Part B_final

B. Collections of Information Employing Statistical Methods

1. Universe and Respondents Selection

The Mortality in Correctional Institutions (MCI) is a full enumeration of deaths occurring in the custody of prisons and local jails, rather than a sample of deaths. Since death is a rare event, sampling is challenging in this context. The original legislation upon which MCI was initiated required a full enumeration of deaths in custody, and BJS has been collecting it that way since then. Sampling would unnecessarily complicate the analysis of trends in causes of death and subpopulation analyses over time, and greatly diminish the quality and utility of the data.

The potential respondent universe for the prison portion of the MCI consists of the 50 state departments of corrections (DOCs). BJS collects administrative data directly from each of the 50 state DOCs. Respondents are DOC staff from the state that has custody of the decedents.

The universe for MCI-Local Jails includes approximately 3,000 local jail reporting units nationwide. Local jails are accustomed to the collection and typically, can rely on their existing information systems to easily provide the data requested. Respondents are local jail staff, including sheriffs, correctional officers, or administrative personnel.

2. Procedures for Information Collection

In November of the year prior to the upcoming calendar year collection, the MCI data collection agent conducts verification calls on behalf of BJS. Through the verification process, the data collection agent confirms that agency contact information is correct, and ensures that jail reporting units meet the criteria for inclusion in the MCI. Changes in prison respondents are uncommon, but there are numerous updates to the jail universe annually. To further reduce burden on respondents, the updated list of eligible jail reporting units is used by MCI as well as other BJS jail collections, including the Annual Survey of Jails (ASJ) and the Census of Jails (COJ).

Using the contact information obtained from the verification calls, the MCI data collection agent sends each potential respondent, either by postal mail or email, an information packet announcing the start of the current year’s collection cycle (Appendix X), typically in mid-January. The mailings (or emails) include reporting instructions, a brief sheet consisting of a short quiz for fun (no responses are collected), MCI facts, and login credentials for submitting data online through the MCI website. Respondents are reminded to submit an annual summary form, even if no deaths occurred in their agencies’ custody during the year, along with any outstanding death report forms from the preceding year. BJS requests that agencies submit their forms by February 28 of the year following the death. Most agencies submit their reports electronically through the MCI website, but some DOCs from larger states, such as California and Texas, submit encrypted bulk files.

Data quality follow-up (DQFU) and reminders to submit data can begin as early as February. Based on several years’ experience, the data collection agent has demonstrated that conducting data quality follow-up on a rolling basis, instead of waiting until a prescribed number of records have been received by a fixed date, yields the best response rates and most complete data. Reminders and data quality follow-up occur by phone, email, and postal mail. Project teams employed by the data collection agent manage the outreach activities and assign agency liaisons to handle correspondence with particular agencies. Agency liaisons (data collection staff) communicate with the largest agencies and those with special requirements. The majority of DQFU calls are made by call center staff, who are usually familiar with the MCI collection and are trained by the data collection agent.

The data collection agent monitors the MCI data and its respondents throughout the active data collection period, which includes tracking the time to complete, number and nature of contacts made to each agency, and reports to BJS on the status and number of forms received from each agency to-date. Typically a data collection cycle ends in September, culminating in the delivery of data files to BJS by late September or early October.

3. Methods to Maximize Response

BJS and data collection agent staff market the mortality data collections to generate interest among stakeholders and encourage participation. Typically BJS receives feedback through email from jail administrators and DOC personnel, at healthcare-related or correctional conferences, and comments from respondents during data collection. BJS staff also present at conferences about the needs and uses of mortality data. Discussions with various stakeholders and respondents reveal general satisfaction with the current questionnaires, including topic relevance and clarity of instructions.

BJS has consistently achieved high response rates for the state prison and local jail inmate mortality data collections. All 50 state DOCs have participated each year since the prison mortality data collection’s inception in 2001. Since BJS initiated the MCI-Local Jails collection in 2000, it had an average annual response rate of 98% from approximately 3,000 local jail jurisdictions. In 2013, the response rate decreased slightly to 92.6% due to a single-year increase in questions about the jail facilities, but rebounded to 95.7% in 2014 (table 3). Given the historically high response rates for most survey items, BJS typically does not report missing data in its statistical tables published annually to report on deaths in custody of prisons and jails. The only exception to this practice is in reporting total number of deaths, where cases missing cause of death are included to ensure an accurate total number of deaths. BJS does not impute missing values of individual death data elements.

Table 4. Response rates for eligible jail jurisdictions in the MCI frame, 2013-2017

	2013	2014	2015	2016	20171
Number of eligible jail jurisdictions in frame	2,949	2,947	2,930	2,913	2,993
Number of jail jurisdictions responding	2,732	2,821	2,850	2,846	2,920
Percent of jail jurisdictions responding	92.6%	95.7%	97.3%	97.7%	97.6%

¹2017 response rates are as of October 8, 2018. The final data will be delivered by the end of the month. Note: numbers represent the number of eligible jail jurisdictions and number reporting as of the end of each collection year. Data may have been revised from previously published statistics.

Data collection staff follow up with emails and phone calls to both promote high response rates and to verify any out-of-range values for critical items and improve data quality. These methods have been effective in attaining high unit response, minimizing item nonresponse, and ensuring data validity.

4. Test of Procedures or Methods

BJS obtains feedback from state DOCs and local jail jurisdictions before proposing significant changes to existing questionnaires. As discussed in Part A, in the next year, BJS will pursue obtaining more detailed cause of death information from the National Death Index (NDI), specifically underlying causes of death, which are not captured on the current questionnaires.

BJS and its data collection agent continuously seek opportunities to innovate and adjust protocols that may increase response rates or reduce respondent burden or costs. For instance, the annual verification calls that occur two months prior to data collection serve to confirm agency eligibility and contact information, and to pre-notify agencies of the upcoming data collection. The data collection agent for MCI has successfully implemented experiments, testing and identifying participants’ responsiveness to reminders sent by email vs. a physical letter that includes hardcopy form(s). Results from these experiments are used to tailor correspondence to agencies based on their historical or expressed preference for email or physical mail. Offering multi-mode data collection (online, mail, email, fax, phone, or bulk file transfer) also contributes to better response rates.

Online submissions benefit from automated validation checks. If data entered appear to be out of bounds or inconsistent with other responses, an alert will prompt the respondent, thus maximizing data quality. During each data collection cycle, the data collection agent accounts for factors that may affect form completions and impact the data collection timeline, such as competing agency priorities, deaths pending autopsy or further investigation, and recent election of officials, which may result in changes to POCs and their willingness to participate.

As BJA develops its new mortality information collection Death in Custody Reporting Act Collection (OMB Number 1140–0101), in compliance with legislation passed in 2014, BJS will assess items that should be removed from the BJS MCI collection to avoid redundancy and minimize burden. Once the BJA mortality collection is firmly established, BJS will submit proposed revisions to the MCI collection to OMB for approval. Presently, however, items previously approved for collection will not change.

5. Contact Information

The BJS Corrections Unit has overall responsibility for the design and management of activities related to the MCI collection. BJS oversees data collection, development of definitions and criteria for inclusion in the study, data analysis and dissemination. The primary BJS contact for the MCI is:

Elizabeth Ann Carson, Ph.D.

Statistician and Acting Chief, Corrections Unit

Bureau of Justice Statistics

810 Seventh Street NW

Washington, DC 20531

(202) 616-3496

Elizabeth.Carson@usdoj.gov

3