B. Statistical Methods
Respondent Universe
The survey will be provided to the central criminal history record repositories in 56 jurisdictions including the 50 States, the District of Columbia, American Samoa, Guam, the N. Mariana Islands, Puerto Rico and the U.S. Virgin Islands. These entities represent the entire population of central state criminal history repositories in the U.S. Thus, the survey typically does not involve any sampling, methods for stratification, imputation-s or similar estimation methods. Every effort will be made to obtain a completed survey from every repository. If a repository does not answer every question on the survey or does not submit responses to the survey, the entry will remain blank in the resulting report unless it is possible to provide an estimated response. For example, Alabama was unable to provide information regarding the number of subjects in their criminal history file for the 2018 survey, but they had reported this information during the three previous cycles. To estimate the number of records for 2020, SEARCH (in consultation with BJS) determined the average biennial growth rate in the criminal history file to calculate the estimated number of subjects added to the file from 2016 to 2018. While it is not anticipated that there will be a need to estimate data for the 2020 cycle, similar estimation methods may be used when preparing the 2020 reporting tables with the consent of the state for which data are estimated. Footnotes will be added to each table indicating “not reported” or “estimated” when applicable. The most recent survey (2018) achieved a 93% response rate with some follow up effort (all 50 states, Guam and Puerto Rico responded, but no responses were received from the District of Columbia, American Samoa, N. Mariana Islands, and the Virgin Islands). A similar response rate is anticipated for the 2020 collection. The primary goal is to receive completed responses from the 50 states given the majority of federal funding has been directed to and volume of criminal records are provided by the states.
Procedures for Collecting the Information
Respondents will have access to online, password-protected reporting forms (three screenshots from the 2018 web-based survey are included as Attachment 4). For the 2018 collection, 47 (90.4%) of the 52 respondents submitted their information via the web-based form (4 submitted via email and 1 via fax). A similar distribution is anticipated for the 2020 collection. The web-based forms will automatically populate databases that will be accessible by respondents online, thereby reducing data entry error. Online edit functionality is also employed by the online reporting tool to detect common errors in data reporting and anomalies against data reported in previous cycles. The functionality will include a built-in utility for respondents to complete and submit individual sections of the survey, and to examine/update previously submitted portions. It will also allow other maintenance and reporting capabilities such as the ability to store survey data in a researchable database to produce tables, dashboards, and graphics. Since it is possible that some respondents will prefer to complete a paper version of the survey, this capability is also being provided.
For most of the numerical data collected, the data collection agent will primarily calculate totals, percentages, and cumulative percentages. In some cases, frequencies will be compiled and some distributional statistics will be calculated including means, medians, modes, and ranges. As stated above, there will be no adjustments made for missing data. Missing data will be properly flagged and any analysis of provided data will footnote any discrepancies in the analysis.
Methods to Maximize Response Rates
Response rates will be maximized through email reminders after the initial contact. Approximately a week before the original deadline for submission, the data collection agent will follow up with respondents who have not submitted a completed survey. If some respondents have not responded by the deadline, a second reminder is emailed a few days after the deadline. Where necessary, respondents will be contacted via telephone. This follow-up will be performed primarily by the data collection agent. The survey is the only existing source for complete and comprehensive data on the status of state-maintained criminal history records. As such, users at both the state and Federal levels have come to depend on its production on a regular basis to assess the effectiveness of programs designed to improve criminal record quality and to identify priorities for federal funding. It is the most cited publication produced by SEARCH and the survey is very popular among respondent agencies. As a result of the follow-up efforts, and due to the survey’s popularity and importance, non-response has not been a significant issue. In a few exceptional cases, follow-up may be conducted by BJS if the data collection agent has not been successful in achieving a response.
Respondents will be asked to verify submitted responses, and will also be given the opportunity to review individual and compiled responses prior to publication.
Tests of Procedures
The changes made to the survey described in Part A, item 3 (“Efforts to Minimize Burden”) were developed and designed by SEARCH staff, in consultation with BJS. SEARCH staff, along with personal experience with prior iterations of the instrument serving in previous roles as state record repository staff, have extensive experience with state criminal record repository operations. The alterations to the instrument and its implementation derive from ongoing discussions and reviews with state criminal record repository staff. Comments and suggestions on the survey are routinely gathered by SEARCH staff through presentations based on survey results to the SEARCH membership and at meetings of the FBI Criminal Justice Information Services Division Advisory Policy Board, its task forces, subcommittees, and regional working groups. The most recent version of the survey has been reviewed by the six states mentioned in Part A, but this version has not been fully tested. Given that this is not a new survey to these respondents a full testing of the instrument did not seem necessary. The estimates of the burden provided by respondents were based on a thorough review of the instrument and completion of past surveys.
Contact information
BJS contact for statistical aspects of the design – Devon Adams (202) 514-9157; [email protected])
Data Collection Agent contact - Dennis DeBacco, SEARCH, 1900 Point West Way, Suite 161, Sacramento, CA, 95815, (916) 392-2550, ext. 325; [email protected].
| File Type | application/vnd.openxmlformats-officedocument.wordprocessingml.document |
| Author | Adams, Devon |
| File Modified | 0000-00-00 |
| File Created | 2021-02-04 |