This appendix describes the approach employed to extract the universe of victims to whom notifications were generated by the VNS during the scope of our review. From this universe, a sample of registrant identification numbers was selected and a questionnaire was sent to the sampled victims. Accordingly, the descriptions of the sample selection process and the survey response review are included in this appendix. Our objective was to determine whether the VNS served its intended purpose and satisfied victims’ expectations.

Scope

The scope of our universe consisted of opt-in victims with approved notification information between October 1, 2004, and September 30, 2006. “Opt-in” is the status of a registered victim or contact allowing them to receive notifications and access the VNS phone line and Internet web page.

Methodology

To determine the number of victims registered in the VNS, we requested specific data fields contained in the VNS database between October 1, 2004, and September 30, 2006. We obtained the data to select a sample of victims for our survey deployment.

The VNS creates a unique registrant identification number (Registrant ID) to identify a victim. The Registrant ID data field was used as the anchor data field in the universe extraction process. We requested data for each notification phase (investigative, prosecution, and incarceration) using the Registrant ID as the primary key for opt-in victims. We requested case information, such as the investigative case number, notification event, USAO agency name, USMS inmate number, custody status; and victim information, such as city, state, and country.

Data Organization and Sample Selection

We identified records associated to the opt-in victims and requested the databases from EOUSA. We constructed the opt-in victim universe from the received data download. We uploaded the databases to summarize the data in order to compile the information and link tables to create a universe from which we could select the sample. There were a total of 621,276 registrant IDs we identified that were at least in 1 of the 3 phases during the scope of our audit. We then removed 4,000 registrant IDs that had previously been surveyed by EOUSA to arrive at our final universe of 618,201 unique sample units. We also created a sub-set of 52,166 registrant IDs from the universe using the incarceration phase’s “Notification Event Types” field with events that were related to the release of inmates.

We separated the universe into two categories for U.S. and non-U.S. Registrant IDs using the victim information fields, state, and country. We then stratified the universe of unique Registrant IDs and the incarceration phase release-related events sub-set. We then created nine strata through various grouping of the three phases and the incarceration release-related events using the U.S. Registrant IDs to group them. The tenth stratum included the Registrant IDs with non-U.S. addresses and those with no address information identified. The sample selection process included the sample sizes for each of the 10 strata that were statistically calculated. The total population size was 618,203 Registrant IDs, and the sample size was 2,783.

Survey Response Review

We received a total of 691 surveys mailed back to us by the recipients. We found that some respondents completed the survey, answered only some of the questions, or included a comment without completing the survey. As a result, we reviewed the data to eliminate surveys that did not contain useful information. Upon closer review of the data, we found records of surveys that did not have responses to all the questions in the survey. We identified and excluded those records with non-responses and arrived at a data set of 531 records. We used the data in these 531 records for analysis using the SPSS software package.

Conclusion

Of the total 618,201 unique VNS Registrant ID numbers opted into the system from October 1, 2004, through September 30, 2006, which were suitable for our survey deployment, our sample size was 2,783. We received a total of 691 surveys, of which 531 were considered to have useful responses.