SSV 2009 Sample Design

Attachment 3
SSV 2009 Sampling Design

Page 1 of 6
Privately-operated State and Federal Prison Sample Design
for the 2009 Survey on Sexual Violence
417 units on the frame
Revision Date: March 15, 2010
The Bureau of Justice Statistics (BJS) instructed the Census Bureau to use the 2005 Prison
Census file as the frame for this sample. The 2005 Prison Census file contained 417 records for
privately-operated state and federal prisons. BJS requested a sample of 125 units. This is an
increase of 40 units from the 2008 private prison sample. The increase in 2009 is an attempt to
produce better standard errors than were achieved in the 2008 Survey on Sexual Violence (SSV).
Some facilities are large compared to the rest, so we had to use a certainty cutoff to select some
of the facilities as certainties due to size. A facility was declared a certainty due to size if it had
average daily population (ADP) of 488 or more. There are 71 size certainties in the 2009
sample.
The rest of the file was serpentine-sorted by region, two-digit state code, and ADP. Region is the
region of the country where the facility is located: Northeast, Midwest, South, or West.
We used PROC SURVEYSELECT in SAS to select a systematic probability proportional to size
sample.
Each noncertainty privately-operated state or federal prison in sample has a weight based on its
measure of size. The weights are shown in Table 1.
We verify the sample weights by using Horvitz-Thompson estimation. We use the sample to
estimate the national ADP. The estimated national ADP is
where yi is the ADP of the ith sample unit. The national sum of the measure of size is 106,014.
Table 2 shows the estimated coefficients of variation (CVs) for this sample design.

Page 2 of 6
Table 1. Sample design table for privately-operated state and federal prisons
Obs

ID

Measure of size

Weight

Measure of size * weight

1 038000000074700000000

625

1.0000

625.0000

2 038000000076000000000

984

1.0000

984.0000

3 038011666072200000000

1,321

1.0000

1,321.0000

4 058000000079992200000

555

1.0000

555.0000

5 058000000079992300000

532

1.0000

532.0000

6 058000000079992400000

529

1.0000

529.0000

7 058015666070200000000

2,293

1.0000

2,293.0000

8 058015666072000000000

2,631

1.0000

2,631.0000

9 068000000072200000000

697

1.0000

697.0000

10 068000000072900000000

733

1.0000

733.0000

11 068000000073100000000

1,100

1.0000

1,100.0000

12 068000000073200000000

752

1.0000

752.0000

13 108000000079881000000

689

1.0000

689.0000

14 108000000079992000000

1,037

1.0000

1,037.0000

15 108000000079996000000

745

1.0000

745.0000

16 108000000079997000000

747

1.0000

747.0000

17 108000000079999100000

1,643

1.0000

1,643.0000

18 118000000075700000000

1,495

1.0000

1,495.0000

19 118000000075800000000

1,500

1.0000

1,500.0000

20 118000000076000000000

1,649

1.0000

1,649.0000

21 118134666079981000000

1,568

1.0000

1,568.0000

22 138000000071700000000

1,250

1.0000

1,250.0000

23 188000000071400000000

805

1.0000

805.0000

24 188000000071900000000

770

1.0000

770.0000

25 198000000071400000000

1,570

1.0000

1,570.0000

26 198000000072300000000

1,557

1.0000

1,557.0000

27 248000000070999100000

919

1.0000

919.0000

Page 3 of 6
Obs

ID

Measure of size

Weight

Measure of size * weight

28 258000000073100000000

993

1.0000

993.0000

29 258000000073500000000

962

1.0000

962.0000

30 258000000073600000000

976

1.0000

976.0000

31 258000000073700000000

867

1.0000

867.0000

32 258000000074000000000

1,002

1.0000

1,002.0000

33 258000000074100000000

883

1.0000

883.0000

34 278000000070991000000

541

1.0000

541.0000

35 318000000075100000000

539

1.0000

539.0000

36 318000000075700000000

500

1.0000

500.0000

37 328000000071300000000

630

1.0000

630.0000

38 328000000071400000000

1,140

1.0000

1,140.0000

39 328000000071500000000

596

1.0000

596.0000

40 328000000071600000000

1,200

1.0000

1,200.0000

41 348046666079111100000

1,300

1.0000

1,300.0000

42 368000000073900000000

720

1.0000

720.0000

43 368000000074100000000

567

1.0000

567.0000

44 368000000074200000000

1,417

1.0000

1,417.0000

45 378000000074700000000

807

1.0000

807.0000

46 378000000074800000000

974

1.0000

974.0000

47 378000000074900000000

952

1.0000

952.0000

48 378000000075000000000

1,892

1.0000

1,892.0000

49 378000000076100000000

1,893

1.0000

1,893.0000

50 438000000072300000000

1,630

1.0000

1,630.0000

51 438000000073500000000

1,970

1.0000

1,970.0000

52 438000000073700000000

1,487

1.0000

1,487.0000

53 448000000073700000000

519

1.0000

519.0000

54 448000000073800000000

518

1.0000

518.0000

55 448000000073900000000

997

1.0000

997.0000

Page 4 of 6
Obs

ID

Measure of size

Weight

Measure of size * weight

56 448000000074100000000

2,040

1.0000

2,040.0000

57 448000000078020000000

832

1.0000

832.0000

58 448000000079200000000

488

1.0000

488.0000

59 448000000079800500000

1,973

1.0000

1,973.0000

60 448000000079930000000

498

1.0000

498.0000

61 448000000079940000000

517

1.0000

517.0000

62 448000000079996000000

1,023

1.0000

1,023.0000

63 448000000079996700000

520

1.0000

520.0000

64 448000000079997000000

999

1.0000

999.0000

65 448000000079999000000

1,047

1.0000

1,047.0000

66 448000000079999200000

2,196

1.0000

2,196.0000

67 448048666070100000000

1,403

1.0000

1,403.0000

68 448085666075110000000

1,105

1.0000

1,105.0000

69 448114666070200000000

2,587

1.0000

2,587.0000

70 448195666070100000000

2,162

1.0000

2,162.0000

71 478000000078500000000

1,820

1.0000

1,820.0000

72 028000000072300000000

102

4.6543

474.7386

73 038000000072800000000

395

1.2019

474.7505

74 038000000073400000000

387

1.2267

474.7329

75 048060666073200000000

70

6.7820

474.7400

76 058000000079800000000

38

12.4931

474.7378

77 058000000079993800000

102

4.6543

474.7386

78 058037666079995000000

226

2.1006

474.7356

79 068000000076100000000

95

4.9973

474.7435

80 068000000077200000000

60

7.9123

474.7380

81 068000000079000000000

184

2.5801

474.7384

82 068000000079200000000

118

4.0232

474.7376

83 068000000079800000000

201

2.3619

474.7419

Page 5 of 6
Obs

ID

Measure of size

Weight

Measure of size * weight

84 068000000079900000000

297

1.5984

474.7248

85 068000000079910000000

40

11.8685

474.7400

86 078000000074700000000

18

26.3744

474.7392

87 078000000076000000000

109

4.3554

474.7386

88 078000000076300000000

57

8.3288

474.7416

89 108000000075850000000

104

4.5648

474.7392

90 108000000079922200000

80

5.9342

474.7360

91 108036666076260000000

40

11.8685

474.7400

92 148000000075500000000

293

1.6203

474.7479

93 148000000076200000000

120

3.9562

474.7440

94 178052666072200000000

145

3.2741

474.7445

95 188000000072450000000

56

8.4775

474.7400

96 188000000072460000000

194

2.4471

474.7374

97 188000000072600000000

445

1.0668

474.7260

98 198000000072010000000

49

9.6885

474.7365

99 238082666079980000000

50

9.4948

474.7400

100 268096666075900000000

117

4.0576

474.7392

101 278000000070600000000

166

2.8599

474.7434

102 298002666073400000000

75

6.3299

474.7425

103 318000000074200000000

395

1.2019

474.7505

104 318000000075000000000

153

3.1029

474.7437

105 318000000076200000000

35

13.5640

474.7400

106 338031666070200000000

90

5.2749

474.7410

107 348000000079700000000

20

23.7369

474.7380

108 368000000075400000000

53

8.9573

474.7369

109 368000000076100000000

72

6.5936

474.7392

110 368000000077500000000

134

3.5428

474.7352

111 368000000077700000000

116

4.0926

474.7416

Page 6 of 6
Obs

ID

Measure of size

Weight

Measure of size * weight

112 378000000075400000000

292

1.6258

474.7336

113 378000000075500000000

280

1.6955

474.7400

114 378000000078900000000

115

4.1282

474.7430

115 398000000075700000000

216

2.1979

474.7464

116 398022666077500000000

96

4.9452

474.7392

117 428018666071000000000

35

13.5640

474.7400

118 438019666074100000000

51

9.3086

474.7386

119 448000000079997400000

460

1.0320

474.7200

120 448071666079159900000

141

3.3669

474.7329

121 448101666071600000000

219

2.1678

474.7482

122 448227666079988200000

75

6.3299

474.7425

123 448254666079992000000

375

1.2660

474.7500

124 488000000072600000000

60

7.9123

474.7380

125 518000000070500000000

171

2.7763

474.7473

TOTAL

106,013.9400

Table 2. Estimated CVs for this sample design
Estimate
Estimated variance
Adult females

2005 total

CV

740,046.42

6,946

12.4%

1,101,074.74

92,578

1.1%

Black

539,793.00

32,961

2.2%

Female ADP

743,898.51

7,656

11.3%

Hispanic

463,736.49

25,754

2.6%

854.59

107

27.3%

Male ADP

807,551.42

98,238

0.9%

One day count

444,826.89

108,884

0.6%

Rated capacity

1,333,010.47

115,152

1.0%

517,917.45

32,115

2.2%

Adult males

Juvenile males

White

Page 1 of 2

Public Jails Sample Design for the 2009 Survey on Sexual Violence
2,867 public units on the 2008 Deaths in Custody file
Revision Date: March 10, 2010

The Bureau of Justice Statistics (BJS) requested a sample size of 700, with the largest public jail
in each state 1 selected with certainty to meet the requirements of the Prison Rape Elimination
Act of 2003. The measure of size is the average daily population (ADP).
We used the 2008 Deaths in Custody file as the frame.
We chose 128 units as certainty due to size (ADP of 1,000 or more). BJS requested that the
remaining 526 units be selected in a stratified systematic random sample. There are three
noncertainty strata for those units with less than 1,000 ADP.
We used the cumulative
method (Cochran, Sampling Techniques, 1977 edition, p. 129)
to determine the noncertainty stratum boundaries. The strata are shown in Table 1.
We used ADP to stratify the sample, with the allocation to the strata based on the number of
confined persons on December 31, 2008. An optimal allocation to the strata was calculated for
the number of confined persons.
The noncertainty strata were serpentine-sorted by region, two-digit state code, and ADP. Region
is the region of the country where the jurisdiction is located: Northeast, Midwest, South, or
West.
Table 1 shows the weights for this sample design.
Table 1. Public jails sample design table
Stratum
Stratum description
number
1
2
3
4
5

1

Largest jail in each state
Certainties due to size (≥ 1,000 ADP)
Jails with 0 to 85 ADP
Jails with 86 to 268 ADP
Jails with 269 to 999 ADP
TOTALS

Units in 2008
Deaths in
Custody file
46
128
1,489
770
434
2,867

Units
in
sample

46
128
99
317
110
700

Sample
weight
1.0000
1.0000
15.0404
2.4290
3.9455

There are public jails in 45 states and the District of Columbia. There are five states with no
public jails: Connecticut, Delaware, Hawaii, Rhode Island, and Vermont.

Page 2 of 2

This sample design produces the estimated coefficients of variation shown in Table 2 below.
Table 2. Estimated coefficients of variation for the public jails sample design
Estimate

Coefficient of variation

Confined males

1.9%

Confined females

2.8%

Newly admitted males

3.0%

Newly admitted females

3.4%

New admissions

2.9%

Male ADP

1.0%

Female ADP

1.9%

Page 1 of 2

Private Jails Sample Design for the 2009 Survey on Sexual Violence
41 private unites on the 2008 Deaths in Custody file
Revision Date: March 10, 2010

There are 41 private jails on the 2008 Deaths in Custody file. The Bureau of Justice Statistics
(BJS) requested a sample of 15 of the private units, with the units being selected with probability
proportional to size. The measure of size is the average daily population (ADP).
Two private jails were selected with certainty because they are so much larger than all the other
private jails. The remaining 13 units in sample were selected with probability proportional to
size after the file was serpentine-sorted by region, two-digit state code and ADP. Region is the
region of the country where the jurisdiction is located: Northeast, Midwest, South, or West.
The weights are shown in Table 1.
The 41 private jails on the frame have a total ADP of 20,322. We verify the sample weights by
using Horvitz-Thompson estimation. We use the sample to estimate the total ADP. The
estimated total is

where yi is the ADP of the ith unit in the sample.
Table 2 shows the estimated coefficients of variation for this sample design.

Page 2 of 2

Table 1. Private jails sample design table
ID

Region

Measure
of size

Sampling
Weight

Measure of size *
Sampling Weight

158049008061000000000

Midwest

1,130

1.1604

1,311.2520

268041041062000000000

Midwest

200

6.5565

1,311.3000

108006006065000000000

South

573

2.2885

1,311.3105

108027027061000000000

South

703

1.8653

1,311.3059

198013013062000000000

South

785

1.6705

1,311.3425

438019003068000000000

South

1,083

1.2108

1,311.2964

438084084062000000000

South

599

2.1892

1,311.3308

448082082061000000000

South

373

3.5156

1,311.3188

448104104062000000000

South

530

2.4742

1,311.3260

448147147062000000000

South

1,096

1.1964

1,311.2544

448233233061000000000

South

1,259

1.0415

1,311.2485

328030030061100000000

West

733

1.7890

1,311.3370

058019001061000000000

West

6

218.5513

1,311.3078

378072072064000000000

South

1,394

1.0000

1,394.0000

398023023063000000000

Northeast

1,881

1.0000

1,881.0000
20,321.9306

Table 2. Estimated coefficients of variation for this sample design
Estimate
Estimated variance
2008 total

CV

Confined females

149,456.51

1,317

29.4%

Confined males

348,020.82

18,018

3.3%

Confined persons

402,677.32

19,335

3.3%

Female ADP

149,386.20

1,531

25.2%

Male ADP

269,293.17

18,791

2.8%

258,077,093.93

24,948

64.4%

Newly admitted males

2,740,829,912.76

152,809

34.3%

New admissions

4,525,904,139.32

196,242

34.3%

Newly admitted females

Page 1 of 3

Tribal Sample Design for the 2009 Survey on Sexual Violence
63 units on the extract of the 2008 Jails in Indian Country file
Date: March 11, 2010
The Bureau of Justice Statistics (BJS) requested a sample of 15 units from the 63 units listed in
the extract of the 2008 Jails in Indian Country file. To be eligible for this sample, units hold
adults only or adults and juveniles. Units that hold only juveniles have been added to the
juvenile sample for the 2009 Survey on Sexual Violence (SSV).
The sample was selected through probability proportional to size, with the adjusted average daily
population (ADP) as the measure of size. The adjusted ADP was the maximum of (1, ADP).
Two units were relatively large compared to the rest of the units in the frame, so they were
selected as certainty units based on size. The size cutoff for certainty units was ADP of 100 or
more.
The rest of the file was serpentine-sorted by two-digit state code and ADP.
The 15 tribal facilities selected for the sample have weights based on their measure of size. The
weights are shown in Table 1.
The 63 tribal facilities on the frame have a total ADP of 1,714. We verify the sample weights by
using Horvitz-Thompson estimation. We use the sample to estimate the total ADP. The
estimated total is
where yi is the ADP of the ith unit in
the sample.
Table 2 shows the estimated coefficients of variation (CVs) for this sample design. We merged
the extract with Appendix Table 3 of the publication Jails in Indian Country, 2008 so that we
could calculate estimated CVs.

Page 2 of 3

Table 1. Tribal sample for the 2009 SSV

ID

Facility

Adjusted
Adjusted
measure of size
measure Sampling
* Sampling
of size
Weight
Weight

37004001070099990000

San Carlos DOC and RehabilitationAdult and Juvenile Detention

92

1.1672

107.38

37007001070099990000

Salt River Pima-Maricopa Department
of Corrections

52

2.0651

107.39

37015001070099990000

Colorado River Indian Tribes Adult
Detention Center

38

2.8259

107.38

37001001071000000000

Navajo Department of CorrectionsChinle

19

5.6518

107.38

67034001070100000000

Southern Ute Police Department and
Adult Detention Center

40

2.6846

107.38

277043001070100000000 Fort Peck Police Department and Adult
Detention Center

25

4.2954

107.39

297004002070100000000 Eastern Nevada Law Enforcement
Adult Detention Facility

20

5.3692

107.38

327033001070100000000 Acoma Tribal Police and Holding
Facility

34

3.1584

107.39

357003001070100000000 Fort Totten Law Enforcement and
Adult Detention Center

28

3.8352

107.39

427043001070100000000 Lower Brule Justice Center-Adult
Detention

22

4.8811

107.38

427061001070299900000 Rosebud Sioux Tribal PD and Adult
Detention

42

2.5568

107.39

487024001070100000000 Colville Adult Detention Center

34

3.1584

107.39

507040001070100000000 Menominee Tribal Detention Facility

52

2.0651

107.39

37011002071599990000

Gila River Department of
Rehabilitation and Supervision-Adult

176

1.0000

176.00

37010001071500000000

Tohono O'odham Adult Detention
Center

142

1.0000

142.00

TOTAL

1,714.00

Page 3 of 3

Table 2. Estimated coefficients of variation for this sample design
Estimate
Estimated variance

2008 total

CV

3,333.99

384

15.0%

41.29

14

45.9%

21,797.79

1,498

9.9%

Male juveniles

330.47

25

72.7%

Rated capacity

125,043.67

2,362

15.0%

35,672.35

1,882

10.0%

420.31

39

52.6%

Female adults
Female juveniles
Male adults

Total adults
Total juveniles

Page 1 of 5
Juvenile Facility Sample Design for the 2009 Survey on Sexual Violence
2,810 non-tribal units on the 2008 Juvenile Residential Facility Census file 1
plus 19 tribal juvenile facilities from the 2008 Jails in Indian Country file
Revision Date: March 22, 2010
For the 2009 Survey on Sexual Violence (SSV) juvenile facility sample, the Bureau of Justice
Statistics (BJS) requested a sample design similar to that used for the 2008 SSV juvenile facility
sample. Note that the tribal juvenile facilities are selected from the 2008 Jails in Indian Country
file, not the 2008 Juvenile Residential Facility Census (JRFC) file.
The 2008 SSV juvenile facility sample was a modification of that used in the 2005 SSV. To
understand this year’s sample design, we need to look at how the facilities are categorized. The
2008 JRFC serves as the frame for the 2009 SSV.
Table 1. 2009 SSV juvenile facility sampling frame
2,819 facilities in the 2008 JRFC
-9 tribal facilities in the 2008 JRFC
2,810 non-tribal facilities in the 2008 JRFC

BJS requested that all 473 state central reporters and facilities that report separately be included
in the sample with certainty. The rest of the sample comes from the remainder of the sampling
frame, to produce a sample of 330 non-state units.
This year, two facilities in the District of Columbia will be treated as a state central reporter or
facilities that report separately.
Of the 330 non-state units in sample, 36 units 2 are in with certainty as the largest locallyoperated facility in the state (as instructed by BJS), and 51 are in with certainty as the largest
privately-operated facility in the state (as instructed by BJS). That leaves 243 noncertainty
sample units to be selected.

1

There is a significant amount of turnover among juvenile facilities from one year to the next, which means that a
new juvenile facility sample should be drawn for this project every year. There were 2,911 facilities on the 2007
Census of Juveniles in Residential Placement file. There are 2,696 facilities on both the 2007 and 2008 files, 215
facilities that are on the 2007 file only, and 123 that are on the 2008 file only.

2

There were 37 states with locally-operated facilities in the 2008 SSV. Florida does not have locally-operated
facilities in the 2008 JRFC. Florida had one locally-operated facility in the 2007 CJRP, which was the frame for the
2008 SSV.

Page 2 of 5
Table 2. Certainty and noncertainty counts on the 2009 frame
473
36
643
51
1,607
2,810

state central reporters and facilities that report separately
local facility certainties (largest in the state)
local noncertainty facilities
private facility certainties (largest in the state)
private facilities
non-tribal facilities in the 2008 JRFC

BJS requested an oversample of non-state detention centers. By law, we need a 10-percent
sample of the non-state facilities. There are 2,337 such facilities in the 2008 JRFC, so a 10percent sample is 234 units.
The oversample of non-state detention centers is the extra 96 units available for the sample (330
non-state sample units – 234 non-state units based on a 10-percent sample = 96 “extra” sample
units for the oversample.)
Table 3. Counts of the non-state facilities on the 2009 frame

523 non-state detention centers
214 local noncertainty facilities (shelters, reception/diagnostic centers, training schools,
halfway houses/group homes, ranches, camps, or farms)
1,513 private noncertainty facilities (shelters, reception/diagnostic centers, training schools,
halfway houses/group homes, ranches, camps, or farms)
36 local facility certainties (largest in state)
51 private facility certainties (largest in state)
2,337 non-state non-tribal facilities in 2008 JRFC

Table 4. Distribution of non-state, noncertainty, nontribal facilities
523
37
177
202
1,311
2,250

non-state detention centers
local non-commitment facilities
local commitment facilities
private non-commitment facilities
private commitment facilities
noncertainty facilities

Table 5. Distribution of non-state units in 2009 sample
36
51
148
95

local facility certainties (largest in state)
private facility certainties (largest in state)
non-state detention centers (10-percent sample plus 96 oversampled units)
local or private noncertainty facilities (shelters, reception/diagnostic centers, training
schools, halfway houses/group homes, ranches, camps, or farms)
330 non-state units in 2009 sample

Page 3 of 5
The non-state detention facilities are stratified by region. BJS requested that the sample be
proportionally allocated by number of persons assigned to beds. Normally we would take a 10percent sample of the non-state detention facilities, or 52 units. The 96 extra units available for
the oversample mean we will select 148 non-state detention facilities for the 2009 SSV sample.
Table 6. Proportionally allocating sample to non-state detention facilities

Stratum number
40A
40B
40C
40D

Description
Detention facilities Midwest
Detention facilities Northeast
Detention facilities South
Detention facilities West
Totals

Persons assigned to beds
4,562
2,053
4,416
7,236
18,267

n

37
16
36
59
148

There are 35 non-state detention facilities that are too large compared to the rest of the facilities
in their strata, so they are declared certainties due to size and reassigned to stratum 40E in the
sample design table.
Table 7. Proportionally allocating sample to local and private noncertainty facilities
Stratum number
51
52
61
62

Description
Local
Non-commitment
Local
Commitment
Private Non-commitment
Private Commitment
Totals

Persons assigned to beds
358
7,464
2,738
28,219
38,779

n

2
18
7
68
95

There are one local facility and one private facility that are too large compared to the rest of the
facilities in their strata, so they are declared certainties due to size and reassigned to strata 53 and
strata 63 in the sample design table.
Once the state central reporters and facilities that report separately (stratum 10), largest locallyoperated facility in each stratum (stratum 20), detention facilities that are certainties based on
size (stratum 40E), local facilities that are certainties based on size (stratum 53), largest
privately-operated facility in each state (stratum 30), and private facilities that are certainties
based on size (stratum 63) were determined, those records were removed from the 2008 JRFC
file. The remaining 2,213 facilities were serpentine-sorted by region, two-digit state code,
collapsed facility type, and persons assigned to beds within each stratum. We used PROC
SURVEYSELECT in SAS to select a systematic probability proportional to size sample.

Page 4 of 5
Table 8.
Stratum
Number
10
15
20
30
40A
40B
40C
40D
40E
51
52
53
61
62
63

Juvenile facilities sample design table
Description
State central reporters and facilities that report separately
Tribal juvenile facilities from 2008 Jails in Indian Country file
Largest locally-operated facility in each state
Largest privately-operated facility in each state
Detention facilities Midwest
Detention facilities Northeast
Detention facilities South
Detention facilities West
Detention facility certainties due to size
Local
Non-commitment
Local
Commitment
Local certainty due to size
Private
Non-commitment
Private
Commitment
Private certainty due to size
Totals

N

473
19
36
51
156
87
144
101
35
37
176
1
202
1310
1
2829

n
473
19
36
51
33
14
33
33
35
2
17
1
7
67
1
822

Calculating coefficients of variation for the sample 3
We use the Hartley – Rao formula to estimate the variance of this sample design. The variance
is given by

where πi is the probability that the ith unit is selected for the sample. For more details, see
equation 5.17 in Hartley and Rao (1962).
Table 9. Estimated coefficients of variation for this sample design
Level of estimate
Estimate
Estimated variance 2008 total CV
National
Juvenile offenders
1,599,524.96
78,973 1.6%
National
Juvenile non-offenders
1,583,501.41
15,575 8.1%
Non-state detention facilities Juvenile offenders
23,679.29
22,107 0.7%
Non-state detention facilities Juvenile non-offenders
11,585.81
385 28.0%

3

The calculations for the coefficients of variation do not include the 19 tribal facilities from the 2008 Jails in Indian
Country (JIC) file because comparable data on juvenile offenders and juvenile non-offenders were not available on
the JIC file.

Page 5 of 5
Verification of the sample file produced by the Statistical Methods Branch
The Statistical Methods Branch of Governments Division selected the locally-operated and
privately-operated facility records for the sample. The Criminal Justice Branch of Governments
Division prepared the mailout records for the state-operated units in stratum 10, and the 20 tribal
juvenile facilities from the 2008 Jails in Indian Country (JIC) file in stratum 15.
The file produced by the Statistical Methods Branch has 330 records. We can verify the sample
selection by estimating the total number of persons assigned to beds for locally-operated and
privately-operated facilities. There are 67,423 persons assigned to beds in locally-operated and
privately-operated facilities in the 2008 JRFC file that are in-scope for the 2009 SSV.
We use the Horvitz-Thompson estimator to estimate the total persons assigned to beds.

where yi is the total number of persons assigned to beds for the ith facility, and weighti is the
weight assigned to the ith facility in the file produced by the Statistical Methods Branch. The
results are shown in Table 10.
Table 10. Estimated total number of persons assigned to beds in the locally-operated and
privately-operated facilities in the juvenile facility sample
Stratum number
Facilities in sample
Estimated total
20
36
4,031.00
30
51
6,346.00
40A
33
4,025.00
40B
14
1,728.00
40C
33
3,907.99
40D
33
2,476.00
40E
35
6,130.00
51
2
358.00
52
17
7,032.00
53
1
432.00
61
7
2,738.00
62
67
27,755.98
63
1
463.00
330
67,422.97
References
“Sampling with Unequal Probabilities and without Replacement”
Hartley, H.O. and Rao, J.N.K.
The Annals of Mathematical Statistics, Vol. 33, No. 2. (Jun., 1962), pp. 350 – 374.