NPRP Draft Part B - Supporting Statement 20220419

SUPPORTING STATEMENT (PART B)

National Pretrial Reporting Program (NPRP)

B. COLLECTION OF INFORMATION EMPLOYING STATISTICAL METHODS

1. Universe and Respondent Selection

The purpose of the NPRP is to understand the pretrial release or detention ordered for defendants with at least one felony charge filed in state courts in the largest 200 counties in the United States. Describing pretrial release and detention may require information from courts (for the initial bond decision, any changes in release or detention status during the case, disposition, and sentencing), jails (for any period the defendant is incarcerated or re-incarcerated during the pendency of the case), and pretrial services agencies (for any supervision during periods of pretrial release).

The target population for the NPRP is all criminal cases filed with at least one felony charge in state courts in the largest 200 counties in calendar year 2019.¹ We will ask the courts to provide case-level data for all cases filed with at least one felony charge in calendar year 2019 through disposition (and, if possible, sentencing). We will ask jails to provide information for all bookings, and all cases opened by pretrial release agencies for calendar year 2019. We will match case-level data across the jail, court, and pretrial services agency files, knowing that some records may not be able to be matched (e.g., cases filed in January 2019 may have been booked in December 2018, and cases filed in December 2019 may not be released to a pretrial services agency until 2020).

The NPRP is based on an earlier data collection series of the same name, later amended to be called the State Court Processing Statistics (SCPS) series. Beginning in 1988, that program used a sample of 40 of the largest 75 counties based on county population size, with a certainty stratum based on the relative number of case filings in previous SCPS collections. The SCPS data collection paused in 2009 to examine different sampling strategies and data collection methods; as a result, BJS does not have an estimate of the number of felony criminal cases filed in the largest counties. However, population size is highly correlated with case filings, and with this NPRP, BJS is selecting the largest 75 counties with certainty and drawing a sample of 50 counties from the next largest 125 counties, to be representative of the largest 200 counties.

NPRP will include felony case filings in state courts of general jurisdiction. Although some felonies may resolve in limited jurisdiction courts, most will be transferred to the court of general jurisdiction for resolution. BJS is not targeting limited jurisdiction courts in the NPRP effort; however, if statewide or centralized data providers are able to provide data from limited jurisdiction courts for cases, then BJS will accept it. BJS is excluding municipal courts since their involvement in felony case processing is highly unlikely.

Overall Study and Sample Design

The overall NPRP sample design will combine a census of the 75 largest counties with a sample-based data collection from 50 of the next largest 125 counties to estimate the pretrial characteristics for the largest 200 counties. Counties 76-200 will be stratified by population size, and the sample of 50 will be drawn proportionate to size. If counties in the sample decline to provide data, BJS will draw a replacement county from the same stratum.

BJS will minimize the number of states, counties, and agencies asked to report data to NPRP. Table 1 shows the sources that BJS anticipates engaging in the NPRP collection. NPRP is county-based, and some counties have centralized reporting repositories for court, jail, and pretrial services data, where most or all agencies report their data to a single source, such as the county court. In those counties, BJS will request data from that single source. For some counties, all counties in a state report data to a central repository, such as a state administrative office of the courts. In these instances, BJS will request the data for the specific counties from the centralized repository and combine the state-reported county data with the data requested from the remaining agencies within the counties.

Regardless of the data sources, BJS will merge the data at the case level to follow an individual’s path from case filing to pretrial release or detention, and to case outcome and sentencing. BJS will request that courts, jails, and pretrial services agencies provide common identifiers between the three agencies, such as a unique defendant identifier or a series of case-linking identifiers. If no common identifier exists, BJS will match individuals based on demographic factors, such as name, date of birth, race, and sex.

The data collection is a census of the 75 largest counties and a sample of 50 of the next largest 125 counties, based on the size of the population aged 18 and over in 2019. The rationale for using 2019 is to avoid most of the disruption of “typical” pretrial release and detention practice that occurred because of the COVID-19 pandemic.

Table 2 details the largest 75 counties. Table 3 lists the remaining counties, 76-200, from which the sample will be drawn.

Table 2. Largest 75 counties by population, 2019

County Size Rank	County	State	2019 Population 18+
1	Los Angeles County	California	7,894,558
2	Cook County	Illinois	4,037,516
3	Harris County	Texas	3,467,885
4	Maricopa County	Arizona	3,432,975
5	San Diego County	California	2,623,532
6	Orange County	California	2,486,016
7	Miami-Dade County	Florida	2,167,261
8	Dallas County	Texas	1,955,358
9	Kings County	New York	1,979,773
10	Riverside County	California	1,856,391
11	Queens County	New York	1,802,531
12	King County	Washington	1,801,166
13	Clark County	Nevada	1,745,918
14	San Bernardino County	California	1,610,447
15	Tarrant County	Texas	1,555,282
16	Bexar County	Texas	1,497,113
17	Santa Clara County	California	1,511,935
18	Broward County	Florida	1,542,840
19	Wayne County	Michigan	1,336,953
20	Alameda County	California	1,331,231
21	New York County	New York	1,396,835
22	Middlesex County	Massachusetts	1,296,600
23	Philadelphia County	Pennsylvania	1,241,810
24	Sacramento County	California	1,188,937
25	Suffolk County	New York	1,167,701
26	Palm Beach County	Florida	1,212,898
27	Bronx County	New York	1,070,144
28	Hillsborough County	Florida	1,146,545
29	Nassau County	New York	1,065,969
30	Orange County	Florida	1,087,438
31	Franklin County	Ohio	1,011,351
32	Oakland County	Michigan	997,704
33	Cuyahoga County	Ohio	980,916
34	Hennepin County	Minnesota	989,821
35	Travis County	Texas	1,004,012
36	Allegheny County	Pennsylvania	989,647
37	Fairfax County	Virginia	880,601
38	Contra Costa County	California	894,142
39	Salt Lake County	Utah	851,291
40	Mecklenburg County	North Carolina	852,208
41	Wake County	North Carolina	849,055
42	Montgomery County	Maryland	808,651
43	Fulton County	Georgia	836,143
44	Pima County	Arizona	831,673
45	St. Louis County	Missouri	776,516
46	Honolulu County	Hawaii	769,689
47	Fresno County	California	717,718
48	Collin County	Texas	769,439
49	Westchester County	New York	757,148
50	Pinellas County	Florida	819,558
51	Marion County	Indiana	727,973
52	Milwaukee County	Wisconsin	720,305
53	Fairfield County	Connecticut	733,670
54	Shelby County	Tennessee	704,794
55	Duval County	Florida	742,210
56	Bergen County	New Jersey	735,892
57	DuPage County	Illinois	715,343
58	Erie County	New York	733,429
59	Gwinnett County	Georgia	686,917
60	Prince George's County	Maryland	707,865
61	Hartford County	Connecticut	705,385
62	Kern County	California	641,082
63	Pierce County	Washington	694,525
64	San Francisco County	California	763,303
65	Macomb County	Michigan	692,117
66	New Haven County	Connecticut	684,132
67	Hidalgo County	Texas	590,120
68	Ventura County	California	655,715
69	El Paso County	Texas	614,939
70	Denton County	Texas	671,750
71	Baltimore County	Maryland	648,363
72	Middlesex County	New Jersey	646,614
73	Worcester County	Massachusetts	657,270
74	Montgomery County	Pennsylvania	652,573
75	Hamilton County	Ohio	630,440

Table 3. Largest 76-200 counties by population, 2019

County size rank	County	State	2019 Population 18+
76	Multnomah County	Oregon	663,188
77	Snohomish County	Washington	637,832
78	Suffolk County	Massachusetts	672,740
79	Essex County	New Jersey	609,597
80	Oklahoma County	Oklahoma	594,839
81	Essex County	Massachusetts	622,724
82	San Mateo County	California	611,781
83	Jefferson County	Kentucky	598,203
84	Fort Bend County	Texas	589,946
85	Cobb County	Georgia	583,597
86	DeKalb County	Georgia	585,187
87	Monroe County	New York	588,820
88	San Joaquin County	California	558,389
89	Lee County	Florida	636,679
90	Denver County	Colorado	588,587
91	Lake County	Illinois	530,410
92	Norfolk County	Massachusetts	559,627
93	El Paso County	Colorado	549,134
94	Jackson County	Missouri	538,783
95	District of Columbia	District of Columbia	577,848
96	Will County	Illinois	521,914
97	Davidson County	Tennessee	551,090
98	Polk County	Florida	565,638
99	Bernalillo County	New Mexico	534,056
100	Hudson County	New Jersey	535,864
101	Jefferson County	Alabama	509,191
102	Kent County	Michigan	499,889
103	Tulsa County	Oklahoma	487,873
104	Arapahoe County	Colorado	504,162
105	Providence County	Rhode Island	507,922
106	Bucks County	Pennsylvania	501,425
107	Monmouth County	New Jersey	489,192
108	Baltimore city	Maryland	473,923
109	Utah County	Utah	426,950
110	Ocean County	New Jersey	460,496
111	Johnson County	Kansas	457,474
112	Washington County	Oregon	466,438
113	Brevard County	Florida	492,569
114	Jefferson County	Colorado	469,684
115	Montgomery County	Texas	448,951
116	Anne Arundel County	Maryland	450,650
117	Delaware County	Pennsylvania	442,201
118	Bristol County	Massachusetts	449,495
119	Douglas County	Nebraska	425,639
120	New Castle County	Delaware	439,396
121	Union County	New Jersey	426,292
122	Williamson County	Texas	440,981
123	Ramsey County	Minnesota	422,367
124	Stanislaus County	California	402,887
125	Summit County	Ohio	428,863
126	Lancaster County	Pennsylvania	417,852
127	Volusia County	Florida	456,552
128	Dane County	Wisconsin	436,428
129	Montgomery County	Ohio	415,349
130	Kane County	Illinois	399,424
131	Guilford County	North Carolina	418,280
132	Pasco County	Florida	441,991
133	Chester County	Pennsylvania	407,023
134	Plymouth County	Massachusetts	410,783
135	Sedgwick County	Kansas	384,757
136	Greenville County	South Carolina	403,474
137	Camden County	New Jersey	392,466
138	Spokane County	Washington	407,948
139	Adams County	Colorado	382,294
140	Passaic County	New Jersey	382,808
141	Sonoma County	California	398,859
142	Morris County	New Jersey	389,366
143	Lake County	Indiana	373,045
144	Polk County	Iowa	369,064
145	Richmond County	New York	372,457
146	Clark County	Washington	373,556
147	Hampden County	Massachusetts	366,727
148	Onondaga County	New York	363,435
149	Tulare County	California	323,943
150	Prince William County	Virginia	344,025
151	Seminole County	Florida	372,855
152	Knox County	Tennessee	371,876
153	Washoe County	Nevada	370,990
154	Ada County	Idaho	369,859
155	Virginia Beach city	Virginia	350,926
156	Burlington County	New Jersey	353,190
157	York County	Pennsylvania	350,419
158	Santa Barbara County	California	348,215
159	East Baton Rouge Parish	Louisiana	339,986
160	Solano County	California	348,758
161	Jefferson Parish	Louisiana	337,196
162	Monterey County	California	320,870
163	Pinal County	Arizona	360,216
164	Lucas County	Ohio	330,356
165	Cameron County	Texas	296,542
166	Dakota County	Minnesota	325,107
167	Sarasota County	Florida	372,984
168	Berks County	Pennsylvania	327,545
169	Mobile County	Alabama	316,868
170	Hillsborough County	New Hampshire	332,756
171	Richland County	South Carolina	326,666
172	Clackamas County	Oregon	329,826
173	Genesee County	Michigan	315,245
174	Charleston County	South Carolina	330,609
175	Waukesha County	Wisconsin	318,146
176	Loudoun County	Virginia	298,272
177	St. Charles County	Missouri	309,611
178	Pulaski County	Arkansas	301,662
179	Orleans Parish	Louisiana	313,010
180	Placer County	California	310,171
181	Manatee County	Florida	330,933
182	Orange County	New York	287,134
183	Butler County	Ohio	293,990
184	Forsyth County	North Carolina	295,459
185	Lane County	Oregon	312,496
186	Allen County	Indiana	282,488
187	Stark County	Ohio	291,678
188	Collier County	Florida	319,864
189	Mercer County	New Jersey	289,368
190	Washtenaw County	Michigan	300,102
191	Lehigh County	Pennsylvania	286,118
192	Madison County	Alabama	292,193
193	Nueces County	Texas	274,352
194	Hamilton County	Tennessee	291,381
195	Brazoria County	Texas	276,764
196	Marion County	Florida	298,327
197	Westmoreland County	Pennsylvania	285,145
198	Osceola County	Florida	285,152
199	Anoka County	Minnesota	272,162
200	Bell County	Texas	263,178
Source: U.S. Census Bureau, Population Division. Table 1. Annual Estimates of the Resident Population for the United States, States, Counties and Puerto Rico Commonwealth and Municipios: April 1, 2010 to July 1, 2019

For the purposes of the overall design, BJS has assumed the following:

Class 1 – Collection from the largest 75 counties

During work under BJS’s generic clearance (OMB Control No.1121-0339), BJS contacted court data leaders, jails, and pretrial service agencies in the largest 75 counties to determine whether their electronic case-level records systems are capable of extracting data elements necessary to support the NPRP. Overall, the data systems vary in terms of geographic coverage (e.g., statewide data system, centralized county with all jail, court, and pretrial records, and county agency-specific data systems). The data systems are used largely for case management and include data elements related to general case information, defendants/inmates/clients, charges, filing/booking/intake, and disposition/release/termination of supervision. Sentencing data are sometimes maintained by the court or jail data systems, and sometimes by both. Some of these data are in free text fields or contained in scanned or paper documents, such as orders of release or orders of supervision.

Any agency or centralized data repository (e.g., centralized data for all agencies within the county or state) will be asked for an electronic file containing all criminal cases filed as felonies in calendar year 2019. Courts will be asked for cases filed with at least one felony charge, jails will be asked for bookings with at least one felony charge, and pretrial services agencies will be asked for cases opened with at least one felony charge. We will ask the agencies to include all information about each case until it is disposed. “Disposed” for courts is defined as a final finding by a judicial officer (typically a judge), and includes dismissal, nolle prosequi, placement on an inactive docket (stay of prosecution), placement in a diversion program, guilty, not guilty, acquittal, or other finding. “Disposed” for jails means that the person is released from custody as a release without a return prior to disposition (i.e., there is no rearrest for pretrial misconduct), sentenced by the courts to the jail or held pending transfer to another incarceration facility, or otherwise unable to be located before the end of the study (e.g., released pretrial, a bench warrant issued for some reason, but had not been rearrested). Often, jails assign unique booking identifiers each time a person is taken into jail, so BJS may need to provide an end date for the jail data extract. BJS will use March 15, 2020 for this purpose. For pretrial services agencies, “disposed” means that the pretrial agency is no longer responsible for monitoring the individual’s release, either because the release was revoked for misconduct or because the person completed pretrial release and was sentenced by the courts.

Courts, jails, and pretrial services agencies may provide data on all such cases in any format. BJS expects most will provide an unformatted data extract, where the data are extracted from the system “as-is” and BJS will work with the state to clean and standardize the data. Rarely, agencies may choose to provide a full system extract (“data dump”) of the entire case records system. In that case, BJS will extract the relevant cases.

Some courts, jails, or pretrial services agencies, or even entire counties in Class 1 may decline to provide data. These counties cannot be replaced, and BJS cannot substitute agency information from other counties (i.e., BJS cannot use data from a responding county as a substitute for a nonresponding county). BJS will use as much of the responding agencies’ data as possible and mark any elements not reported as missing. If the entire county fails to respond, BJS will either (a) have to adjust the coverage of the data; for example, to represent 73 counties rather than 75, or (b) use the participating largest 75 counties to represent those who do not participate. Once the nonparticipating counties are known (i.e., at the end of data collection), a determination will be made about each nonparticipating county as to whether any of the participating 75 can be used to represent it.

Class 2 - Sampling of Non-Certainty Counties

The goal of Class 2 of the NPRP is to develop representative estimates related to the pretrial release or detention ordered for defendants with at least one felony charge filed in state courts within one of the largest 200 counties not included in Class 1, or the largest 75 U.S. counties. As such, the Class 2 inferential population consists of the 76^th to 200^th largest counties in the country based on the 2019 American Community Survey 5-year population estimates (Table 3).

Sample Design. A random sample will be drawn such that the counties in which information is collected can be used to make inferences about all 125 counties. The sample size of Class 2 will be 50 counties in which all criminal cases filed with at least one felony charge will be collected.

Sample Stratification. While not much is known about the type and quantity of criminal cases filed with at least one felony charge in state courts in advance of data collection, a correlation with county population is assumed. Because county population size ranges from approximately 800,000 to 350,000 (Table 1), the sample will stratify the 125 counties by population size. Population size will be the only variable used to stratify the counties for two reasons. First, any other demographic information about the counties is likely to be highly correlated to population size and, therefore, will not add any additional information. Second, characteristics beyond county demographics are not known for all 125 counties.

Five strata will be created consisting of 25 counties each based on the rank ordering of the counties. In other words, the first stratum will consist of the 76^th to 100^th largest counties and the fifth stratum will consist of the 176^th to 200^th largest counties. These strata are designated strata 2 – 6 (Table 4) as stratum 1 is the Class 1 counties. Five strata were selected for two reasons. First, it kept the size differential between the largest and smallest county in a stratum relatively small. Second, five strata allow for an equal number of counties to be in each stratum (i.e., quintiles).

Table 4. Sample stratification

Stratum	Smallest County Population	Largest County Population
2	535,864	663,188
3	428,863	509,191
4	344,025	417,852
5	318,146	372,855
6	263,178	298,272

Sample Allocation. The sample will be allocated in a balanced fashion. This means an equal number of counties (i.e., 10) will be selected from each stratum. A balanced allocation is recommended to ensure there is representation from the smaller counties which may be different in terms of the outcomes of interest or characteristics of the pretrial population. Additionally, because the strata are of equal size, a balanced allocation of the sample produces an equal probability of selection for each sampled county.

Sample Selection. Within each stratum, a replicate/replacement design will be used for selected counties. Under a replicate design, the 25 counties within each stratum will be randomly assigned to a replicate. To form the replicates the 25 counties will be assigned a random number and ordered in descending fashion based on their random number. The initial replicate will consist of the first 10 randomly ordered counties. The remaining 15 counties will be assigned to a replicate of size one and used to replace one of the initial 10 counties if there is nonresponse (see next section).

Under this design, within each stratum, counties will be treated equally regardless of their population size. As such the probability of selection for each county in a stratum (h) will be

In other words, each county within a stratum will have an equal probability of selection.

An alternative to this design is a more traditional approach where a nonresponse rate is assumed and a larger than needed sample is selected. However, because the nonresponse rate is unknown and both a larger and smaller than desired sample size within each stratum is not desirable, this approach has too much uncertainty to be a viable option.

Accounting for Nonresponse. Nonresponse is likely to occur in both cycles of the study. Because the selection methods are different for each cycle, the method for addressing nonresponse will be tailored to the specific cycle.

Class 1. In Class 1, the largest 75 counties are treated as self-representing. That is, each county is selected with certainty and only represents itself. However, it is likely that some of these counties will not participate. This leaves two options: BJS will either (a) have to adjust the coverage of the data; for example, to represent 73 counties rather than 75, or (b) use the participating largest 75 counties to represent those who do not participate. Once the nonparticipating counties are known (i.e., at the end of data collection), a determination will be made about each nonparticipating county as to whether any of the participating 75 can be used to represent it.

For those with similarities to the participating counties, weighting classes (i.e., counties grouped together for the purpose of creating a weight adjustment) will be formed consisting of participating and nonparticipating counties. The weighting classes will be defined based on similar county-level characteristics such as population size, county demographic profile, and expected similarities in the types of felonies which occur. Within each weighting class, a ratio adjustment will be formed and applied to the sum of the base weights of each participating county (the base weight for each county is 1 since they are self-representing). In other words,

Where is the base weight for a responding county in weight class c and is an indicator of response for a given county.

Class 2. While it is anticipated that a high percentage of counties will participate, some counties – or a high number of agencies within the county – may not be able or willing to provide the requested information. Because a final sample of 50 counties is desired, a plan will be put in place to replace counties who cannot participate using the replicate design. The plan for accounting for nonresponse will be tied to the sample selection process. Specifically, because each county within a stratum has the same probability of selection and are considered similar to the nonparticipating county, the replicate counties in each stratum (i.e., counties 11 – 25 under the random ordering) will replace each nonparticipating county. The replacement counties will be selected in their random order (i.e., randomly ordered county 11 will be used first, county 12 second, etc.). Once 10 participating counties are identified, no further counties will be selected.

To adjust for nonresponse, a ratio adjustment of the participating counties over the total counties in the stratum (i.e., 25) will be applied. However, because each county has an equal probability of selection this adjustment will yield the same equal weights within each stratum.

2. Procedures for Collecting Information

In work done under BJS’s generic clearance (OMB Control No. 1121-0339), BJS interviewed county court, jail, and pretrial services agency leaders, many of whom reported that they would be able to provide most information in the form of data extracts from case management systems. A data extraction guide will be provided to all respondents (see Attachment 2).

At the start of the collection, BJS will email the state court, jail, and pretrial leaders in states with centralized statewide data. The letter will describe the purpose and importance of the collection, introduce the data collection agents (RTI International (RTI) and the National Center for State Courts (NCSC)), and invite the court, jail, and pretrial services agency to participate in the collection (Attachment 5). The following week, the same letter will be sent to county court, jail, and pretrial leaders in the counties without centralized court data systems. The same letters will be sent in staggered mailings to state and county leaders where some of the data are centralized at the state level and some of the data are maintained at the county level (e.g., the court data is held by a state agency, but pretrial and jail data are kept at the county agency level).

Once permission to collect data is obtained from the relevant contacts, RTI and NCSC will work with staff who manage the agency’s information system to obtain data files (Attachment 6). All data files will be submitted to RTI via a secure AWS GovCloud drive, RTI’s secure FTP, the agency’s secure FTP, or BJS’s secure BOX account. BJS is providing multiple options for submission to avoid difficulties in agency firewall or security issues. RTI will process the jail and pretrial services agency files, and NCSC will process the court data files on RTI’s secure AWS GovCloud drive, working with the respondent to evaluate data quality and completeness. NCSC is conducting the initial file processing because its analysts are more familiar with state court data from other NCSC projects, such as the Court Statistics Project. All identifiable files will be maintained on the AWS GovCloud drive during the data processing and merging. After NCSC conducts the preliminary processing of the court data, RTI will combine the court files with the pretrial and jail files.

After the files are processed, RTI will link the court, jail, and pretrial services agency data files using the personal identifiers provided. Once the files are linked, RTI will create a crosswalk of unique identifiers to replace any personally identifiable information (PII). The de-identified files will remain on the AWS GovCloud drive for further analysis, while the crosswalk will be moved to RTI’s secure project network. The de-identified file and crosswalk will not be stored in the same location unless it is necessary to update the de-identified file. In that event, a copy of the crosswalk will be moved to the AWS GovCloud, the data updated, and the crosswalk moved back to the RTI secure project drive.

As the data collection progresses, some courts, jails, and pretrial services agencies may decide not to participate. If this occurs, NCSC and RTI will continue the request from the remaining agencies in the county and will use as much of the data as possible to describe pretrial release from that county. The completeness of the data collection depends on how many agencies refuse in each county.

3. Methods to Maximize Response Rates

Every attempt will be made to collect complete information on felony criminal cases filed in state and county courts in 2019, to collect detention data from jails, and to collect pretrial release information from pretrial services agencies. BJS developed a project factsheet that has been circulated among court, jail, and pretrial services agencies in the largest 75 counties (Attachment 15). BJS also hosted a webinar, and provided links to the recorded webinar, available at RTI’s website (https://youtu.be/c1QFRxJldnA) and NCSC’s website (https://vimeo.com/604855587).

RTI and NCSC have already spoken with many of the data providers as part of the work done under BJS’s generic clearance. RTI and NCSC asked court, jail, and pretrial services agency leaders about their data systems and the policies that affect how they record the data. During these interviews, RTI and NCSC were able to explain the importance of the NPRP collection, and describe the products that may be published from the data collection.

The data extraction guides clearly articulate the data elements requested in the collection and the various acceptable data formats. RTI also maintains two main submission methods: AWS GovCloud and secure FTP. If agencies cannot access either, RTI can use the agency’s own FTP and move the data to the secure drive for processing. A final option is to allow the agency to submit data using BJS’s BOX account.

It is assumed that BJS will enter into data use agreements with some or all the state and county courts, jails, and pretrial services agencies. During the interviews, most of the agencies indicated that they would require both a data use agreement and some method of secure file transfer to participate in NPRP. Further, many agencies indicated that several personnel would need to review the data use agreements prior to agreeing to participate in the project and noted that time to review the agreement and data extract requests is important when considering participation in research projects.

A team of RTI and NCSC staff members will be assigned to act as the point of contact for each respondent. The data extraction guides for courts and for pretrial services agencies and jails include direct phone and email contact information for respondents. Additionally, RTI maintains a project email ([email protected]) monitored by the project director and data manager to respond to any technical questions.

4. Testing of Procedures

During the data interviews conducted under an earlier generic clearance, we asked whether the agency would be willing to provide a sample of their extracted data. Eight jurisdictions agreed, but BJS and RTI decided to follow up with seven.² The sites varied in terms of the agencies that were requested to provide data and the size of the population covered.

Table 8. Pilot test sites

Pilot Test Site No	County	Data systems	State	Region	Population
1	Allegheny County	Court, Jail, Pretrial Services	Pennsylvania	Northeast	989,647
2	El Paso County	Jail, Pretrial	Texas	South	614,939
3	King County	Jail, Pretrial	Washington	West	1,801,166
4	Middlesex County	Jail	Massachusetts	Northeast	1,296,600
5	Bexar County	Jail	Texas	South	1,497,113
6	New York City Criminal Justice Agency (multiple counties – Bronx, Queens, Kings, New York)	Pretrial	New York	Northeast	6,621,740
7	Orange County	Jail	Florida	South	1,087,438

RTI sent a follow-up email to the seven sites that reminded them of their voluntary participation in the pilot, the purpose of the NPRP and the pilot project, the BJS template data use agreement, and the draft data extraction guide that contained the data elements discussed in the data capacity interviews in November 2021. RTI followed up with reminder emails rather than following a more aggressive plan, in case the counties failed to respond and RTI would have to reach out again for the data after completing the OMB review process. Two jurisdictions (Allegheny and King) requested phone conversations to discuss the DUA requirements and the data extraction guides.

As of the end of February, RTI adjusted the nonresponse contact to every two weeks, and then in March to every week. As summarized in Table 9 below, our approach to information gathering yielded varying outcomes.

Table 9: Summary of Pilot Results

Jurisdiction	No Response	Held Call	Reviewed Data Request	Completed DUA	Closed Reason
Allegheny County, PA		X	X		Still negotiating DUA
El Paso County, TX			X		Still negotiating DUA
King County, WA		X	X		Experiencing backups due to Covid and IT emergencies (1/28/22)
Middlesex County, MA				X	Has login, has not submitted data
Bexar County, TX					Data received 4/1/2022
New York City Criminal Justice Agency (multiple counties – Bronx, Queens, Kings, New York), NY				X	Data received 3/10/2022
Orange County, FL			X		Still negotiating DUA.

As of April 4, 2022, New York City Criminal Justice Agency and Bexar County, TX submitted data. Middlesex County, MA completed the data use agreement with BJS and has the login information to submit data, but has not been responsive to email requests for a status update. Orange County, FL, Allegheny County, PA, and El Paso County, TX are still negotiating the data use agreement with BJS, but have agreed to submit data. King County, WA remains non-responsive to follow-up emails.

5. Contact for Statistical Aspects and Data Collection

The prosecution and judicial statistics unit staff at BJS are responsible for the overall design and management of the NPRP data collection, including the development of the data extraction guide and the analysis and publication of the data.

Erica Grasmick, Statistician

Judicial Statistics Unit

Bureau of Justice Statistics

810 7^th Street, NW

Washington, D.C. 20531

(202) 307-1402

Attachments (from CCSC to be updated to NPRP by BJS)

1. 34 USC § 10132

2. Data extraction guide

3. 60-day notice

4. 30-day notice

5. BJS introduction letter

5a. FAQs

6. Request for data

7. Initial follow-up script

8. Second follow-up

9. BJS final follow-up

10. Confirm data script

11. Thank you email

12. Collection closing script

13. Tyler Technologies Comments

14. Legal Rights Center Comments

15. Minnesota Freedom Fund Comments

16. NPRP Factsheet

17. Letter of Support

1 This definition excludes misdemeanors (other than those charged in addition to a felony charge), violations of probation and all civil cases, including traffic offenses (if charged civilly instead of criminally), municipal ordinance violations, infractions, fish and game commission charges, and habeas corpus petitions.

2 Harris County, Texas offered to be a pilot jurisdiction, but noted that pretrial data extracts would require court review and approval. BJS and RTI determined it would be burdensome to ask the court to review a data request for a pilot study and decided not to request data from Harris County until the final data collection.

24