SJCACC_SUPPORTING STATEMENT Part B

B. Collection of Information Employing Statistical Methods
1. Universe and Respondent Selection
The purpose of the SJCACC is to better understand the number, circumstances, and
characteristics of young offenders processed in criminal courts. States vary considerably, not just
with respect to the transfer of juveniles to adult court, but also with respect to the ability of their
courts to provide case-level data on these matters easily and electronically. The overall SJCACC
study and sample design attempts to both take advantage of and address this variability in data
availability by combining a census and sample-based data collection and estimation approach.
Research Questions, Analytic Objectives, and Data Items
The principal SJCACC research questions ask for the number of youth (i.e., persons under age
18) charged in criminal court, the characteristics of these youth, the pathways that led them to
criminal court, and the final disposition of their cases. In order to help answer these research
questions, the SJCACC will estimate the annual number of juveniles processed in adult criminal
court nationally, demographic case characteristics, their method of transfer (age, other), and final
case dispositions. Producing these estimates requires nationally representative case-level data
with variables pertaining to these estimates.
Population of Inference, Target Population, and Precision Requirements
The population of inference and target population for the 2014 SJCACC is all cases (felony and
misdemeanor, excluding traffic, fish and game commission charges) disposed in state criminal
courts in 2014 involving a defendant who was younger than the age of 18 at the time of the
crime.
Estimates of the total U.S. population, the population younger than the age of 18, the population
of inference and target population are provided in Table 1.

Page 1 of 20

Table 1.

2012 Population of inference and target population
2012 Population Estimates

N
1

Total U.S. population
Total U.S. juvenile population aged 10 to 171
Estimated number of persons under age 18 in adult courts2
Estimated number of persons under age 18 in adult courts via age exclusion2
Estimated number of juveniles in adult criminal courts via other exclusions2

313,914,000
33,253,200
195,800
174,500
21,300

This information suggests that approximately 0.6% (195,800/33,253,200) of the total U.S.
juvenile population aged 10 to 17 is involved in the adult criminal courts. Persons under age 18
alleged to have committed a crime can have their cases handled in criminal court if:
1) they live in the two states that define adult court jurisdiction to include all persons aged
16 or above or in the ten states that includes all persons aged 17 or above [otherwise
known as Age Exclusion] or
2) their cases were placed in adult court by means of juvenile court waiver, prosecutorial
discretion, or legislative mandate [otherwise known as Other Exclusions].
It is estimated that 89.1% (174,500/195,800) of cases of persons under the age of 18 are in the
adult system due to age exclusion.
Overall Study and Sample Design
The overall SJCACC study and sample design acknowledges the state-to-state variability in the
availability and accessibility of statewide, electronic court case-level records systems by
combining a census and sample-based data collection and estimation approach. In states with
such statewide, electronic court case-level records systems, all in-scope cases will be processed.
In other states, more traditional sample-based data collection methods will be appropriate. For
planning purposes, the SJCACC study and sample design assumes the following four
classifications (or primary strata) of states with respect to the availability and accessibility of
state-wide, electronic court case-level records systems:

1

National Center for Health Statistics (2013). Vintage 2012 postcensal estimates of the resident population of the
United States (April 1, 2010, July 1, 2010-July 1, 2012), by year, county, single-year of age (0, 1, 2, .., 85 years and
over), bridged race, Hispanic origin, and sex. Prepared under a collaborative arrangement with the U.S. Census
Bureau. Available online from http://www.cdc.gov/nchs/nvss/bridged_race.htm as of June 13, 2013, following
release by the U.S. Census Bureau of the unbridged Vintage 2012 postcensal estimates by 5-year age group on June
13, 2013. [Retrieved 7/1/2013]
2

Estimates of cases or persons under the age of 18 processed in criminal courts were developed by the National
Center for Juvenile Justice based on an understanding of state-specific juvenile populations, the legislation that
determines the range of juvenile court jurisdiction, and available state-specific estimates of juvenile transfers to
adult court.

Page 2 of 20

Stratum 1.

States with expected 100% coverage (county) and complete (data items)

Stratum 2.

States with expected 100% coverage (county) and some incomplete (data
items) – later to be treated as item-missing data

Stratum 3.

States with most but not all counties – incomplete coverage statewide –
uncovered counties to be included in the Primary Sampling Unit (PSU)
sampling frame

Stratum 4.

States where the statewide approach will not work and all such counties are
to be included in PSU sampling frame

A census of cases will be obtained from all counties in the states in strata 1 and 2 and all counties
in the state electronic record system in Stratum 3. A sample of PSUs and cases will be used to
represent the balance of states and counties in strata 3 and 4. Figure 1 depicts the availability of
state electronic data.

Page 3 of 20

Figure 1.

SJCACC - Diagram of Electronic Data Availability by State

Suppose that the United States consisted of nine states as shown in Figure 1 above. All data will
be collected for SJCACC from states with easily accessible statewide electronic court record
systems (i.e., states 1, 4, 6 and 8). Some states will have easily accessible electronic court record
systems for the majority of the counties (i.e., state 5), and therefore all the counties with these
data will be collected. These states are considered states with partial electronic data. Once this
electronic data harvesting is complete (represented by the blue shaded states and areas), PSUs

Page 4 of 20

will be formed, and a PSU sample will be selected to represent the balance of the United States
(i.e., the non-blue shaded states), in order to allow for unbiased national estimates.
Figure 2 depicts how this approach will actually be implemented, with estimates and
assumptions on the number of states and counties involved. The statewide approach (i.e., census)
and the probability sample are further described in the sections below.

Page 5 of 20

Figure 2.

SJCACC - Data Availability Decision Tree

Cases will be divided into two groups: 1) those that are retrieved from statewide electronic
record systems and 2) all others. The precision of estimates of the combined groups will be
sensitive to the percentage of the population covered by each group. An expression for the
variance of a proportion assuming two groups would be as follows:

Page 6 of 20

2

𝑊ℎ2 𝑝ℎ 𝑞ℎ
𝑉(𝑝) = ∑
𝑛ℎ
ℎ=1

Where:
𝑝
𝑞
ℎ

=
=
=

proportion
1–𝑝
group

𝑊ℎ2

=

the group population percentage, 𝑊ℎ2 = (𝑁 ℎ )

𝑛ℎ

=

group case sample size

𝑁

2

This calculation, and calculations like it, will be used to estimate the precision of the proposed
study and sample design. For the purposes of this discussion, BJS has assumed the following:
Table 2.

Target population by census/sample approach

Population
Total Population (juveniles in adult criminal courts)
Population in first 3 strata (where statewide approach is to be
attempted)
Statewide approach success rate assumed
Population covered by successful statewide approach
Balance of population (to be covered by PSU sample)

N
195,800
165,100

%
100%
84%

-123,800
72,000

75%
63%
37%

The Statewide Approach
A final list of states for which the statewide approach is possible will be developed and the
statewide approach will be attempted within these states. Under the statewide approach, the state
will be asked to confirm that it has a statewide (or nearly statewide), electronic case-level records
system which can provide all or most of the relevant SJCACC data items. The state will then be
requested to provide all such cases in one of several ways. The data obtained from the statewide
approach is a census of all cases within and across such states.
The 26 States where BJS expects to attempt the statewide approach are:
Alabama
Arizona
Colorado
Connecticut
DC

Page 7 of 20

Illinois
Iowa
Maryland (part)
Massachusetts
Michigan
Minnesota
Montana
Nebraska
New Hampshire
New Jersey
New Mexico
New York
North Carolina
Pennsylvania
Rhode Island
South Carolina
South Dakota
Vermont
Virginia
Washington
Wisconsin
Probability Sampling – PSUs and Cases
States and counties where the statewide approach will not work [part of Stratum 3 and all of
Stratum 4] will be divided into primary sampling units (PSUs) that will consist of one or more
contiguous counties. The formation and sampling of PSUs for the SJCACC will consist of the
following four steps:
1.

Calculating the minimum PSU measure of size (MOS)

2.

PSU Formation

3.

Stratifying PSUs

4.

Sampling PSUs and cases within PSUs

Calculating the Minimum PSU MOS
In probability proportional to size (PPS) sampling, the variance of an estimator is reduced if the
outcome variable(s) is approximately proportional to the measure of size (MOS) used. For the

Page 8 of 20

SJCACC, a number of outcome variables are of interest, and a good measure of size will be
correlated with many of the variables of interest, and will also be relatively stable. For the
SJCACC, possible MOS include the total population aged 10-17, and the number of juvenile
arrests. The former is available for all counties in the U.S. and with reasonable accuracy, while
the latter is not. Therefore, for the SJCACC, total population aged 10-17 is used as the MOS.
A minimum PSU MOS is needed as the first step in forming primary sampling units (PSUs).
Forming all PSUs larger than the minimum MOS ensures that the PSUs selected for the sample
will have sufficient numbers of cases to support the research objectives of the SJCACC. The
minimum MOS is calculated such that a self-weighting sample, overall or within subgroups can
be selected across the sampled PSUs. The minimum MOS requirement for PSUs is usually
determined by solving the following equation:
ℎ
∑𝑁
𝑖=1 𝑀𝑂𝑆𝑖
𝑀𝑖𝑛(𝑀𝑂𝑆𝑖 ) ≥ 𝑓𝑚𝑎𝑥
,
𝑛
Where
𝑀𝑖𝑛(𝑀𝑂𝑆𝑖 )
𝑓𝑚𝑎𝑥
𝑀𝑂𝑆𝑖
𝑛

=
=
=
=

minimum measure of size for PSUα;
largest overall sampling rate;
measure of size for PSUα; and
number of PSUs to be selected.

PSU Formation
In order to draw a PPS sample of PSUs, a frame of PSUs has to be created first. The SJCACC
PSUs will be formed according to the following criteria:


PSUs will be formed as counties or groups of adjacent counties



PSUs will be required to achieve a minimum (with few exceptions) PSU MOS



PSUs will respect region and state lines

To form PSUs, the Westat proprietary software WESPSU will be used. The software uses an
optimization approach that either:


minimizes the end-to-end PSU distance of the formed PSU as a proxy for travel and
other within PSU labor-related costs, or



maximizes the within PSU heterogeneity in an attempt to reduce the design effect
due to clustering

Page 9 of 20



subject to a maximum distance constraint while meeting the county contiguity,
minimum MOS, region and state constraints listed above.

For the SJCACC, BJS will choose to minimize the end-to-end PSU distance as a precaution, in
case more labor-intensive approaches are required to obtain the data within at least some
sampled PSUs.
The PSU formation algorithm starts with designating each county as a sufficient or a deficient
county. A sufficient county is one that meets all the requirements of becoming a PSU by itself
(as determined by the user), and a deficient county is one that fails to meet one or more of the
requirements. A deficient county is combined with one or more other contiguous deficient
counties, if available. Otherwise, a deficient county is combined with a contiguous sufficient
county. As there is often more than one choice of combining counties to form a sufficient PSU
around a deficient county, the software uses the specified objective (i.e., to minimize distance) to
identify the optimal combination for each deficient county. Once an optimal solution is found for
each deficient county, the counties constituting that PSU are moved from the deficient and
sufficient list and into a solved list. With these counties removed, the process is repeated for all
remaining deficient counties, using all remaining deficient and sufficient counties. Through this
iteration, ideally all counties in the deficient list are eventually moved into the solved list. In
some cases, when a deficient county cannot be merged with any of the contiguous counties
because of other constraints, the software leaves that county as unresolved. These counties are
then combined outside the software with other counties to form PSUs.
Since the success of the statewide approach is an unknown, PSUs will be formed in all states,
with the resulting PSUs respecting state lines and with each PSU containing a minimum measure
of size. Approximately 420 PSUs will be formed for the 50 states and DC, with an estimated 80
of these PSUs located in Strata 3 and 4. This PSU formation will be used where the state-wide
approach does not succeed. The PSU sample will be drawn separately for each state where the
state-wide approach does not succeed, or across a group of such states. The latter will be
preferable for sample design efficiency if the number of such states is relatively large.
Stratifying PSUs
PSUs within states that do not have available, accessible, and complete electronic data systems,
where data collection costs are expected to be higher, will be stratified further into substrata
using two categories based on the nature of the handling of juveniles within the state. One
substrata will include those states with juvenile court jurisdiction ending at age 15 or 16 (i.e.,
states with age exclusion laws); the other will include states in which original juvenile court
jurisdictions ends at age 17. Eight PSUs will be selected within each of the two substrata using a
systematic PPS sampling with size as an implicit sort variable. Two additional PSUs will be
sampled containing small PSUs as described below.
Page 10 of 20

Sampling PSUs and Cases within PSUs
Eighteen PSUs will be selected with probability proportional to size. The total number of PSUs
will be determined by the success of obtaining statewide electronic case-level data and the
percent of the population covered by these data. In selected counties in which electronic data
systems exist, all cases will be processed. In other instances, a sample of cases will be drawn
from physical files at fixed overall rates in order to provide unbiased estimates. Within a county
cases will be divided into 2 lists: 1) those in criminal court due to age exclusion and 2) those in
criminal court due to some form of transfer. Equal number of cases will be sampled from each
list.
Very small counties (defined in terms of the population of 10-17 year olds) in these states will be
assigned to their own PSU and to a specific substratum. Only one or two such counties will be
selected with the expectation that no or a very small number of study-eligible cases will be found
within them. The study estimates will, therefore, be unbiased and the effect of small counties on
the study sample sizes and the resulting precision of the estimates will also be controlled via the
stratification.
Note that the number of states where the statewide approach is not successful may exceed 18, in
which case some such states will not receive any sampled PSUs. This is not uncommon in
national, multistage probability samples and has no real implications for the national
representativeness or inferential capabilities of the sample.
A precision objective of a standard error of 1 percent on a proportion with a value that could
range from 30 percent to 70 percent was set for comparison and discussion purposes, given the
overall study and sample design framework laid out earlier in the Overall Study and Sample
Design section. Under those assumptions, a sample random sample size of at most 341 cases
from a simple random sample would be needed, in conjunction with the statewide census, to
achieve that precision objective (see Table 3). Since the proposed design is a two-stage design,
the actual sample sizes required will have to exceed the values in Table 3, in order to take into
account the design effect due to clustering, and produce an effective sample (i.e., the actual
sample size / design effect) size greater or equal to the sample sizes given in Table 3. Given 18
PSUs with an average of 40 cases per PSU, and a maximum of 50 cases per PSU, the total actual
sample size for the SJCACC would be 720 cases. Assuming an intraclass correlation of 0.025
(which is a plausible value for clusters of counties or groups of counties), a design effect of
1.975 and an effective sample size of 365 cases results from solving the following equations:
𝐷𝐸𝐹𝐹 = 1 + 0.025 (40 − 1)

Page 11 of 20

720
1.975
The effective sample size of 365 cases exceeds all values shown in Table 3, indicating that the
probability sample is sufficient to meet these precision objectives, under the assumptions laid out
above.
𝑒𝑓𝑓𝑓 𝑛 =

Table 3.

P
30.00%
40.00%
50.00%
60.00%
70.00%

Sample sizes required for a .01 ste, for various proportions

Q
70.00%
60.00%
50.00%
40.00%
30.00%

N (covered by
PSU sample)
72,045
72,045
72,045
72,045
72,045

N
(PSU sample)
286
327
341
327
286

Var
(overall)
0.0001
0.0001
0.0001
0.0001
0.0001

Ste (overall)
0.010000008
0.01
0.01
0.01
0.01

RSE
3.33%
2.50%
2.00%
1.67%
1.43%

LCI
28.04%
38.04%
48.04%
58.04%
68.04%

UCI
31.96%
41.96%
51.96%
61.96%
71.96%

Nonresponse and Substitution
A minimum of 90 percent response rate is assumed for this study. Since court records are
involved, virtually all nonresponse will be at the PSU level. Three different approaches can be
used to compensate for PSU nonresponse, depending on the expected and actual response rate.
Nonresponse adjustment can be used to adjust the PSU weights of responding PSUs to the PSU
weights of all sampled PSUs. Although unbiased estimates might result, a shortfall in cases and
consequently precision might also result. Alternatively, the actual PSU sample could be inflated
by the inverse of the expected response rate. This works well when the actual response rate can
be estimated accurately in advance. A third possibility is to use release groups of PSU samples,
the exact number of groups released to be determined as the sample is implemented, and actual
response rates manifest themselves.
It should be noted that partial response of a PSU can occur when some, but not all, of the courts
within the counties constituting the PSU participate in the study. Under these circumstances, if
the courts not participating are small (in juvenile case counts), then weight adjustments might be
used. If the courts not participating are large, then this is tantamount to PSU nonresponse, and
the above procedures would be used.
Alternative Scenarios and Contingency Plans
The plan described above assumes a certain level of success with the statewide approach.
Specifically, Table 2 assumes that the statewide approach success rate is 75 percent. Since the
likely success of the statewide approach is uncertain, it would be useful to consider a few
Page 12 of 20

alternative scenarios and demonstrate the contingency plans that are essentially built into the
overall design.
Two alternative scenarios will be considered. Scenario 2 will assume that the statewide approach
success rate is 25 percent. Tables 4 and 5 below, which are similar to Tables 2 and 3 presented
above, show the effects of this assumption.
Table 4.

Target population by census/sample approach – scenario 2

Total Population (juveniles in adult criminal courts)
Population in first 4 strata (where statewide approach is to be attempted)
Statewide approach success rate assumed
Population covered by successful statewide approach
Balance of population (to be covered by PSU sample)

Table 5.
P
30.00%
40.00%
50.00%
60.00%
70.00%

N

%

195,800
165,100
-41,300
154,500

100%
84%
25%
21%
79%

Sample sizes required for a .01 ste, for various proportions – scenario 2

Q N (covered by PSU sample) n (PSU sample) Var (overall) Ste (overall)
70.00%
154,577
1,311
0.0001
0.01
60.00%
154,577
1,499
0.0001
0.01
50.00%
154,577
1,562
0.0001
0.01
40.00%
154,577
1,499
0.0001
0.01
30.00%
154,577
1,311
0.0001
0.01

RSE
3.33%
2.50%
2.00%
1.67%
1.43%

LCI
28.04%
38.04%
48.04%
58.04%
68.04%

UCI
31.96%
41.96%
51.96%
61.96%
71.96%

Under scenario 2, a sample random sample size of at most 1,562 cases from a simple random
sample would be needed, in conjunction with the statewide census, to achieve the stated
precision objective (see Table 5). The actual sample sizes required, taking into account the 1.975
design effect due to clustering, would be 3,085. Assuming an average of 40 cases per PSU, 77
PSUs would be required.
Scenario 3 will assume that the statewide approach success rate is 50 percent. Tables 6 and 7
below, which are similar to Tables 2 and 3 presented above, show the effects of this assumption.

Page 13 of 20

Table 6.

Target population by census/sample approach – scenario 3

Total Population (juveniles in adult criminal courts)
Population in first 4 strata (where statewide approach is to be attempted)
Statewide approach success rate assumed
Population covered by successful statewide approach
Balance of population (to be covered by PSU sample)

Table 7.
P
30.00%
40.00%
50.00%
60.00%
70.00%

N
195,800
165,100
-82,500
113,300

%
100%
84%
50%
42%
58%

Sample sizes required for a .01 ste, for various proportions – scenario 3

Q N (covered by PSU sample) n (PSU sample) Var (overall) Ste (overall)
70.00%
113,311
706
0.0001
0.01
60.00%
113,311
808
0.0001
0.01
50.00%
113,311
841
0.0001
0.01
40.00%
113,311
808
0.0001
0.01
30.00%
113,311
706
0.0001
0.01

RSE
3.33%
2.50%
2.00%
1.67%
1.43%

LCI
28.04%
38.04%
48.04%
58.04%
68.04%

UCI
31.96%
41.96%
51.96%
61.96%
71.96%

Under scenario 3, a sample random sample size of at most 841 cases from a simple random
sample would be needed, in conjunction with the statewide census, to achieve the stated
precision objective (see Table 7). The actual sample sizes required, taking into account the 1.975
design effect due to clustering, would be 1,660. Assuming an average of 40 cases per PSU, 42
PSUs would be required.
In the event of such statewide approach success rates, the overall two stage design will apply;
however finer stratification of PSUs would be employed to reflect the increased importance of
the probability sample and its role and contribution to the overall precision.
Weighting, Estimation and Variance Estimation
The data obtained for estimation purposes will be obtained from essentially a combination of a
census and a probability sample. Data weighting, acknowledging and reflecting the probability
sampling, will be required in order to enable unbiased estimation. The data weighting proposed
will consist of calculating base weights, nonresponse adjustment factors, and poststratification
factors. Base weights will be calculated as the inverse of the overall probability of selection, for
all sampled cases, reflecting any variability in the probabilities of selection across all stages.
Nonresponse adjustment factors will be calculated reflecting PSU or case nonresponse, as the
ratio of the sum of base weights for all units (i.e., at a given stage) to the sum of base weights for
all responding units. Note, however, that case nonresponse is less likely for studies involving
court records. Finally, poststratification or raking factors will be calculated to adjust weighted
Page 14 of 20

(via the nonresponse adjusted weights) estimates to known population counts, where the known
counts are obtained from the sampling frames and statewide files, or possibly counts from the
2014 National Survey of Prosecutors (a census with 90+ percent response rate). The 2014 NSP is
asking each prosecutor office in the nation to provide counts of juvenile transfer cases disposed
by their office in 2014 by the three methods of transfer (i.e., the three excluding jurisdictional
age laws).
The final resulting weight will be used for estimation. Variance estimation will be supported by
providing strata and PSU variables for SUDAAN or other Taylor Series methods, as well as a set
of replicate weights. The jackknife or balanced repeated replication (JK2 or BRR) approach
would be used since variance estimation for medians might be used.

2. Procedures for Collecting Information
In 2010, during initial research in preparation for preparing the competitive application for this
project, the National Center for Juvenile Justice (NCJJ) discovered that two-thirds of the states
had some sort of automated case-level data system that included all of the counties in their states.
Additionally, almost all of those states had a way to extract the necessary data and would be able
to provide a data dump, non-uniform extract or a uniform extract. Operating under the
assumption that improvements in technology have had a positive effect on data availability, we
believe that most of the necessary data will be available through electronic downloads.
For those respondents that are able to provide electronic data, they will be able to submit the files
as an unformatted “data dump,” as a non-uniform data extract, or as a uniform extract (see
Section 4, Testing of Procedures). A data extraction guide will be provided to all respondents so
they may determine which submission format is most feasible and what information will be
necessary for those providing electronic data (see Attachment 4).
Many of the data suppliers were identified early on through BJS’ State Court Processing
Statistics (SCPS) and National Judicial Reporting Program (NJRP) collections. At the state level
and in many counties these contacts will consist of the Court Administrator. The contractor will
work with Administrative Offices of the Court to update the list of data suppliers and supplement
the list for those counties in which contact information is missing. Once the list is complete, the
contacts will be mailed an invitation from BJS that will explain the importance and purpose of
the collection, introduce the data collection agents, and invite the court to participate in the
collection.
Once permission to collect data is obtained from the relevant contacts, the contractor will work
with staff who manage the courts’ information system to obtain data files, for those submitting
electronic data. The contractor will process the files, working with the respondent to evaluate

Page 15 of 20

data quality and completeness. The contractor will determine the unit of analysis in the files
(case or charge) and restructure the files into a standardized unit of count as needed.
For those respondents who do not have statewide electronic systems and whose data are selected
by PSU, they will be asked to enter case-level data onto a hard copy TeleForm survey. A survey
guide and a draft of the hardcopy survey with the proposed questions and response options is
attached (Attachments 5 and 6). The final formatted TeleForm survey will differ from the draft
survey in that the formatting will enable receipt of the data, review, scanning and data
verification, as well as quality control. The scanned data are stored directly into a database that
tracks the disposition and date of each survey. Extracted data from scanned questionnaires are
subject to pre-determined field validations that are programmed in the design phase according to
project specifications. If a data value violates validation rules, the data are flagged for review by
verifiers who can interactively review the images and the corresponding extracted data, and
resolve validation errors.

3. Methods to Maximize Response Rates
Every attempt will be made to collect complete and accurate information on all sampled persons
under age 18 processed in adult criminal court. In order to maximize the response rate and
minimize non-response bias, the data collection agents will work closely with respondents to
ensure that they understand every step of the data collection process, including determining the
most appropriate format for data submission and obtaining the necessary approvals for providing
case level data. A help desk will be developed to address respondents’ technical questions and to
proactively address survey nonresponse throughout data collection.

4. Testing of Procedures
The variables selected for the study were chosen by BJS, with input from an advisory committee
from OJJDP, and through a conference on criminal case processing of juveniles.
After the selection of relevant variables, BJS and Westat staff attended a court data specialists’
workshop in Williamsburg, VA in September of 2012. This was a yearly workshop funded by
BJS, conducted by the National Center for State Courts and attended by approximately 20 states.
At this meeting, data specialists who manage state court information systems discussed the
content and coverage of their systems and how those systems can be improved. BJS gave a
presentation to discuss the redesign of its courts program and received feedback regarding the
feasibility of conducting collections using the methods employed by the SJCACC.
In 2014, four states (Pennsylvania, New York, New Jersey, and Maryland) were interviewed by
BJS and Westat about their data systems, whether they contained the variables of interest, and
Page 16 of 20

potential challenges accessing the data. Each state discussed their procedures for releasing data,
if one was in place, and the contents and accessibility of its information system.
In Pennsylvania and New York it was determined that a file already available to researchers and
others contained all needed variables for the SJCACC collection. The process to obtain the data
for each state was explained. In Maryland a file that contained most counties in the state and all
needed variables for the SJCACC was available and Maryland officials indicated that it was
probable that the three missing counties would have the needed information. New Jersey
officials indicated that a file containing all of the needed variables could be assembled from
multiple existing files.
The survey instrument was developed in consultation with court officials. Feedback was
provided regarding different terms in different jurisdictions and the ability and time needed to
complete the survey.
NCJJ conducted a pretest of the hardcopy data collection survey, Survey guide, and Electronic
extraction guide (see Attachments 4 and 5) in an effort to gather feedback on the project’s data
collection protocols, and selected eight state contacts. Of the eight contacts, five responded to
questions about their ability to identify the target population, answer the survey questions and the
length of time to respond to the survey.
All but one the states indicated that they could identify defendants under 18 at the time of arrest
and would be able to provide some or all of the requested data. Focusing on the remaining four
states that could identify this population and had some or all of the data elements, they responded
in the affirmative to the ability to answer the questions and the ability to provide the data in
electronic format. Most, however, would not be able to provide a uniform extract and would only
be able to provide a data dump, a non-uniform extract, a combination of the two, or a hardcopy
survey.
5. Contacts for Statistical Aspects and Data Collection
The prosecution and adjudications staff at BJS take responsibility for the overall design and
management of the SJCACC data collection, including the development of the questionnaires
and the analysis and publication of the data.
a) BJS contacts include:
Howard Snyder, Deputy Director
Bureau of Justice Statistics
810 7th Street, NW
Washington, D.C. 20531
(202) 616-8305

Page 17 of 20

Tracey Kyckelhahn, Statistician
Prosecution and Adjudications Unit
Bureau of Justice Statistics
810 7th Street, NW
Washington, D.C. 20531
(202) 353-7381
b) Persons consulted on statistical methodology:
Jim Green
Westat
1600 Research Boulevard
Rockville, MD 20850
(301) 251-1500
c) Persons consulted on data collection and analysis:
Elissa Rumsey, Social Science Program Analyst
Office of Juvenile Justice and Delinquency Prevention
810 7th Street, NW
Washington, D.C. 20531
(202) 616-9279
Kathi Grasso, Senior Policy/Legal Advisor
Office of Juvenile Justice and Delinquency Prevention
810 7th Street, NW
Washington, D.C. 20531
(202) 616-7567
Janet Chiancone, Associate Administrator
Office of Juvenile Justice and Delinquency Prevention
810 7th Street, NW
Washington, D.C. 20531
(202) 353-9258
Patrick Griffin, Senior Research Associate
National Center for Juvenile Justice
3700 South Water Street, Suite 200
Pittsburgh, PA 15203

Page 18 of 20

Melissa Sickmund, Director
National Center for Juvenile Justice
3700 South Water Street, Suite 200
Pittsburgh, PA 15203
Megan Kurlychek, Associate Professor
University at Albany
135 Western Avenue
Albany, NY 12222
Richard Schauffler, Director of Research Services
National Center for State Courts
300 Newport Avenue
Williamsburg, VA 23185

Frank Broccolina, State Court Administrator (since retired)
Maryland Judiciary, Administrative Office of the Courts
Maryland Judicial Center
580 Taylor Avenue
Annapolis, MD 21401
Andrew Ginder, Senior Researcher
Maryland Judiciary, Administrative Office of the Courts
Court Operations
2001 E/F Commerce Park Drive
Annapolis, MD 21401
Barbara Holmes, Enterprise Design Architect
Administrative Office of the Pennsylvania Courts
5035 Ritter Road
Mechanicsburg, PA 17055
Kim Nieves, Director of Policy and Research
Administrative Office of the Pennsylvania Courts
1515 Market Street, Suite 1414
Philadelphia, PA 19102
Mark Davies, Chief
Quantitative Research Unit

Page 19 of 20

New Jersey Administrative Office of the Courts
P.O. Box 037
Trenton, NJ 08625
Chip Mount, Director of Research and Technology
New York, Office of Court Administration
25 Beaver Street, Room 852
New York, NY 10004
Ron Younkins, Director
New York, Office of Court Administration
25 Beaver Street, Room 852
New York, NY 10004
Gene Meyers, Director of Operations
New York, Office of Court Administration
25 Beaver Street, Room 852
New York, NY 10004

Attachments
1.
2.
3.
4.
5.
6.
7.
8.

Title 42
BJS Participant Letter
Juvenile and U.S. Population Counts
2014 SJCACC Electronic Extraction Guide
2014 SJCACC Hard Copy Extraction Guide
2014 Survey of Juveniles Charged in Adult Criminal Court
Follow up letter for electronic data
Follow up letter for survey

Page 20 of 20