are designed to prevent errors, omissions, and outdated registry data. In addition, because our report was linked to the inclusion of offender residential addresses on the Internet SOR and some members of the General Assembly have expressed reservations about the accuracy of offender addresses that may be posted on the Internet SOR, we are also providing a section at the end of this report that provides our observations on this topic. Our methodology for the first objective included (1) employing automated data analysis techniques on an electronic copy of the SOR as of February 18, 1 The Internet SOR can be found at http://www.dps.state.vt.us/cjs/s_registry.htm. 2Section 28 of Act 58 states that Section 14 “shall not take effect until the state auditor, in consultation with the department of public safety and the department of information and innovation technology, has provided a favorable performance audit regarding the Internet sex offender registry to the senate and house committees on judiciary, the house committee on corrections and institutions, and the joint committee on corrections oversight.” Our role is to conduct the audit and report on its results, but policy decisions, such as what constitutes a favorable audit in the context of Act 58, is the role of the Legislature and the Administration. Page 1 2010 and (2) extracting a statistical sample of community-based3 offenders from this same electronic copy of the SOR and tracing data in individual records to source documentation. With respect to our sample, we used statistical random attribute sampling4 with a 95 percent confidence level, and 5 percent tolerable rate of records not in compliance with the statutorily established requirements and zero expected population deviation rate (i.e. expected error rate).5 The audit work for our second objective largely consisted of reviewing the SOR statute and rules promulgated by VCIC and DOC, reviewing the various forms and mechanisms used to convey data to the SOR, and conducting walkthroughs at VCIC, DOC, and the District Courts, which collectively provide much of the data to the SOR. We also interviewed DOC probation and parole officers (PO). Additional detail on our scope and methodology can be found in appendix I. 3 Community-based offenders are those offenders who are currently residing, employed, or in school in Vermont or whose last known residence was in Vermont. 4 Attribute sampling is a type of statistical sampling used to reach a conclusion about a population in terms of a rate of occurrence. Statistical sampling uses the laws of probability to measure sampling risk while the random selection of the sample ensures that every sampling unit (in this case a sex offender record) has the same probability of being selected as every other sampling unit in the population (e.g., community-based offenders). 5 We defined an error as encompassing (1) records that were incorrectly omitted, added, retained, or deleted from the registry, (2) data in the registry that differed from the source documentation, (3) omission of data in a registry field, when applicable, and (4) inaccurate registry calculations or decisions (e.g., whether a record should be on the Internet registry). Page 2 Highlights: Report of the Vermont State Auditor Sex Offender Registry: Reliability Could Be Significantly Improved (June 25, 2010, Rpt. No. 10-05) Why We Did This Audit Findings Act 58 required that we conduct a performance audit of Vermont’s sex offender registry. Our objectives were to assess (1) the extent to which the data in the State’s SOR is reliable and current and (2) whether the State’s controls are designed to prevent errors, omissions, and outdated registry data. There were a sizeable number of serious errors in the SOR and the currency of the system’s data could not be determined. With respect to the SOR’s reliability, one or more errors were found in the records of most of the 57 community-based offenders in our randomly selected statistical sample in which we traced data in the system to supporting documentation and assessed whether offenders were properly categorized (e.g., were or were not appropriately on the Internet SOR). Moreover, about three-quarters of these records had errors that were critical or significant in nature (see figure 1), in that offenders SOR or Internet SOR status was incorrect or would have been incorrect if not fixed after being brought to VCIC’s attention (critical) or that data used to identify and locate offenders or conveyed to the public or law enforcement was incorrect, incomplete, or omitted (significant). What We Recommend Figure 1: Effect of the Errors in the Community-Based Sample (57 records) Our report makes numerous recommendations to address specific types of errors and process weaknesses. We also made a recommendation that VCIC, DOC, and the Courts form a working group to reassess and possibly redesign the processes related to the Vermont sex offender registry to include possible system solutions to more effectively and efficiently submit information to the SOR. Critical Significant Other No ne Further, using an automated data analysis tool to identify anomalies in the SOR database as a whole, we found an additional 195 critical and significant errors. VCIC corrected these errors as they were brought to its attention. The errors in the SOR resulted from a variety of causes, including data entry errors and inaccurate calculations, inaccurate or incomplete information provided by DOC, and SOR system weaknesses. On the positive side, a few tests, including verification that the SOR contained all sex offenders who had been convicted and sentenced in Vermont over a 3-year period, yielded no or few errors. With respect to the currency of data in the SOR, data was either not available or not reliable enough to perform a systematic analysis of whether information was being received and entered into the SOR in a timely manner. An exception was the annual process used to verify offenders’ location data. In this case, VCIC was generally receiving data from offenders within required timeframes. The organizations (i.e., VCIC, DOC, and the Courts) and processes used to support the SOR did not work in a seamless manner, which limited the State’s ability to prevent errors, omissions, and outdated registry data. Specifically, the processes were largely manual and the controls were not always documented or consistently applied. For example, the VCIC and DOC processes in place to ensure that sex offender treatment status in the SOR is up to date were not always utilized or documented. Moreover, the SOR system lacks features, such as electronic interfaces, logic edits, and audit trails, to help prevent errors or identify what changes were made, when, and by whom. Although VCIC and DOC have taken action to address some of the process and control weaknesses, a comprehensive approach involving all of the organizations that provide most of the data to the SOR is more likely to achieve meaningful and sustainable improvement. Page 3 Background Vermont’s SOR has undergone significant changes in the past few years as new laws have been enacted that have made a variety of changes, including changes to the definition of a sex offender, the addition of lifetime registration for some registrants, and requirements to post certain offenders to the Internet. These changes have not only affected VCIC, which is responsible for the SOR, but the other State organizations that provide significant data to the SOR, namely the DOC and the Courts. The resulting systems and processes used to support the SOR are complicated and interwoven among these three organizations. SOR Statutory Requirements Since the 1996 law that first mandated the establishment of the SOR,6 the legislature has made several significant revisions to 13 VSA, Chapter 167, Subchapter 3, which provides the legislative framework for management of the Registry. For example, • Act 49, which was approved on June 12, 2001, required some sex offenders to register for life7 (all others are required to register for 10 years after release from prison or discharge from parole, supervised release, or probation, whichever is later). This Act also authorized DPS to participate in the National Sex Offender Registry Program.8 • Act 157, which was approved June 8, 2004, required that certain sex offenders be posted to an Internet Registry and specified the information that should be posted. This Act also authorized DOC to evaluate sex offenders for purposes of determining whether the offender is high risk in order to identify offenders who should be subject to increased public access as to his or her status as a sex offender. 6 Act 124 approved on April 25, 1996 and effective on September 1, 1996. 7 The lifetime registration requirement of Act 49 became effective September 1, 2001. 8 Currently, the U.S. Department of Justice hosts the Dru Sjodin National Sex Offender Public Website, which is a cooperative effort between the jurisdictions hosting public offender registries and the federal government. The national website can be found at http://www.nsopw.gov. Page 4 • Act 192 (effective as of May 26, 2006) amended the Internet Registry provision of the SOR statute, by expanding the types of offenders that are required to be posted to the Internet SOR. • Act 77, which was approved on June 7, 2007, established the noncompliant high-risk category of sex offenders.9 These offenders are subject to additional SOR reporting requirements. • Act 58, approved in 2009, modified the definition of a sex offender (see appendix II for this definition), greatly expanded the number of crimes for which an offender would be posted to the Internet SOR,10 and required that additional information be included in the Internet SOR. In particular, this Act requires that addresses of offenders be posted to their Internet records if the offender (1) has been designated as high risk by DOC, (2) has not complied with sex offender treatment, (3) has an outstanding warrant for his or her arrest, (4) has a conviction for a sex offense against a child under 13 years of age, or (5) has been electronically posted for an offense committed in another jurisdiction that required the person’s address to be electronically posted in that jurisdiction.11 DPS and DOC have issued administrative rules to implement the SOR statutory requirements. DPS issued a Sex Offender Registry rule effective December 31, 2004 and DOC issued a Determination of High Risk and Failure to Comply with Treatment for Purposes of Sex Offender Internet Registry rule effective June 8, 2005. In addition, DOC issued an internal directive the purpose of which was to establish procedures for (1) ensuring that sex offenders under the supervision of DOC are registered with the SOR 9 13 VSA §5411d defines noncompliant high-risk sex offenders as those who (1) were incarcerated on or after June 7, 2007 for lewd and lascivious conduct with a child, sexual assault, aggravated sexual assault, or any attempt to commit these crimes or a comparable offense in another jurisdiction in the United States, (2) are not subject to indeterminate life sentences under 13 VSA §3271, (3) are designated as a high-risk sex offender, (4) are noncompliant with sex offender treatment. 10 This law inadvertently omitted offenders with out-of-state convictions from these new Internet posting requirements. Act 66 approved on February 24, 2010 corrected this omission. 11 The earliest that this section of the Act could take effect is July 1, 2010. Page 5 and (2) the determination of high-risk sex offenders and those offenders who are not in compliance with DOC treatment criteria.12 Overview of Process Used to Provide Data to SOR While managed by VCIC, the source of the information in the SOR is largely from DOC caseworkers or POs,13 the Courts, and the offenders themselves. Figure 2 provides a simplified view of the main types and sources of the data in the SOR. Except for offender identification and conviction and sentencing information, this information is generally received by VCIC via the U.S. mail, faxes, and email and entered into the SOR via manual processes. Moreover, much of the sex offender conviction and sentencing information is manually input into another VCIC system, the Vermont Criminal History Information Program (VCHIP) before it is made electronically available for input into the SOR. 12 The DOC directive that establishes procedures related to the SOR was originally issued in 1998 and revised in 2005 and 2008. The most recent version was signed on May 10, 2010 and was effective June 14, 2010. Unless stated otherwise, the DOC SOR directive mentioned in this report is referring to the 2008 version. 13 Caseworkers are responsible for offenders’ SOR paperwork when the offenders are confined in a facility while POs are responsible for offenders who have been released to the community under various types of supervision, such as probation, parole, or furlough. Page 6 Figure 2: Simplified Diagram of the Sources and Types of Data Sent to the SOR Court Administrator’s Office and District Courts: • Conviction and sentencing data, such as the docket number, offense literal (narrative description of the charge), and type and length of sentence • Violation of Probation report Vermont State Police VINE System: • Offenders’ photos VCIC Fingerprint System: • Offender identification data, such as name, aliases, social security number, date of birth, sex VCIC’s VCHIP System: • Conviction and sentencing data • Offender identification data District Courts: • Order to Provide Information, which includes the data required by 13 VSA §5403, such as residential, employment, and postsecondary school data DOC: • Registration forms • Changes to residential, employment, and postsecondary school enrollment data • Changes in supervision status, including when offender is discharged • Risk designations • Treatment information • Offenders’ photos Offenders: • Registration forms • Changes to residential, employment, and postsecondary school enrollment data • Annual verification letter (e.g., verifies residential address, employment and postsecondary school data) • Signed treatment forms from provider VCIC staff adds data to SORa a VCHIP is the only system that provides data electronically to the SOR although the VCHIP data exchange also requires human intervention in that the SOR coordinator must accept each individual record before the data is electronically transferred. Page 7 Further complicating the process is that the responsibility for providing this information to VCIC can vary and the forms can be completed and sent to VCIC intermittently as the offender goes through various court and supervision processes during his or her time on the registry. Table 1 outlines the major forms that are submitted, when, and by whom. a Table 1: Major Forms Submitted to VCIC Related to the SOR Form Criminal Charge Disposition Report Order to Provide Information Violations of Probation Registration Form Change of Address, Employment, Education Form Notification of High-Risk Offender Responsible Submitter Court Administrator’s Office Types of Data Elements on Formb Timing of Submission Docket information; disposition of the Monthly electronic transmission of case, and sentence; offense code (the the prior month’s cases statute under which the offender was convicted) and offense literal (narrative description of charge). District Courts Weekly or bi-weekly Docket information and offender date District Courts Within 10 days after sentencing of birth, social security number, residential address, name and address of post-secondary school in which the offender is enrolled, and employer name and address.c Disposition of the violation and District Courts Weekly or bi-weekly sentence imposed, if applicable. Offender identification information; DOC Caseworker No later than 5 days prior to release conditions of release; treatment from confinement compliance status, high risk designation request; offender location DOC PO Initial registration within 5 days of data (e.g., physical residential address, sentencing and later forms at the employer name and address); discretion of the PO conviction and sentencing information; age and gender of victim. Offender convicted in Within 10 days of arrival in another jurisdiction Vermont Offender residential physical address; DOC PO For offenders under DOC mailing address; employer name and supervision, within 24 hours of the address; post-secondary school change attending and address; supervising caseworker; DOC field office. Offender For offenders not under DOC supervision, within 3 days of the changed Name and address of offender that has DOC Sex Offender No later than 5 work days after such been determined to pose a high degree a determination has been made Review Committeee of dangerousness to others. Page 8 Form Responsible Timing of Submission Submitter DOC caseworker No later than 5 days prior to release from confinement Types of Data Elements on Formb DOC Sex Information on the compliance status Offender of offenders under DOC supervision. Treatment Compliance and Non-Compliance Checklist Noncompliant High-Risk Offender Form Change of Treatment and Supervision Status Formf Registration Information Verification Notice Information similar to the Registration Form with the addition of vehicle make, model, color, registration, and license plate number. Date probation is closed or parole or furlough expired and treatment status at time of discharge. DOC PO Within 60 days of an offender being placed on community supervision the compliance determination should be made and the form submitted to VCIC within 5 days. Also, within 5 days of a change in status DOC caseworker Prior to offender being released from total confinement DOC caseworker or PO When an offender’s supervision status has changed (e.g., has been discharged from probation) Offender Annually sent out to offender within 10 days of birthday and must be signed and returned to VCIC within 10 days Signature of offender attesting to accuracy of the information in the SOR regarding physical address, mailing address, employment information, and school information. a This table provides information on the current forms. These forms have changed over time and the forms in VCIC files for SOR registrants may not have had all of the data elements shown in this table. b This column provides representative information on what is on each form and does not include all data elements. c This form also requires that offenders obtain fingerprints, a photograph, and physical description from an agency identified by the court. d Offenders designated as (1) high-risk are required to report a change within 36 hours and (2) noncompliant high-risk are required to report a change of address prior to the change or, if the change was unanticipated, within 1 day of the change. e This is a 5-member committee appointed by the Commissioner of Corrections. f This form was added in early 2010. Based on the data received about a particular sex offender, VCIC’s SOR coordinator makes certain decisions regarding the offender and manually enters the result into the system. Specifically, the coordinator decides whether the offender meets the statutory criteria to be (1) on the Registry, (2) posted to the Internet and, if so, the reason for such posting, and (3) a lifetime registrant and calculates the date of the end of the offender’s registration period if the offender does not meet this criteria. In addition, the SOR coordinator categorizes the registration status of each offender, as outlined in table 2. Page 9 Table 2: Offender SOR Status Categories Status Name Pending Active MIA Homeless Status Description Convicted sex offenders for whom VCIC has received conviction and/or sentencing information from the court, but for whom a signed Registration Form has not been submitted to VCIC. VCIC does not consider offenders in the “pending” status to be registered with the SOR and they are not on the Internet SOR. All registered sex offenders (i.e., those for whom a signed Registration Form was submitted to VCIC) who reside in a Vermont community. These offenders are eligible to be on the Internet SOR if they also meet the criteria of 13 VSA §5411a. Sex offenders that are currently in violation of Vermont SOR requirements. These offenders are eligible to be on the Internet SOR if they also meet the criteria of 13 VSA §5411a. Registered sex offenders who have no actual physical address and must report daily to local police and/or the registry. These offenders are eligible to be on the Internet SOR if they also meet the criteria of 13 VSA §5411a. Student/Employed Registered sex offenders that reside in another state or country but who work or are students in Vermont. These offenders are eligible to be on the Internet SOR if they also meet the criteria of 13 VSA §5411a. Inactive Sex offenders who are incarcerated (whether on a sex offense or non-sex offense conviction). These offenders are not on the Internet SOR. Out-of-State Sex offenders with either a Vermont or out-of-state conviction who have registered in Vermont and who currently reside outside of the state. These offenders are eligible to be on the Internet SOR if they also meet the criteria of 13 VSA §5411a. Reg Exp Sex offenders who have finished their registration obligation – registration has expired. These offenders are not on the Internet SOR. # of Offenders as of 2/18/10 136 1,443 75 7 24 423 361 164 During the course of an offender’s registration period, his or her status category and Internet posting category may change. For example, an offender who was in active status and was posted to the Internet SOR that is later incarcerated on another charge would have his or her status changed to inactive and would no longer be posted to the Internet SOR until such time as he or she is no longer incarcerated. With respect to the process of posting offenders to the Internet SOR, every evening the vendor who maintains the SOR for VCIC runs a program to extract the Internet-eligible offenders from the SOR. The vendor then posts this extract file to the Internet SOR, which is located on a server in the Department of Information and Innovation’s data center. This process results in a complete replacement of the file posted on the previous night. See Page 10 appendix III for an example of what an offender’s record looks like on the Internet SOR. Errors in SOR Were Pervasive and Timeliness Generally Could Not Be Gauged The number and diverse types of errors14 in the SOR indicate that the SOR has a pervasive reliability problem.15 Specifically, there were (1) offenders on the SOR who did not meet the statutory requirements for registration, (2) errors in dates that effected how long an offender had to be registered, which resulted in offenders being registered too long or not long enough, (3) offenders who were listed on the Internet SOR who did not meet the requirements for such posting as well as offenders who should have been posted to the Internet SOR and were not, and (4) a wide variety of data elements that included incomplete, inaccurate, or omitted data. VCIC corrected these errors as they were brought to its attention. There was not one primary reason for these errors. For example, some errors resulted from data entry errors while others stemmed from incorrect or unclear information from DOC. Moreover, a system error resulted in 174 offenders that should have been omitted from the Internet being posted to the national sex offender website (but not the Vermont Internet SOR) for about 7-1/2 months.16 In addition to reliability concerns, it was not possible to gauge the timeliness of data entered into the SOR. Specifically, with one exception, data was not available to determine with certainty whether data was being sent to, and entered by, VCIC in a timely manner. The exception was related to the timely completion of annual verifications of offender residential and mailing addresses, employer names and addresses, and school names and addresses. 14 We defined an error as encompassing (1) records that were incorrectly omitted, added, retained, or deleted from the registry, (2) data in the registry that differed from the source documentation, (3) omission of data in a registry field, when applicable, and (4) inaccurate registry calculations or decisions (e.g., whether a record should be on the Internet registry). 15 For purposes of this audit, we defined reliability as (1) all offenders who fulfill the statutory requirements for being on the SOR are on it and those that do not are not, (2) each SOR record has a full set of information, and (3) data in each record is accurate. 16 As allowed in Act 58, these offenders had submitted petitions to DOC requesting a waiver from the Act’s requirement that their records be posted to the Internet SOR. Offenders were not supposed to be on the Internet while their petitions were being reviewed. This system error was corrected in May 2010 and the offenders with outstanding petitions were removed from the national sex offender website. Page 11 In this case, 85 percent of the verification letters were submitted to VCIC within expected timeframes. Registry Had Incomplete, Inaccurate, and Omitted Data The SOR had a substantial number of errors, including incomplete, inaccurate, and omitted data. Some of these errors were not material in that they did not affect whether an offender was on the SOR or Internet SOR or the accuracy of information that was provided to the public or law enforcement. However, as illustrated in figure 3, there was a sizeable percentage of the 57 records in the community-based statistical sample that contained errors in which the effect of the errors was critical or significant. (See appendix I for an explanation of how we categorized the effect of errors. In general, records with critical errors indicate that offenders’ SOR or Internet SOR status was incorrect or would have been incorrect if not fixed after being brought to VCIC’s attention and records with significant errors indicates that data used to identify and locate offenders or conveyed to the public or law enforcement was incorrect, incomplete, or omitted.) Figure 3: Effect of the Errors in the Community-Based Sample (57 records) 7% 28% 14% Critical Significant Other None 51% The statistical sample of community-based offenders (i.e., offenders in active, MIA, homeless, and student/employed status) was based on a 95 percent confidence level, and 5 percent tolerable rate of records not in compliance with the statutorily established requirements and zero expected population deviation rate (i.e. expected error rate).17 By tracing the data in 17 Appendix IV provides a profile of the offenders that were included in our random statistical sample. Page 12 the SOR to supporting documentation and evaluating VCIC decisions related to the offender we were able to identify errors and estimate an error rate (plus or minus 5 percent) for the community-based SOR population. The error rate for each data element reviewed in the sample can be found in appendix V. We also utilized automated data analysis techniques on an electronic version of the SOR as of February 18, 2010. These techniques allowed us to test the entire database (including offenders that were in pending, inactive, out-ofstate, and “reg exp” status) for obvious errors (e.g., missing data, incorrect calculations) and logical inconsistencies (e.g., offenders required to register for life who have a date in the end registration date field) that were then checked against supporting documentation to determine whether the test exceptions were true errors. The results of our automated data analysis also yielded a substantial number of critical and significant errors. Specifically, the automated data analysis found an additional 165 and 30 offender records with errors18 that had a critical and significant effect, respectively.19 There was no single cause of the errors in the SOR. Among the reasons for errors were (1) data entry and calculation mistakes, (2) incorrect or unclear information provided to VCIC by DOC, (3) the misapplication of statutory requirements, (4) SOR system anomalies and glitches, and (5) incomplete conviction information sent from VCIC’s VCHIP (criminal history) system. In addition, it is more difficult to maintain up-to-date data in a system in which there are changes in the underlying requirements (e.g., the SOR statute). Among the data elements that were added to the SOR because of statutory changes were victim ages, sex offender treatment compliance, and high risk designations. It appears that some of the errors (particularly the omissions) were caused by an incomplete data collection process to backfill the records of offenders already in the SOR. The most critical errors in the SOR were those in which offenders were registered who did not meet the statutory requirements for registration or those offenders who have not been registered as required even though they have been in the community for years. With respect to the offenders who should not have been registered in the SOR, (1) one offender was convicted 18 We removed the records of the offenders that were in our community-based sample from the results of the automated data analyses so as not to count errors multiple times. 19 Our automated data analysis also found many records with other errors, but the effect of those errors did not meet our definitions of critical and significant so we did not include them in our count. Page 13 and discharged prior to the inception of the SOR;20 (2) three offenders were convicted of offenses that are not listed in the statutory definition of a sex offender21 and there was no evidence that the judges ordered them to register; (3) three sex offenders convicted in other jurisdictions had already exceeded the 10 year registration requirement prior to moving to Vermont; and (4) one offender met the criteria in statute that limited registration of offenders who were younger than 18 at the time of the offense.22 These eight offenders have subsequently been removed from the SOR. Also, the SOR included 14 offenders who fell into the second category of the most critical errors—offenders who have not been registered even though some were released into the community years ago. In one of these cases a registration form was submitted in 2003 and was inadvertently overlooked by VCIC staff, but in the other cases there was no evidence that a registration form was received by VCIC. For example, an offender maxed out23 his sentence in 2002, but there was no evidence in either the VCIC or DOC files that a registration form was completed before he was released from the correctional facility. Of the eight other offenders in this category that were supervised by DOC, (1) DOC’s files had no evidence that a registration form had been completed as of February 18, 2010 in four cases and (2) DOC could not find four of the offenders’ files so we do not know whether these files included the required forms. The 14 offenders who are not registered have a record in the SOR (most of them are listed in MIA status) and in some cases are posted to the Internet SOR (although their Internet SOR records generally did not include a photograph and other identifying information). Nevertheless, the lack of a registration form is important because (1) the form includes the notification 20 Offenders are required to register with Vermont’s SOR if (1) convicted in Vermont on or after July 1, 1996, (2) convicted in Vermont or another state prior to July 1, 1996, confined under the custody of the Commissioner of Corrections, and released from confinement in Vermont on or after July 1, 1996, (3) convicted in Vermont or another state prior to July 1, 1996, and was being supervised in the community in Vermont by the Commissioner of Corrections on July 1, 1996, or (4) convicted or released from confinement in another state on or after July 1, 1986 and who establishes residence in Vermont on or after July 1, 1996. 21 13 VSA §5401(10). 22 13 VSA §5401(10)(B) states “A person who is convicted of any of the following offenses against a victim who is a minor, except that, for purposes of this subdivision, conduct which is criminal only because of the age of the victim shall not be considered an offense for purposes of the registry if the perpetrator is under the age of 18 and the victim is at least 12 years old.” 23 An offender is considered to have maxed out his sentence when he or she completes his or her maximum sentence while incarcerated and is released without further supervision by DOC. Page 14 of the offender’s duty to comply with registration requirements (including his or her signature acknowledging these requirements) as required by 13 VSA §5406 and (2) VCIC did not know where the offenders were going to be located once they left DOC supervision for later follow-up. Offenders who knowingly do not comply with registration requirements are subject to arrest and the penalties outlined in 13 VSA §5409, which can include fines and imprisonment. If an offender is not compliant with registry requirements, the SOR coordinator can submit an affidavit to the applicable State’s Attorney’s Office attesting to the noncompliance and requesting an arrest warrant. If the State’s Attorney’s Office decides to prosecute, an arrest warrant is issued. In the cases of the 14 offenders who are not registered, an arrest warrant was issued for three offenders. In the other 11 cases there was no evidence that an affidavit was ever completed and submitted requesting an arrest warrant. The following subsections highlight some of the other pervasive and important errors in the SOR. We brought errors to the attention of VCIC as we found them and changes were made to the SOR, as needed. Discharge Dates and End of Registration Dates 13 VSA §5407(e) requires that, except for offenders that meet the requirements for lifetime registration, offenders are required to register for 10 years subsequent to their release from incarceration or discharge from supervision, whichever is later. According to DPS’s SOR Rule, DOC is required to forward a discharge notice to VCIC when a sex offender is discharged from the supervision of DOC. VCIC, in turn, generally24 uses this date to calculate the date that an offender’s registration requirement ends, except for those offenders who meet the criteria for lifetime registration. When an offender’s registration period is finished, a VCIC staff member changes the offender’s status to “reg exp” and sends a letter to the offender informing him or her that he or she is no longer required to comply with Registry requirements. Out of 57 offenders in our community-based statistical sample, 16 (28 percent) had discharge dates that were inaccurate or missing. In addition, our automated data analysis found that another 47 offenders had inaccurate or 24 Offenders with out-of-state convictions were treated differently. In the past VCIC had used the start registration date as the basis for the beginning of the 10-year registration period rather than the discharge date. However, recently VCIC changed its process and is now using the discharge dates for these offenders unless the date is not available. Page 15 missing discharge dates. Because the discharge dates were incorrect or omitted, the dates in which these offenders’ SOR registration obligations ended were also wrong—in some cases by years. Documentation was not always available at VCIC to determine the origin of the inaccurate discharge dates in the SOR, although in at least 2 cases the discharge date was wrong because VCIC used a discharge date for an unrelated offense committed by an offender, not the date from the relevant sex offense conviction. Most of the cases in which the discharge date was missing related to offenders convicted outside of Vermont (although many were supervised by DOC under an interstate compact).25 In these cases, VCIC’s standard practice was to use the offender’s registration start date as a basis for calculating the 10year end registration date rather than the discharge date. However, offenders’ discharge dates were generally available at DOC or other jurisdictions and VCIC’s files did not indicate that this information had been requested. The registration end dates calculated from the discharge dates often significantly differed from those calculated based on when an offender started registration. For instance, the difference in the registration start date and the discharge date for one offender convicted out of state but supervised by DOC was about 3 years. An additional 13 offenders had the correct discharge date information in SOR, but their end of registration date was miscalculated. For example, one offender’s discharge date was in 2004, but the end of registration date was miscalculated and showed that his registration would end in 2017 instead of 2014. Another important element in determining an offender’s end of registration is whether he or she meets the requirements of 13 VSA §5407(f) to register for life (applies only to offenders who were convicted after September 1, 2001). Offenders are required to register for life if they have (1) multiple sex offense convictions, (2) been convicted of aggravated sexual assault (13 VSA §3253), (3) been convicted of sexual assault (13 VSA §3252),26 (4) been designated as a sexually violent predator, or (5) been designated as a noncompliant high-risk offender. There were 22 offenders that met the requirement for lifetime registration, but had a date in the registration end field (it should have said “NONEXP” to indicate that the record does not 25 The Interstate Compact for Adult Offender Supervision governs the transfer of offenders, including sex offenders, among different jurisdictions. 26 The lifetime registration requirement does not apply to offenders convicted of sexual assault if the offender is not more than 6 years older than the victim and the victim is at least 14 years old and if the age of the victim was the basis for the conviction. Page 16 expire). Since the report that is generated to identify offenders whose registration is due to expire is based on the end of registration date, we believe that these offenders would likely have had their registration prematurely ended. Moreover, there were an additional 38 offenders that were listed as lifetime registrants that did not meet the statutory requirement for this designation. These offenders had out-of-state convictions. As a result of these various types of discharge date/end registration date errors, some offenders had their registration requirements extended beyond the statutory requirement or were prematurely removed from their registration obligations. For example, • 10 offenders were registered for a year or more past the statutory requirement. • 12 offenders had their registration prematurely expired (i.e., were no longer registered) by VCIC. The premature expiration of registration for 11 of these offenders ranged from 4 months to over 7 years. There was also one offender who should have been listed as a lifetime offender but whose record had been expired by VCIC. In other cases, the discharge and end registration dates that were in error had not yet resulted in offenders having their registration requirements extended beyond the statutory requirement or prematurely removed from their registration obligations. Nevertheless, this would have occurred had the errors identified in the audit not been corrected by VCIC. Recently VCIC and DOC have implemented some changes in their processes related to discharge dates. For instance, in early 2010 a new SOR form was added (Change of Treatment Status * Supervision Status) that now contains fields to indicate a parole/furlough date of expiration and a probation closed date. DOC also recently modified its internal directive related to the SOR (effective June 14, 2010), which requires that caseworkers or POs submit this form within 24 hours for offenders who have maxed out, have had their probation discharged by the Court, or have had their parole or sentence expire. The SOR coordinator also told us that she is more proactively seeking discharge information from other states’ law enforcement authorities and sex offender registries for sex offenders living or convicted out of state. Internet Status 13 VSA §5411a sets requirements for certain offenders to be posted to the Internet SOR. For example, offenders who have been convicted of sexual Page 17 assault, lewd and lascivious conduct with a child, or sexual exploitation of a minor are required to be posted to the Internet. Other criteria for Internet posting includes, but is not limited to, offenders with multiple convictions or those who have been designated as high risk by DOC. Our statistical sample included two of 57 offenders (4 percent) that were posted to the Internet, but did not meet the statutory requirements for such posting. In both of these cases there were errors in the offense description in the SOR which erroneously showed them to have been convicted of offenses that qualified them for being posted to the Internet. In addition, our automated data analysis found another 14 offenders who were erroneously posted to the Internet. Ten of those offenders were wrongly identified as eligible to be posted for the reason of having a qualifying offense (they also did not meet any other criteria for Internet posting). The other four offenders erroneously posted to the Internet SOR had offenses that met the statutory requirement for being on the Internet SOR, but they also had an outstanding petition requesting a waiver for such posting. Offenders were not supposed to be on the Internet SOR while their petitions were being reviewed.27 The records of the offenders’ that were not eligible to be posted to the Internet SOR were removed when it was brought to VCIC’s attention. While only four offenders requesting an Internet waiver were on the Vermont SOR site, the names and city/town of all 174 offenders that have outstanding waiver requests were posted on the national sex offender website between October 1, 2009 and May 17, 2010. After we brought this situation to the SOR system vendor’s attention, we were told that it was caused by a system error in that the selection logic used to exclude these offenders from the Vermont SOR website was not applied to the process that extracts offenders for the national site. The vendor fixed the system error and the offenders with outstanding waiver petitions were removed from the national sex offender website. There were also four offenders that should have been posted to the Internet, but were not. For example, two offenders who were convicted of lewd and lascivious conduct with a child and sexual exploitation of children, respectively—offenses required to be posted to the Internet under 13 VSA 27 In addition to expanding the list of sex offenses that would qualify for posting to the Internet SOR, Act 58 allowed certain categories of sex offenders to petition the DOC Commissioner to remain off the Internet SOR. Eligible petitioners were supposed to remain off the Internet SOR until their petition was reviewed and a final decision was made by the DOC Commissioner. As of June 2, 2010 final decisions on offender petitions had not been made. Page 18 §5411a(a)—were not on the Internet SOR as of February 18, 2010. In addition, two offenders were erroneously marked as having outstanding eligible petitions to the DOC Commissioner to remain off the Internet SOR. These four records were subsequently posted to the Internet. In addition to whether an offender was appropriately posted to the Internet, our statistical sample of community-based offenders found numerous instances in which the information on the Internet was incorrect or incomplete. For example, there were four instances (12 percent of those offenders in our sample that were on the Internet) in which the Internet SOR showed that the offender had been convicted of multiple counts when the offender had only been convicted on one count. According to VCIC’s SOR system vendor, this is the result of a system glitch that causes some duplication on the Internet SOR, which has been known about for some time. He added that a workaround had been implemented to avoid this problem. This workaround did not appear to be effective in all cases. The Internet SOR also often did not include an offender’s middle name or name suffix (e.g., Jr.). In the case of an offender’s middle name, the SOR was missing information that was generally readily available in VCHIP. In the case of the name suffix, the current SOR system does not have a field on the input screen to capture this information. Name suffixes used to be collected in the SOR so some offenders’ records included this data. Nevertheless, even for these offenders the name suffix was not included in their Internet SOR record. It appears that the usage of the name suffix data element was dropped years ago when the original SOR system was replaced. Having the full names of offenders available to the public is important to ensure that the right person is identified as a sex offender (or not erroneously identified as a sex offender). Another example in which the Internet SOR did not accurately reflect supporting documentation pertained to risk designations. 13 VSA §5411a(b)(12) requires that the Internet record of applicable offenders who have not undergone a risk assessment by DOC include language stating that they are presumed to be high risk. The risk designation information in the SOR for 20 offenders in our sample (35 percent) was incorrect or omitted. For those offenders who were posted to the Internet, (1) two had the presumed high risk language included in their record even though they had undergone a risk assessment and were not considered high risk by DOC and (2) eight did not include the presumed high risk language on their Internet records even though they had not undergone a risk assessment. Page 19 Age of Victim Act 58 requires offenders who meet certain requirements to have their address posted on the Internet as early as July 1, 2010. In particular, offenders who are on the Internet SOR whose victims were younger than 13 will be required to have their addresses posted. Accordingly, a field for victim ages was added to the SOR data entry screen in 2009 and VCIC has been in the process of collecting victim ages through inquiries of DOC, local law enforcement, and other research. Our community-based statistical sample of 57 records found that the victim ages were incorrect or omitted when documentation was available that had this information in nine (16 percent) and 12 cases (21 percent), respectively. In most cases, corrections or additions to the SOR related to the victim’s age would not have had an effect on how the offender was treated with respect to having his or her address posted on the Internet. However, in some cases, if the SOR had not been corrected, offenders who should have had their addresses posted to the Internet would not have had them posted. For example, ● An offender was convicted of sexual assault on a minor younger than 16. The SOR showed the victim’s age as 15, but there was no documentation in the VCIC files that indicated the origin of this age. We found a copy of the investigator’s report in the offender’s DOC file that indicated that the victim was 12 years old at the time of the incident. The confusion in this case may have occurred because the same document recounted an incident in which the victim was identified as 15 years old. ● An offender who was convicted of aggravated sexual assault and lewd and lascivious conduct with a child did not have the victim’s age entered into the SOR. According to an October 2004 registration form, the victim was 10 years old. In addition, one offender would have had his address posted because DOC reported to VCIC that the victim was 10 years old and this age was recorded in the SOR, but the documentation in the offender’s DOC file did not support this age. Some Tests Yielded Few Errors While we found numerous critical and significant errors in our statistical sample and automated data testing, some of the tests that we ran resulted in no or few errors thereby indicating that these processes were working Page 20 relatively well. For example, we validated that the number of offenders listed on the Internet SOR equaled the number of offenders listed on the electronic copy of the SOR that we received February 18, 2010 that were coded as requiring posting to the Internet. In addition, the SOR contained almost all of the Vermont convictions for sex offenses defined in 13 VSA §5401(10)(A) and 13 VSA §5401(10)(B) in which the offender was sentenced between January 1, 2007 and December 31, 2009. During this 3-year period VCIC recorded in the SOR all but one of the 309 offenders that had been sentenced during that timeframe for an offense that met these statutory criteria (99.7 percent complete). There was a slightly higher number of offenders whose records did not reflect all of the sex offense charges for which they had been convicted—eight of 309 offenders (but 97.4 percent of records had the correct number of charges). According to the SOR coordinator, in most cases the reason for the discrepancy in the number of charges was an incomplete transfer of conviction information from VCHIP to the SOR. Lastly, we extracted a statistical random sample28 of 53 incarcerated offenders (i.e., those in inactive status) from the February 18, 2010 SOR file to confirm that their status was correct. There were two offenders (4 percent) that were listed in inactive status in the SOR, but who were not in jail as of February 18, 2010. One offender’s registration obligation had expired in 2009 while he was incarcerated on non-sex offense charges so his status should have been in “reg exp” rather than inactive status. The second offender was released on probation on February 14, 2010 and his status should have been changed to active. The DOC PO told us that it was an oversight that she did not send VCIC a notice that this offender had been released. Registration Process Does Not Lend Itself to Timeliness Analysis The SOR statute and the DPS rule related to the operation of the SOR contain several timeframes for when activities need to be performed. For example, both the statute and the DPS rule require that the courts submit sex offenders’ conviction records within 10 days after sentencing. In addition, the rule requires DOC to submit changes in offender residential addresses, employment, and school enrollment to VCIC within 24 hours for those 28 We used statistical random attribute sampling with a 95 percent confidence level, and 5 percent tolerable rate of records not in compliance with the statutorily established requirements and zero expected population deviation rate (i.e. expected error rate). Page 21 offenders under supervision. Offenders that are no longer under DOC supervision are required to submit such changes to VCIC within 3 days of the change. The process used by the State to establish, maintain, and remove offenders from the SOR did not allow us to perform a systematic analysis of whether the timeframes related to certain data being reported to VCIC are being met. Specifically, • The origination and maintenance of offender records in the SOR is largely a manual process in which data is generally received via U.S. mail, fax, and email and entered by VCIC staff (either directly into the SOR or input into VCHIP and then downloaded into the SOR). Only a few dates, such as the date a record is first established, are automatically generated by the system and all of them can be modified manually. • The SOR system and forms do not capture dates that would allow some types of timeliness analysis. For example, the system does not include dates of when changes are made so it is not possible to assess whether data entry was close to the date of receipt of the information. In addition, the forms do not always include information as to the date of the event (e.g., when the address changed), just the date that the form was signed or received by VCIC. • Incomplete date fields in the system and errors in dates in the system found during our testing reduced our confidence that accurate conclusions could be drawn based on the data in the system alone. For example, the dates that an offender was released from incarceration and when he or she started SOR registration are important to determining whether registration forms are being submitted in a timely manner. However, the release date field in the SOR was generally empty and we were told that this field is no longer used. Moreover, our statistical sample of community-based offenders found 14 errors (25 percent) in the start registration date. Some of the dates were off by only a period of days, but others were different by many months. While we do not generally consider an incorrect registration start date to be an important error by itself, the number of errors meant that we could not rely on this date in the system to be sufficiently accurate to draw conclusions. In addition, VCIC does not have a written standard with which to judge the timeliness of data entry into the SOR. Specifically, DPS’s current SOR rule (which is in the process of being revised) does not specify how quickly data should be entered into the SOR. According to VCIC management, they have Page 22 an unofficial standard of entering data on the same day that it is received. Nevertheless, an explicit performance measure related to the timeliness of SOR entry that is tracked or periodically checked would provide greater assurance that information is being entered into the SOR in a timely manner. Although we could not perform a systematic analysis of timeliness, the system had data that could be used to analyze the extent to which offenders were returning annual verification letters.29 Specifically, 13 VSA §5407(g) requires VCIC to verify the residential addresses of most offenders annually.30 VCIC sends a verification letter to offenders within 10 days of their birthday and offenders must sign and return the form within 10 days. This form includes the offender’s residential and mailing addresses, employer name and address, and school name and address. As shown by Figure 4, the offenders who were sent the verification letters generally returned them within 20 days of their birthday (85 percent of the time for verification letters sent out between February 18, 2009 and January 18, 2010). 29 To perform this analysis, we concluded that the offender’s date of birth and annual verification receipt date in the system were sufficiently reliable for the purpose of this analysis. We based this conclusion on the results of our community-based offender statistical sample, in which we found (1) no records with an error in the date of birth field and (2) four records with errors in the annual verification receipt date field (7 percent). 30 VCIC sends out annual verification forms to registrants that are in the active, homeless, and student/employed statuses. These forms are not sent to registrants that are in pending, out-of-state, inactive, MIA, or “reg exp” status. Page 23 Figure 4: Timeframes for VCIC Receipt of Offender Annual Verification Letters a (sent between February 18, 2009 and January 18, 2010)b 1,200 1,000 800 600 400 200 0 1-20 days 21-40 days 41-60 days over 60 days Unreturned a In some cases, VCIC received confirmation of the offender’s address through a confirmation letter or other means rather than through the annual verification letter. Those offenders are not part of this analysis. b We did not include annual verification letters sent after January 18, 2010, because (1) some offenders would not have had the opportunity to submit their signed forms and (2) VCIC may not have had the time to input the date of receipt of the forms. Of the 30 offenders who did not return their verification letters, five offenders were subsequently located. The remaining 25 offenders had addresses in the SOR that remained unverified as of the end of April 2010. This number of offenders with unverified addresses is in addition to the 75 offenders listed in MIA status in the SOR as of February 18, 2010. Significant Control and System Limitations Warrant Urgent Attention The controls and systems that are in place to prevent errors, omissions, and outdated SOR data are limited. In particular, there were control weaknesses in the processes that affected (1) how long offenders were to be registered and (2) whether offenders are eligible to be posted to the Internet SOR and, if so, what data is posted. Moreover, the process used to initiate and maintain sex offender records in the SOR is largely manual and the SOR system does not have features that would prevent basic errors (e.g., incorrectly calculated dates) or aid in their identification and correction when they occur. Page 24 Some Processes Have Not Been Consistently Implemented or Documented The reliability and timeliness of offender data in the SOR rests on the effective implementation of processes and controls at the Courts and DOC, as well as at VCIC. Some of the processes utilized by these organizations to collect and submit data to the SOR were not always (1) reflective of SOR requirements, (2) documented, or (3) consistently applied, as highlighted below. Offender Supervision Changes The status of an offender’s DOC supervision is critical. It determines an offender’s SOR status category, which, in turn, can affect whether an offender is posted to the Internet SOR or is subject to the annual verification process. Some offenders were incorrectly listed as being in pending or inactive status when they had been released from incarceration into the community and so should have been listed in active status. Because DOC had not notified VCIC of their change in supervision status, these offenders were not on the Internet SOR. In addition, other offenders were incarcerated, but were listed in active, rather than inactive, status. DOC’s ability to notify VCIC of changes in an offender’s supervision status is hindered by its system, which does not explicitly flag offenders who are, or should be, registered with the SOR. One area in which offender status notification is particularly important is when the offender is discharged from DOC supervision altogether. This date is key to determining how long an offender is required to be registered. DPS’s SOR rule requires that DOC provide VCIC with this date. However, until recently, DOC’s directive that lays out its internal procedures related to the SOR did not include this requirement. In addition, until recently there was no form or other formal mechanism in place to communicate discharge information to VCIC. Instead POs told us that they generally called or emailed the SOR coordinator with this information. This informal process was not effective. For example, 16 of the 57 SOR records of our statistical sample of community-based offenders (28 percent) had discharge date errors (inaccuracies or omissions). DOC and VCIC have taken steps to improve the process used to report discharge dates. In particular, in February this year a new form was introduced in which this information is now explicitly required to be submitted to VCIC. For those offenders still under DOC supervision, another supervision status area in which we found errors was in the name of the responsible PO and DOC field office. In our community-based statistical sample, there were five Page 25 PO name errors (9 percent) and two DOC field office errors (4 percent). The PO name is important because this is the SOR coordinator’s primary source of information on an offender supervised in the community. In addition, the DOC field office and phone number is posted on offenders’ records in the Internet SOR as the contact point for further information on an offender. In our discussions with POs, it was clear that they were not sure of their responsibilities related to notifying VCIC of such changes. When asked who was responsible for notifying VCIC that the PO or field office had changed, five responded that it was the sending PO, five stated that it was the receiving PO, one said both the sending and receiving PO were responsible, and three stated that they did not know. DOC recently issued guidance to its staff that clarifies that it is the sending office that is responsible for the VCIC notification. Sex Offender Treatment Status Reporting Effective March 1, 2005, 13 VSA §5411a (a)(5) required that offenders who do not comply with sex offender treatment or are ineligible for such treatment to be posted to the Internet SOR. DOC’s directive that addresses its responsibilities related to the SOR lays out the process of providing treatment information to VCIC. Specifically, the directive assigns such responsibility to the correctional facility caseworkers or POs depending on the offender’s supervisory status, or to offenders themselves, in cases when they have been discharged from supervision, but have not completed sex offender treatment. For the supervised offenders, the DOC Directive requires DOC personnel to notify VCIC of the initial determination of treatment compliance status. Offenders no longer under DOC supervision are required to provide monthly updates of their treatment status to VCIC.31 Once treatment compliance information is received at VCIC, it is entered in the SOR and is used to determine the Internet posting status of sex offenders. The Internet SOR has four treatment compliance statuses – compliant, noncompliant, not applicable, and unavailable. The not applicable status is designed to be used for sex offenders that were convicted prior to March 1, 2005 and were not under DOC supervision as of March 1, 2005. The unavailable status is used for offenders for whom treatment compliance is applicable but treatment information has not been received by VCIC. 31 Treatment information that comes from offenders is required to be certified by their treatment providers. Page 26 The SOR contained 79 offenders who met the statutory criteria for having their treatment compliance reported, but whose treatment status was marked either as not applicable or unavailable.32 In up to 21 of these cases, treatment information could result in the offender being posted to the Internet SOR when they are not otherwise eligible, if they have not complied with their treatment requirements. In the remaining cases, the offenders are already posted to the Internet SOR, but their treatment information may need to be updated to indicate either compliance or noncompliance. The SOR coordinator explained that treatment compliance information in the SOR has been difficult to accurately maintain in part because of misunderstandings related to the interpretation of the data that DOC was providing. For example, in the past when DOC caseworkers indicated that an offender leaving prison had completed treatment, the SOR coordinator interpreted this to mean that the offender’s treatment obligations were done when in fact the caseworker may have only meant that the treatment requirements related to the offender’s incarceration period had been completed. The coordinator did not realize until recently that an offender may have continuing treatment responsibilities after release. In addition, less than half of the VCIC offender files reviewed as part of the community-based statistical sample included the forms that are supposed to be submitted by DOC to inform VCIC of an offender’s treatment status, namely, the treatment compliance and noncompliance forms. The lack of these forms can hinder VCIC’s ability to track and accurately report a sex offender’s treatment status. For instance, in one case DOC submitted to VCIC the required noncompliance checklist in January 2008 for an offender. The offender later became compliant with treatment, but a treatment compliance form was not sent to VCIC to update his status. Instead, in the fall of 2009, the offender himself contacted VCIC and provided documentation to support that he was wrongly listed as noncompliant with treatment, namely a letter from a treatment provider supporting his compliance claim.33 Based on this information, VCIC changed his record to reflect that he was now compliant.34 Nevertheless, since the PO did not submit the compliance checklist to VCIC, it caused the offender’s Internet 32 Although the treatment status in these records was incorrect, we did not ascertain what their correct status should be. As a result, we did not include these 79 records in our count of critical and significant SOR errors. 33 We confirmed with the PO that the offender is currently in compliance with treatment requirements. 34 Although the record was changed to indicate that the offender is now compliant with treatment, other related codes were not changed in the SOR because of an oversight, which caused the offender to be listed on the Internet SOR as both compliant and noncompliant. Page 27 record to be wrong for a period of time. The PO told us that he was uncertain when the treatment compliance/noncompliance checklist is required to be submitted. The inconsistency of the application of DOC’s treatment notification process was confirmed by the POs that we interviewed. In particular, about half of the 14 POs we interviewed stated that they send the treatment compliance and noncompliance forms required at the onset of the community supervision process less than 25 percent of the time. Moreover, five POs stated that they were either unaware of the form or did not know when these forms were supposed to be submitted to VCIC. These PO responses may be because DOC’s directive did not clearly indicate when treatment status information needed to be sent to VCIC. DOC recently clarified the treatment reporting requirements in its revised directive.35 In the summer of 2009, the lack of updates on sex offenders’ treatment statuses caused VCIC to request that DOC research and submit the compliance status for a lengthy list of sex offenders for whom treatment information was either missing or appeared to be outdated. As the result of this data call, the treatment information in the SOR at that particular point of time was updated. However, in the absence of effective ongoing DOC treatment information update process; there is no guarantee that sex offenders’ treatment statuses in the SOR will remain current and accurate. It is too early to tell whether the revised DOC SOR directive will result in more reliable treatment data in the SOR although the SOR coordinator reported that she is receiving many more treatment compliance/noncompliance forms recently. VCIC also does not have an effective process to track treatment information for offenders who are no longer under DOC supervision. For instance, one offender was discharged from DOC’s supervision in July 2008 while still in a treatment program. He submitted a certification of compliance36 with sex offender treatment to VCIC in late July 2008. Subsequent to that certification VCIC did not receive any information on his treatment status until early March 2009 when VCIC found out that the offender was noncompliant with treatment and changed his SOR record to reflect this status. Later in March 2009 VCIC received another certification with treatment, indicating that the 35 The new directive was approved May 10, 2010 and was effective June 14, 2010. 36 A certification of compliance is a form that a non-supervised offender who is still undergoing sex offender treatment is required to submit to VCIC on a monthly basis to attest to his/her treatment compliance. The certification is required to be signed by the offender and the treatment provider. Page 28 offender had re-entered the treatment program. VCIC’s files did not indicate that it had received any information on the offender’s treatment status subsequent to March 2009 until we inquired about his status in April 2010. Such a large time gap in the documentation of treatment status is not necessarily an indication of an error. However, gaps in when a nonsupervised offender reports his treatment status to VCIC can indicate that he or she is no longer attending treatment. At this time VCIC does not have a process for tracking the treatment status for offenders who are no longer under DOC supervision and have not completed sex offender treatment. According to the SOR coordinator, she does not make inquiries of sex offenders or providers in order to obtain information on treatment status to check that the offender is remaining current. In addition, VCIC does not have criteria to determine when a previously compliant offender becomes noncompliant (i.e., such as not submitting a certification of compliance form as required to remain in good standing). As a result, the current treatment status of non-supervised offenders that had not completed sex offender treatment while under DOC supervision may not be accurately captured in the SOR. Sex Offender Photo Updates On April 15, 2009, VCIC sent a letter to offenders who were in active, homeless, and student/employed status that, pursuant to the SOR statute, they were required to provide a photograph to the registry each year during the annual verification process. The April letter instructed offenders that if they were still under DOC’s supervision that the POs would take the photographs. Offenders no longer under DOC supervision were directed to go to the nearest Vermont State Police barrack. Of the 55 offenders in our community-based sample that this requirement applied to, almost half did not have a picture on the SOR taken in the timeframes set forth in VCIC's April 2009 letter (10 were taken before 2008). In the case of the offenders with outdated photographs who were still under DOC supervision, POs provided various explanations as to why more recent photographs were not sent to VCIC. For example, some did not realize that VCIC was expecting a photograph while others stated that it was an oversight. For those offenders who were not under DOC supervision, the SOR coordinator did not systematically check the State Police system that contains photographs to determine that some offenders had not fulfilled their obligation to obtain a new photograph. The SOR coordinator stated that as part of the annual verification process she is now confirming that offenders have had the required photographs taken. Page 29 Since an offender’s physical appearance can change over time, it can be important to have up-to-date photographs. In addition, to aid the public in assessing the age of photographs posted on the Internet, some states include the date that the offender’s photograph was last taken on their Internet registries. High-risk Designations for Women 13 VSA §5411b requires DOC to evaluate sex offenders to determine whether they are high risk for the purpose of identifying those offenders who should be subject to increased public access to his or her status. For example, an offender who has been designated as high risk is posted to the Internet SOR. DOC does not perform risk assessments for women offenders and, therefore, none have been designated as high risk. According to the DOC codirector of the Vermont Center for the Prevention and Treatment of Sexual Abuse, the tools used to perform risk assessment of men are not validated for use in assessing women offenders. DOC has not developed an alternative risk assessment process for women. Although there are far fewer women sex offenders in the SOR than men (as of February 18, 2010, the SOR included 55 offenders whose gender was identified as female), the SOR statute does not limit this requirement to male offenders. Noncompliant High-Risk Offenders 13 VSA §5411d(e) requires DPS to conduct periodic unannounced registry compliance checks on noncompliant high-risk offenders37 to verify the accuracy of registration information.38 DPS’s SOR rule does not address this requirement because it was issued prior to the law that established this requirement (this rule is in the process of being updated). Nevertheless, VCIC officials informed us that they fulfill this obligation by relying on “sweeps” periodically conducted by law enforcement agencies in which sex offenders’ addresses are physically checked (not just those for noncompliant high-risk offenders). We were told that prior to conducting a 37 13 VSA §5411d defines noncompliant high-risk offenders as those who (1) were incarcerated on or after June 7, 2007 for lewd and lascivious conduct with a child, sexual assault, aggravated sexual assault, or any attempt to commit these crimes or a comparable offense in another jurisdiction in the United States, (2) are not subject to indeterminate life sentences under 13 VSA §3271, (3) are designated as a high-risk sex offender, and (4) are noncompliant with sex offender treatment. 38 As of February 18, 2010, there were seven offenders that have been designated as noncompliant high-risk. Page 30 sweep; a law enforcement agency will request offenders’ address information from the SOR coordinator in a particular geographical area. In addition, the SOR coordinator stated that she generally receives the results of sweeps from the law enforcement agency (e.g., a new address for an offender), which she uses to update the SOR. However, VCIC does not keep this documentation. Accordingly, VCIC did not have evidence to support that these periodic sweeps included all of the noncompliant high-risk offenders. Largely Manual SOR System/Process Not Conducive to Preventing Errors The SOR system was initially developed in-house, but a new system was purchased from a vendor, Computer Projects of Illinois (CPI), in 2003. However, this new system was simply the import of the functionality of the old system to a new system platform. In addition, while the SOR has changed since 2003 as new functionality has been built (e.g., programming for the Internet SOR was added in 2004), the system’s base design and underlying capability is limited. One of the areas of limitation is the lack of electronic interfaces with other systems. While the data used in the SOR comes from a variety of sources, the only electronic interface that provides input into the SOR is the transfer of offender identification and conviction data from VCIC’s VCHIP system. However, even the process of entering sex offender conviction information into VCHIP is largely manual. Specifically, a VCIC clerk manually enters data from Charge Disposition Reports submitted by the 14 District Courts for offenders convicted of sex offenses into VCHIP and clicks on a button on the screen to send that individual’s record to the SOR. If the SOR coordinator accepts the record, the VCHIP data automatically populates the SOR. This manual process is in place even though the Court Administrator’s Office sends electronic files of its criminal charge dispositions to VCIC. However, since these files are only sent once a month, VCIC chooses to perform manual data entry into VCHIP to update the SOR in a timelier manner. According to the Court’s Director of Research and Information Services, VCIC has requested more frequent electronic submissions, but the Court is in the process of replacing its current case management system, which is a higher priority. Another major source of data for the SOR is DOC. However, there is not an electronic interface between the DOC systems and the SOR. Among the barriers to implementing such an interface, are that (1) DOC and VCIC use different codes to identify offenders, (2) the DOC system does not explicitly identify registered sex offenders, (3) the DOC system is old and not easily Page 31 modified, and (4) the SOR does not have the ability to easily interface with outside systems while ensuring data integrity. Nevertheless, there may be alternatives that could be considered to more effectively submit data to VCIC and lower the possibility of error. For example, based on our observation of the forms that are submitted, most are filled in by hand and, according to the POs we interviewed, are mailed or faxed to VCIC. Even if an interface between the DOC system and the SOR is not feasible at this point, other transmission alternatives that may be possible could include the use of automated forms and/or the electronic submission of such forms to the SOR. In our discussions with POs, the most common response that we received to our question as to what actions could be taken to improve the accuracy, completeness, and timeliness of the SOR was the automation of the process of sending information to VCIC and other electronic sharing of data between DOC and VCIC. In addition to the lack of electronic interfaces, the SOR system does not utilize important business process controls39 that are designed to prevent errors. In particular, the SOR does not have effective data validation and edits built into the system. According to the U.S. Government Accountability Office (GAO), input data should be validated and edited to provide reasonable assurance that erroneous data are prevented or detected before processing.40 Edits are used to help assure that data are complete, accurate, valid, and recorded in the proper format. For example, systems can have edit controls that check for (1) missing data, (2) erroneous dates, (3) limits and reasonableness, (3) valid, or invalid, combinations of related data field values, and (4) mathematical accuracy. Many of the errors that we found in the SOR could have been prevented with the application of such edit controls. Our automated data analysis found: ● ● ● ● Duplicate or missing Social Security Numbers, Duplicate state identification numbers, Missing dates, and Illogical combinations, such as offenders designated as lifetime registrants in the system who also have a date in the end registration date field. 39 Business process controls are the automated and/or manual controls applied to business transaction flows and related to the completeness, accuracy, validity, and confidentiality of transactions and data during application processing. 40 U.S. Government Accountability Office, Federal Information System Controls Audit Manual (FISCAM), (GAO-09-232G, February 2009). Page 32 In addition, the SOR system is not programmed to make, or at least assist, in decisions related to an offender’s registration record. For example, the system does not calculate the end registration date for 10-year registrants; instead the SOR coordinator manually performs and enters the calculation. There were 13 records with calculation errors in the SOR as of February 18, 2010, which caused their end registration dates to be incorrect—sometimes by many years. In addition, the system does not have logic that identifies those offenders who meet the criteria for being posted to the Internet SOR. Instead, the SOR coordinator manually enters the codes that post an offender on the Internet. If the codes are not added, which occurred in a few cases, the offender would not to be posted to the Internet. Conversely, in 12 instances, codes were erroneously added that caused offenders’ records to be incorrectly posted to the Internet. Edit controls can be utilized to prevent these types of errors. Another type of business process control is automated audit trails. An audit trail shows who has accessed a system and what operations the user has performed during a given period of time. According to GAO, data that should be included in an automated audit trail are who initiated each of the transactions, the date and time of the transactions, and the location of the transaction origination.41 The SOR system does not track what changes to the data in the system have been made, when, and by whom. Without such information it can be very difficult, if not impossible, to link a specific action in the system to a user. In addition, because the SOR does not utilize an audit trail that indicates when and what adjustments to an offender’s record were made, with the exception of residential address data, there is no historical information available in the system. The lack of effective SOR business process controls is of particular concern because of VCIC’s overreliance on a single person to appropriately apply, without much system support, complicated requirements, such as which offenders should be on the SOR, posted to the Internet SOR, and how long registration should last. VCIC’s SOR function is staffed by a full-time employee (the SOR coordinator) who also has other duties, most notably the management of the Medical Marijuana Registry. The coordinator spends about 75 percent of her time on the SOR and is assisted by another VCIC employee who works 20 hours a week. The staff member who provides assistance performs largely clerical tasks and does not make some of the 41 U.S. Government Accountability Office, Federal Information System Controls Audit Manual (FISCAM), (GAO-09-232G, February 2009). Page 33 critical decisions related to an offender’s record, such as whether he or she meets the criteria to be a lifetime registrant. In addition, VCIC’s SOR procedures do not include some significant functions or what documentation should be retained. For example, the procedures do not address determining an offender’s (1) registration eligibility, (2) status category (or how they differ in terms of requirements), (3) end of registration, or (4) treatment status. These documentation weaknesses, along with the system’s inability to identify common errors or guide the user through the decision-making processes involved with an offender’s registration, means VCIC’s reliance on a single individual to make these decisions is a high-risk approach. VCIC has recognized its vulnerability and its need to implement a more highly automated system and has applied for a $150,000 federal grant to replace the current system. According to VCIC’s grant request, a new system would (1) contain the most current, efficient, and versatile interfacing tools and (2) minimize the amount of human intervention required in order to minimize the chance of data entry and transcription errors. We agree that a new system with more electronic interfaces and business process controls would help prevent inaccuracies, omissions, and untimely VCIC data. However, layering a new system over the current process is unlikely to fix the all of the many and varied types of problems that have adversely affected the reliability of the data in the SOR since the causes of errors were not solely system-related. For example, the SOR statute does not require offenders to register if they were under the age of 18 at the time of the crime, the victim was at least 12 years old, and the conduct that was criminal was only because of the age of the victim. This set of information is difficult for VCIC to discern because it does not routinely receive detailed information on the offense that was committed, including the date of the offense and the age of the offender as of this date. During the course of the audit, we found this information in documents filed with the Courts42 and contained in DOC files, but neither of these organizations routinely provides this information to VCIC. This example also highlights that the reliability and timeliness of the SOR is dependent not only on VCIC, but also on information, processes, and systems utilized by other entities. Taking a broader view of what currently constitutes the SOR process to include other organizations could help VCIC address not only reliability 42 Court procedures state that two of the types of documents that we used to determine offender and victim ages (dockets and affidavits) are supposed to be supplied to VCIC within 10 days of sentencing, but the Courts were not consistently sending these documents. Page 34 concerns, but could have the added benefit of improving the efficiency of the process. A comprehensive approach to developing a new SOR process/system could even include recommending changes to the procedures required by the Registry’s current statutory framework in light of the technological advances that have occurred since this framework was first put into place in 1996. For example, since the inception of the statute that instituted the SOR, the courts have been required to submit the Order to Provide Information to VCIC for sex offenders defined under 13 VSA §5401(10). However, according to the SOR coordinator, she does not generally rely on the Order to Provide Information and often did not even receive them from the Courts. Indeed, the courts were not sending this information to VCIC consistently in part because the Court system did not prompt the clerk to print the Order to Provide Information form for all offenses listed under 13 VSA §5401(10).43 Accordingly, alternatives to the current Order to Provide Information process could be considered, such as requiring the offender to acknowledge his or her obligations to report to the SOR at the time of conviction or sentencing so that if a registration form is not later submitted for this offender, he or she has been notified of the obligation to report. The Court Administrator’s Office is currently in the process of implementing a new system so it would be prudent to consider changes to the process now rather than after the implementation is complete. Other critical considerations in the development of a new system are the importance of analyzing and documenting requirements prior to its development and security. In layman’s terms, requirements development and management involves establishing and maintaining agreement on what the system is to do (functionality), how well it is to do it (performance), and how it is to interact with other systems (interfaces). Without such an analysis, an organization risks that the new system will not fully meet its operational and performance needs. With respect to security, the current SOR system does not meet DPS’s or the Federal Bureau of Investigation’s password policies related to password length and complexity. According to the DPS policy, passwords are the foundation of the security of DPS’s automated systems. The VCIC deputy director told us that VCIC recognized this deficiency in the summer of 2009 and requested that CPI change the password requirements of the system that authenticates users of VCIC systems, including the SOR. As of May 24, 2010, this system change request remained outstanding. 43 The Court reported that it has fixed this system glitch and the SOR coordinator told us that VCIC is receiving more of these forms than it has in the past. Page 35 Observations Pertaining to the Posting of Sex Offender Addresses on the Internet Some members of the General Assembly have expressed reservations about the accuracy of offender addresses that may be posted on the Internet SOR as early as July 2010.44 Our statistical sample of community-based offenders did not disclose a substantial number of errors pertaining to residential addresses (5 percent).45 However, it is important to note that this analysis was limited to checking whether the address reported by the offender was accurately entered into the SOR. Our analysis does not assess whether the addresses accurately reflect where the offenders reside. Because of the importance of this issue, we make the following observations regarding the process in place to provide and update offender addresses: ● 13 VSA §5407(a) requires most offenders46 to report their new addresses within 3 days after any change in residence. However, except for offenders that are under DOC supervision, there is no mechanism to systematically validate and document that the offender actually lives at that address or even that the address is valid. VCIC staff members explained that law enforcement periodically performs sweeps of sex offenders in specific geographic areas. VCIC provides them lists of the offenders in the area being checked. Although the law enforcement agency may send VCIC the results of the sweep (i.e., offenders who were not found at the address reported in the SOR), VCIC does not keep this documentation. Without this documentation, it is not possible to know how often the residences of sex offenders are physically checked or how often the residence address in the SOR is wrong. 44 Act 58 requires that the following offenders have their addresses posted to the Internet SOR: offenders who (1) have been designated as high risk, (2) have not complied with sex offender treatment, (3) have an outstanding warrant for their arrest, (4) are subject to the registry for a conviction of a sex offense against a child under 13 years of age, and (5) have been electronically posted for an offense committed in another jurisdiction which required the person’s address to be electronically posted in that jurisdiction. 45 In addition to the address errors found in the community-based statistical sample, the automated data analysis found a few other records in the SOR that had incorrect residential addresses. 46 13 VSA §5407(a) requires offenders who have been designated as high risk by DOC to report within 36 hours and 13 VSA §5411d requires noncompliant high-risk offenders report changes in residence prior to the change being made. If the change is unanticipated, noncompliant high-risk offenders must report within 1-day of the change. Page 36 ● In the case of offenders under DOC supervision, the POs that we talked to generally reported that they routinely physically verify (or have another DOC or law enforcement official physically verify) the residential addresses of offenders. Our statistical sample of communitybased offenders included 22 offenders under DOC supervision. POs provided evidence in 20 cases (91 percent) that the offender’s residence had been physically checked. These checks had occurred from 1 day to about 11 months prior to the date of our visit (half of the checks had occurred within a month of our visit). ● 13 VSA §5407(g) sets out the requirements for the annual verification process in that it requires that DPS send out a non-forwardable address verification form to each registrant at the address last reported by the registrant. Accordingly, VCIC sends out a non-forwardable annual verification letter to offenders at the time of their birthdays that they must sign and return indicating that their residential address has not changed or providing their new address. However, this letter is sent to offenders’ mailing addresses, which can be a post office box or in care of another individual. As a result, although non-forwardable, these letters are simply a self-certification that the offender lives at an address and does not necessarily indicate that mail has been delivered and received at his or her physical address. ● VCIC’s SOR rule states that offenders who move out of state have completed their Vermont registry requirements (although it is expected that they would notify VCIC if they returned to Vermont). Accordingly, these offenders are not required to submit changes of address and the annual verification letter is not sent to them. Such offenders can be posted to the Internet SOR and we have been told that their addresses will be included on the Internet if they meet the statutory requirement for such disclosure. Since these offenders are no longer tracked by VCIC their addresses could be years out-of-date. One way that other states that post addresses to the Internet have used to inform the public of the reliability of the addresses that are on their Internet sites is to include the date that the address was last verified or updated. This type of information could help mitigate misunderstandings in the public if an offender’s listed address is not the one that he or she is actually using. In addition to the process that VCIC uses to obtain and maintain valid offender addresses, another critical aspect to ensuring that the new requirements are properly implemented relates to the SOR system. As of Page 37 mid-May 2010 the system changes related to the new address requirements had not been tested by VCIC. According to VCIC’s deputy director, the Internet SOR is being transitioned to a new website. Once this occurs testing of the system’s functionality related to addresses will commence. Given that system glitches have contributed to some of the problems outlined in this report, it is important that this testing be robust and thorough. Conclusion We believe that the errors we found are preventable with stronger controls and a more effective system. In particular, the large and diverse number of errors found in the SOR warrants a fresh look at how convicted sex offenders are identified and how data is collected and submitted to VCIC. The SOR process relies on three primary organizations, VCIC, DOC, and the Courts to ensure that the reliability and timeliness of the data in the system. By working together these organizations may be able to develop a more effective solution than the current process. Moreover, it is important that a more automated approach to the SOR be considered to reduce the errorprone manual processes that are currently in place. However, this is a longterm solution. Of more immediate concern are the offenders’ records that we did not look at that could have some of the critical and significant errors that we identified when we compared the SOR data to supporting documentation. Once these types of errors are addressed (e.g., incorrect end registration dates) then the State can have confidence that any new system or process that is adopted will not perpetuate the same errors as the old system/process. In addition, short-term solutions to address some of the specific process weaknesses that we found, such as a method to track and report treatment compliance status to VCIC could also enhance the reliability of the SOR. Recommendations We recommend that the Commissioner of the Department of Public Safety, Commissioner of the Department of Corrections, and Court Administrator form a working group to reassess and possibly redesign the processes related to the Vermont Sex Offender Registry to include possible system solutions to more effectively and efficiently submit information to the SOR. Page 38 We recommend that the Commissioner of the Department of Public Safety direct the Director of the Vermont Criminal Information Center to: ● Review all discharge dates, end registration dates, internet status, and risk assessment and lifetime registrant flags to confirm that the system accurately reflects supporting documentation and appropriately applies the statutory standard. This review should initially focus on those offenders whose records are posted on the Internet and may have their residential addresses added to the site. ● Work with the SOR system vendor to identify and correct the records of offenders that are shown on the Internet SOR as erroneously having been convicted of more counts than is factual. ● Perform a requirements analysis for the acquisition or development of a new SOR system (or redesign of the existing system) that includes, at a minimum, (1) improved electronic communication with DOC and the Courts, (2) a more robust set of edits, (3) an audit trail, (4) features in which the system automatically performs (or prompts the user to take) actions that are currently performed manually, such as the calculation of the end registration date, and (5) improved security features. ● Develop performance standards for the timely entry of data into the SOR and periodically assess whether these standards are being met. ● Modify the SOR procedures to include all SOR functions and documentation retention standards, including requirements to retain the results of the “sweeps” conducted by law enforcement when they physically check the residences of sex offenders. ● Develop a process to identify and track the treatment progress of offenders that are no longer under DOC supervision. ● Add the date the offender’s photograph was last updated to the Internet SOR records. ● Add the date last verified to the residential addresses posted to the Internet SOR. Page 39 We recommend that the Commissioner of the Department of Corrections: ● Explore, in conjunction with VCIC, system solutions to submit SOR forms electronically. ● Develop a mechanism to identify, and flag in its system, sex offenders in DOC custody who are registered, or required to register, with the SOR and prompts DOC personnel to submit required information to VCIC as necessary (e.g., the submission of a change of address form when a registered sex offender is sent to a DOC facility). ● Monitor the effectiveness of the department’s new SOR directive, particularly whether it results in more accurate and timely data submissions to VCIC and, if not, implement additional mechanisms to achieve this end, such as specialized training in areas of noncompliance. ● Develop a process to perform a risk assessment for women sex offenders that would meet the requirements of 13 VSA §5411b. Management Comments and Our Evaluation The Commissioner of the Department of Public Safety, Commissioner of the Department of Corrections, and Court Administrator provided written comments on a draft of this report, which are reprinted in appendices VI, VII, and VIII, respectively. The Commissioner of DPS’s June 22, 2010 response stated that overall the department concurs with the findings and recommendations in the report. In addition, the Commissioner outlined actions that the department is taking with respect to our recommendations. Notably, the department plans to form a working group along with the Commissioner of Corrections and the Court Administrator to reassess the need to redesign the workflow process associated with the SOR in an effort to develop a more accurate and efficient registry. Moreover, the Commissioner reported on other actions that DPS is taking. For example, the Commissioner noted that DPS formed an internal data quality improvement/audit team that is tasked with addressing many of our recommendations. For instance, the Commissioner reported that this team is expected to review the discharge date, end registration date, internet status, and risk assessment and lifetime registration flags for every individual on the registry. This is expected to take 8 weeks to complete. Page 40 Lastly, the Commissioner of DPS asserted that the efforts that are outlined in his response to this report would support a “favorable performance” that would prepare DPS to include addresses on the Internet SOR as directed by the legislature. While we cannot opine on the efficacy of actions that are not yet complete, the explanation of the planned measures contained in the Commissioner’s response to our draft report would seem to largely address the reliability problems that we found with the current SOR. Nevertheless, it is critical that these short-term efforts be coupled with the development of a more efficient and effective SOR process and system to provide added assurance that improved reliability would continue in the future. In his June 22, 2010 response to our draft report, the Commissioner of the Department of Corrections stated that he agreed with our conclusion that a more effective SOR is needed. The Commissioner stated that it is his intent to continue to work with the Commissioner of Public Safety and the Court Administrator to identify short and long-term solutions to more effectively and efficiently submit information to the registry. In the interim, the Commissioner reported that his department is taking other actions. For example, he reported that at his direction, DOC staff members are in the process of conducting a full-scale audit of the registry process to ensure that the department is in compliance with existing policy, with an eye towards the errors identified in our report. This process is expected to be completed within the next 90 days. The Commissioner added that periodic reviews will be required to monitor compliance and staff training needs. Other actions noted by the Commissioner are the development of a “tickler” system to notify staff to submit information to DPS and a risk assessment instrument for women sex offenders. The Court Administrator’s June 21, 2010 comments on a draft of our report stated that the office agrees with our findings, conclusions, and recommendations as it pertains to the Courts. He noted that a working group formed by the Commissioners of DPS and DOC, and the Court Administrator would likely succeed in developing a more accurate and efficient registry. In addition, the Court Administrator stated that in response to the findings in the report that the district courts have been instructed to provide VCIC with the Order to Provide Information, docket, and affidavit for all cases qualifying for the SOR. The Court Administrator added that regular reminders of these requirements will be made to ensure that these documents are consistently provided by the courts. Lastly, the Administrator added that these activities will be automated once the Court’s new case management system is operational, which is expected to be in about 2 years. Page 41 - - - - In accordance with 32 VSA §163, we are also providing copies of this report to the Secretary of the Agency of Administration, Commissioner of the Department of Finance and Management, and the Department of Libraries. In addition, the report will be made available at no charge on the State Auditor’s web site, http://auditor.vermont.gov/. Page 42 Appendix I Scope and Methodology In planning our work with respect to our first objective, we defined the terms reliable and current for purposes of the audit. We defined SOR reliability as follows: (1) all offenders who fulfill the statutory requirements for being on the registry are on it and those who do not are not, (2) each record has a full set of information, and (3) data in each record is accurate. The registry was considered to have current information if (1) information was received by VCIC from offenders within statutory timeliness requirements (e.g., address verification), (2) information was received by VCIC from other State organizations (e.g. Corrections, Judiciary) in a timely manner, and (3) data was entered into the registry a timely manner.47 In designing our approach to assess whether the SOR was reliable, we utilized guidance from the GAO.48 Specifically, GAO’s Data Reliability Guide outlines a variety of approaches to assessing the reliability of data, including performing data testing and tracing to and from source documentation. Data testing relates to applying logical tests to electronic data files, such as looking for duplicate records or values outside of a designated range. With respect to tracing to and from source documents, the GAO guide refers to tracing a random sample of data records to help determine whether the computer data accurately and completely reflect these documents. Consistent with GAO’s guidance, we (1) employed automated data analysis techniques on an electronic copy of the Registry as of February 18, 2010 and (2) extracted a representative sample of community-based offenders from this same electronic copy of the SOR and traced data in individual records to source documentation. Prior to performing these tasks we gained an understanding of the Registry’s data elements, the interrelationships among data elements, and valid values. We also scanned the data in the electronic copy to assess its reliability for purposes of our analyses (e.g., that dates were in the date fields and that data was not garbled). We concluded that the electronic SOR files that we received were sufficiently reliable for the purpose of conducting this audit. With respect to the automated data analysis that we performed on the electronic copy of the SOR, we downloaded the February 18, 2010 copy into an automated analysis tool. Using this tool, we performed a variety of tests, as follows: 47 We reviewed the information available regarding the timeliness of the data in the SOR and concluded that, except in the case of the annual verification process, we could not perform a systematic analysis of the currency of VCIC data. 48 Assessing the Reliability of Computer-Processed Data (U.S. Government Accountability Office, GAO-09-680G, July 2009). Page 43 Appendix I Scope and Methodology ● We tested for anomalies, such as incorrect or missing values in critical fields, dates that appeared to be erroneous, and the inconsistent or incorrect application of Registry rules. For example, we ran tests to determine whether (1) all offenders who had been designated high risk by DOC were on the Internet SOR, (2) any annual verification letters had not been received, and (3) the end registration date was calculated correctly. ● We chose judgmental samples of offenders in statuses that were not covered by our community-based and incarcerated offender samples to determine whether the offenders were in the correct categories. In particular, we scanned offenders in the pending and “reg exp” statuses looking for anomalies in the data that would indicate that the offender was in the wrong status. ● We compared the SOR data to data from the Court’s automated system to test whether the SOR contained all recent convictions that met statutory criteria. Specifically, the Court Administrator’s Office provided an electronic file of all sex offenders [as defined by 13 VSA §5401(10)] that had been sentenced between January 1, 2007 and December 31, 2009.49 Using our data analysis tool, we compared the Court’s file to the SOR file to determine whether there were offenders who had been convicted and sentenced for crimes under 13 VSA §5401(10) that were not in the Registry. ● We compared the SOR file provided to us as of February 18, 2010 to a copy of the Internet SOR that we downloaded the following day (the Internet SOR file is updated overnight so changes on February 18th would be shown on February 19th on the Internet). The purpose of this test was to confirm that the system logic used to extract sex offenders for the Internet actually resulted in those offenders, and only those offenders, being posted to the Internet SOR. For each data analysis test that yielded results (i.e., a potential error), we sought explanations from the VCIC SOR coordinator; DOC probation and parole and treatment personnel; and the Court’s information technology and court administration personnel, as applicable. When necessary, we also 49 Prior to performing this test we gained a general understanding of the Court system’s data elements, interrelationships among the data, and valid values and scanned the data in the electronic copy to assess its reliability for purposes of our analyses (e.g., that dates were in the date fields and that data was not garbled). We concluded that the file that we received from the Courts was reliable for the purpose of our test. We did not, however, review the internal controls related to this system. Page 44 Appendix I Scope and Methodology reviewed relevant documentation, such as court dockets, DOC records, and various SOR forms as part of researching and validating our results. Regarding our statistical sample of community-based offenders,50 we used the data from the SOR as of February 18, 2010 as the basis for pulling the sample. We used statistical random attribute sampling with a 95 percent confidence level, and 5 percent tolerable rate of records not in compliance with the statutorily established requirements and zero expected population deviation rate (i.e. expected error rate). Our universe consisted of the sex offenders with status codes of active,51 MIA,52 homeless,53 and student/employed.54 There were 1,549 SOR records that met these criteria on February 18, 2010 and our sample size was 57 records. For each record in the community-based sample, we traced data elements related to the offenders’ identification, conviction, sentencing, supervision, and location to supporting information from VCIC, DOC, or the Courts. Supporting information generally consisted of reviewing (1) the offender’s record in VCIC’s criminal history system, VCHIP; (2) the offender’s movement history and risk assessment history in the DOC system; (3) completed SOR forms, such as the registration form, change of address form, and treatment compliance/noncompliance forms; (4) court dockets; (5) affidavits filed in support of the charges, and (6) parole documentation. In addition, we evaluated VCIC’s decisions related to (1) whether the offender should be on the SOR and the Internet SOR, (2) the status of the offender, and (3) whether the offender was required to be a 10-year or lifetime registrant. We also performed a more limited review of incarcerated offenders. We limited our review of these offenders to checking whether a statistical sample of offenders listed in the SOR as being in inactive status were correctly in this status. To pick our sample, we used statistical random attribute sampling with a 95 percent confidence level, and 5 percent tolerable rate of records not in compliance with the statutorily established requirements and zero expected 50 These are offenders who live, work or go to school in Vermont. 51 All registered sex offenders (i.e., those for whom a signed Registration Form was submitted to VCIC) who reside in a Vermont community. 52 These are sex offenders who are currently in violation of Vermont SOR requirements. 53 These are registered sex offenders who have no actual physical address and must report daily to local police and/or the registry. 54 These are registered sex offenders that reside in another state or country but who work or are students in Vermont. Page 45 Appendix I Scope and Methodology population deviation rate (i.e. expected error rate). There was a universe of 423 SOR records that were in inactive status on February 18, 2010 and our sample size was 53 records. Our analysis of the sample of incarcerated offenders was limited to verifying that the DOC system recorded the offenders as being in a DOC correctional facility as of February 18, 2010. We followed up on those cases in which the offender was recorded as not being in a DOC facility by obtaining an explanation from the VCIC SOR coordinator or applicable probation and parole officer. We also confirmed that those offenders listed by DOC as being released from the DOC correctional facility to other jurisdictions were listed by those jurisdictions as being incarcerated. We performed this check by using the applicable jurisdictions’ Internet-based inmate locator search tools. As part of evaluating the results of our data analysis test and communitybased and incarcerated offender statistical samples, we defined an error as encompassing (1) records that were incorrectly omitted, added, retained, or deleted from the registry, (2) data in the registry that differed from the source documentation, (3) omission of data in a registry field, when applicable, and (4) inaccurate registry calculations or decisions (e.g., whether a record should be on the Internet registry). We also used the following categories to characterize the effect of the errors found in an offender’s record.55 ● Critical. Errors that have resulted, or would have resulted if not corrected, in a sex offender (1) being incorrectly omitted, added, retained, or deleted from the Registry or (2) being incorrectly omitted, added, retained, or deleted from the Internet Registry. ● Significant. Errors related to (1) sex offender identification (e.g., name and Social Security Number), (2) sex offender location information related to his or her residence, employment, or school, (3) other data that is on the Internet or provided to law enforcement agencies, or (4) incorrect 55 An offender’s Registry record could have more than one error and our process of categorization considered all of the errors taken together. In addition, errors of a similar nature may be categorized differently depending on the circumstances of an individual sex offender record. To illustrate, in the case of an offender whose registry record incorrectly showed him or her as compliant with treatment requirements; if a change from treatment compliance to noncompliance resulted in an offender being listed on the Internet registry, the error would be categorized as critical, but if the offender was already listed on the Internet registry for another reason, the error would be categorized as significant because the error would effect the accuracy of the data reported on the Internet. In another example, errors in discharge dates would be categorized differently depending on whether the offender was required to register for 10 years after discharge from supervision or for life. If the offender was in the 10-year category, we generally considered errors in discharge dates to be critical because the offender could be on the Registry for a longer or shorter period of time than required. However, if the offender was required to register for his or her lifetime, a discharge date error would be categorized as “other” because it would not effect how long the offender was on the registry. Page 46 Appendix I Scope and Methodology coding in the system that would have caused an offender’s address to be incorrectly added or excluded from the Internet Registry after July 1, 2010 if not corrected. ● Other. Information that is omitted or incorrect in the registry, but does not directly effect whether an offender is on the Registry or Internet Registry or data that is provided to the public or law enforcement agencies. In performing work in support of our second objective, we gained an understanding of the control processes that were used to ensure the reliability and timeliness of data in the SOR. To gain this understanding, we (1) reviewed the statute and laws related to the Registry; (2) reviewed the DPS Sex Offender Registry Rule effective December 31, 2004; (3) reviewed DOC’s sex offender registry directive effective September 9, 2008; (4) reviewed the DOC rule pertaining to risk assessments and treatment effective June 8, 2005; (5) performed walkthroughs of processes used to submit and maintain data in the SOR with the VCIC SOR coordinator, the VCHIP coordinator, a DOC caseworker and probation and parole officers, and District Court staff, and (6) interviewed other VCIC, DOC, and Court Administrator officials, including the VCIC Director and Deputy Director, DOC Field Services Executive and Co-Director of the Vermont Center for the Prevention and Treatment of Sexual Abuse, and the Office of the Court Administrator’s directors of the Division of Trial Court Operations and Division of Research and Information Systems. We also reviewed system documentation from the vendor that maintains the SOR and obtained clarifications of how it works from the vendor’s representative. Using all of the above sources, we flowcharted the various manual and automated processes used to initiate, maintain, and remove a sex offender’s record from the SOR. We validated this flowchart with appropriate officials from VCIC, DOC, and the Courts. Using the flowchart we identified control gaps and inconsistent applications of controls. We performed our work between October 2009 and May 2010 primarily at VCIC and DOC’s offices in Waterbury. We also performed site visits to DOC’s Probation and Parole Offices in Barre, Brattleboro, Burlington, Hartford, Morrisville, Newport, St. Albans, Rutland, and Springfield and to DOC’s Southern State Correctional Facility in Springfield. Except for the exception described below, we conducted this performance audit in accordance with generally accepted government auditing standards, which require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based Page 47 Appendix I Scope and Methodology on our audit objectives. The standard that we did not follow requires that our system of quality control for performance audits undergo a peer review every three years. Because of fiscal considerations, we have opted to postpone the peer review of our performance audits until 2011. Notwithstanding this exception, we believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. Page 48 Appendix II Statutory Definition of a Sex Offender 13 VSA §5401(10) defines a sex offender as the following: (A) A person who is convicted in any jurisdiction of the United States, including a state, territory, commonwealth, the District of Columbia, or military, federal, or tribal court of any of the following offenses: (i) sexual assault as defined in 13 VSA §3252. (ii) aggravated sexual assault as defined in 13 VSA §3253. (iii) lewd and lascivious conduct as defined in 13 VSA §2601. (iv) sexual abuse of a vulnerable adult as defined in 13 VSA §1379. (v) second or subsequent conviction for voyeurism as defined in 13 VSA §2605(b) or (c). (vi) kidnapping with intent to commit sexual assault as defined in 13 VSA §2405(a)(1)(D). (vii) aggravated sexual assault of a child in violation of section 3253a of this title; and (viii) a federal conviction in federal court for any of the following offenses: (I) Sex trafficking of children as defined in 18 U.S.C. §1591. (II) Aggravated sexual abuse as defined in 18 U.S.C. §2241. (III) Sexual abuse as defined in 18 U.S.C. §2242. (IV) Sexual abuse of a minor or ward as defined in 18 U.S.C. §2243. (V) Abusive sexual contact as defined in 18 U.S.C. §2244. (VI) Offenses resulting in death as defined in 18 U.S.C. §2245. (VII) Sexual exploitation of children as defined in 18 U.S.C. §2251. (VIII) Selling or buying of children as defined in 18 U.S.C. §2251A. (IX) Material involving the sexual exploitation of minors as defined in 18 U.S.C. §2252. (X) Material containing child pornography as defined in 18 U.S.C. §2252A. (XI) Production of sexually explicit depictions of a minor for import into the United States as defined in 18 U.S.C. §2260. (XII) Transportation of a minor for illegal sexual activity as defined in 18 U.S.C. §2421. (XIII) Coercion and enticement of a minor for illegal sexual activity as defined in 18 U.S.C. §2422. (XIV) Transportation of minors for illegal sexual activity, travel with the intent to engage in illicit sexual conduct with a minor, and engaging in illicit sexual conduct in foreign places as defined in 18 U.S.C. §2423. (XV) Transmitting information about a minor to further criminal sexual conduct as defined in 18 U.S.C. §2425. (ix) an attempt to commit any offense listed in this subdivision (A). (B) A person who is convicted of any of the following offenses against a victim who is a minor, except that, for purposes of this subdivision, conduct which is criminal only because of the age of the victim shall not be Page 49 Appendix II Statutory Definition of a Sex Offender considered an offense for purposes of the registry if the perpetrator is under the age of 18 and the victim is at least 12 years old: (i) any offense listed in subdivision (A) of this subdivision (10). (ii) kidnapping as defined in 13 VSA §2405(a)(1)(D). (iii) lewd and lascivious conduct with a child as defined in 13 VSA §2602. (iv) slave traffic as defined in 13 VSA §2635. (v) sexual exploitation of children as defined in 13 VSA chapter 64. (vi) procurement or solicitation as defined in 13 VSA §2632(a)(6). (vii) aggravated sexual assault of a child as defined in 13 VSA §3253a. (viii) sex trafficking of children or sex trafficking by force, fraud, or coercion as defined in 13 VSA §2635a. (ix) sexual exploitation of a minor as defined in 13 VSA §3258(b). (x) an attempt to commit any offense listed in this subdivision (B). (C) A person who takes up residence within this state, other than within a correctional facility, and who has been convicted in any jurisdiction of the United States, including a state, territory, commonwealth, the District of Columbia, or military, federal, or tribal court, for a sex crime the elements of which would constitute a crime under subdivision (A) or (B) of this subdivision (10) if committed in this state. (D) A person 18 years of age or older who resides in this state, other than in a correctional facility, and who is currently or, prior to taking up residence within this state, was required to register as a sex offender in any jurisdiction of the United States, including a state, territory, commonwealth, the District of Columbia, or military, federal, or tribal court; except that, for purposes of this subdivision, conduct which is criminal only because of the age of the victim shall not be considered an offense for purposes of the registry if the perpetrator is under the age of 18 and the victim is at least 12 years old. (E) A nonresident sex offender who crosses into Vermont and who is employed, carries on a vocation, or is a student. Page 50 Appendix III Example of the Information Posted on the Internet SOR Page 51 Appendix IV Profile of Sample of Community-Based Offenders Because of concerns over confidentiality, we are not providing the names or characteristics of individual offenders in our community-based sample. However, the following provides summary-level information on the offenders in this sample: Sex 56 males 1 female Internet Status 33 on Internet 24 not on Internet DOC Supervisory Status 22 under supervision 35 not under supervision Place of Conviction 50 were convicted in Vermont 6 were convicted in another jurisdiction 1 was convicted both in Vermont and another jurisdiction Vermont Statute Convicted Under56 12 were convicted of lewd and lascivious conduct 14 were convicted of lewd and lascivious conduct with a child 23 were convicted of sexual assault 1 was convicted of aggravated sexual assault 3 were convicted of sexual exploitation of a child 5 were convicted of other Vermont crimes County of Residence as of February 18, 2010 Addison—1 offender Bennington—2 offenders Caledonia—2 offenders Chittenden—6 offenders Essex—1 offender Franklin—7 offenders 56 Some offenders had convictions under multiple Vermont statutes. Also, this category does not include out-of-state convictions. Page 52 Appendix IV Profile of Sample of Community-Based Offenders Grand Isle—1 offender Lamoille—4 offenders Orange—4 offenders Orleans—1 offender Rutland—8 offenders Washington—6 offenders Windham—8 offenders Windsor—3 offenders Unknown—2 offenders Not in Vermont—1 offender Page 53 Appendix V Summary of Errors in the Community-Based Statistical Sample The following table summarizes the errors found in our statistical sample of 57 community-based sex offenders. Table 3: Number and Types of Errors Found in Sample of 57 Community-based Offenders Type of Errora Data Element Name Name alias Date of birth Date of birth alias Sex Race Eye color Height Social Security Number Social Security Number alias Photo existence Docket number Offense literal description Conviction date Sentence Age of victim Discharge date Registration start date Registration end date Lifetime Registrant flag Annual verification letter receipt date Residential address County of residence Employer name Employer address School name School address High risk designation flag Treatment compliance flag Noncompliant high-risk offender flag 30 b 0 0 0 0 0 0 0 0 0 0 20c 20c 20c 20c 0 0 0 0 0 0 2 2 0 0 0 1 0 0 4 0 0 5 11 10 11 9 7 13 15 1 3 0 3 0 4 0 0 1 1 0 2 0 1 0 0 0 12 9 1 1 0 1 32 5 0 4 0 1 1 1 4 2 0 24 26 26 27 21 16 14 16 1 4 Error percentage (based on 57 records in sample) 56% 9% -7% -2% 2% 2% 7% 4% -42% 46% 46% 47% 37% 28% 25% 28% 2% 7% 0 0 1 2 0 0 0 0 0 3 3 4 4 0 0 4 3 0 0 0 0 1 0 0 16 2 0 3 3 5 7 0 0 20 5 0 5% 5% 9% 12% --35% 9% -- Number Incomplete Number Inaccurate Page 54 Number Omitted Number of Records with Errora Appendix V Summary of Errors in the Community-Based Statistical Sample Type of Errora Data Element DOC Probation and Parole Office Probation and Parole Officer SOR status Notification Internet category Number Incomplete Number Inaccurate Error percentage (based on 57 records in sample) 2 4% Number of Records with Errora Number Omitted 0 2 0 0 0 7 7 4 5 4 4 1 0 2 2 5 5 13 13 9% 9% 23% 23% a The types of errors will not always add to the total number of records with an error because some offenders had multiple types of errors related to a particular data element. For example, in some cases the offender’s record may not include a conviction (incomplete) and the conviction data included in the SOR may not be correct (inaccurate). b The incomplete error type for name was due to missing name suffixes (e.g., Jr.) or middle names when they were available. c The incomplete error type for these data elements generally related to violations of probation. Page 55 Appendix VI Response of the Commissioner of the Department of Public Safety Page 56 Appendix VI Response of the Commissioner of the Department of Public Safety Page 57 Appendix VI Response of the Commissioner of the Department of Public Safety Page 58 Appendix VII Response of the Commissioner of the Department of Corrections Page 59 Appendix VII Response of the Commissioner of the Department of Corrections Page 60 Appendix VIII Response of the Court Administrator Page 61