Appendix K EPICCS Center Sample Selection Memo #1

Appendix K EPICCS Center Sample Selection Memo#1.pdf

Erroneous Payments in Child Care Centers Study (EPICCS)

Appendix K EPICCS Center Sample Selection Memo #1

OMB: 0584-0618

Document [pdf]

Download: pdf | pdf

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Date:

February 22, 2016

To:

Chanchalat Chanhatasilpa

CC:

Devin Wallace-Williams, Lisa Southworth

From:

Roline Milfort, Ron Klinger, Adam Chu, Mamadou Diallo

Subject:

Selection of Centers for the Erroneous Payments in Child Care Centers Study
(EPICCS) - Contract # AG-3198-C-14-0015

Introduction

In the initial stage of the sample selection, 25 States were selected as described in the memorandum
sent to Chanchalat Chanhatasilpa on April 29, 2015. Sampling frames (universe lists) of eligible
Child and Adult Care Food Program (CACFP) child care centers were then constructed for each of
the 25 selected States. This memo provides an overview of the procedures used to create the center
and primary sampling unit (PSU) frames, and subsequent selection of PSUs and centers from the
universe lists submitted by the States. This memorandum serves to provide a summary of the
procedures. A second follow-up memo will be submitted on March 14, 2016 with greater details.
Section 2 provides a summary of the creation of the sample frame for centers and PSUs. One of the
first tasks was to identify and exclude out of scope centers. Second, Average Daily Attendance
(ADA) was verified or imputed as this is a key variable for both PSU and center sampling. Third, the
PSUs were created based on Core Based Statistical Areas (CBSAs) and non-CBSA counties. In some
cases, it was necessary to combine CBSAs and non-CBSA counties to form PSUs with a minimum
number of eligible centers. Section 3 provides a summary of PSU selection, including the assignment
of composite measure of size (MOS). Section 4 provides a summary of center selection. Appendix
B presents a summary of the sampled PSUs presenting the geographical area covered and the
number of sampled centers for each of the 50 PSUs.

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Center and PSU Frames Creation

The first step in the selection of child care centers (CCCs) was to create the frame of CCCs for each
of the 25 States in the EPICCS sample. The necessary information required to create the frame of
CCCs was obtained from designated State Agencies. Only one State, Tennessee, was not able to
provide a list of centers; instead Tennessee provided a list of sponsoring organizations. In the case
of Tennessee, an additional sample selection step was used to accommodate the non-availability of a
list of CCCs (see Section 4 for more details).
The State agencies submitted data files with a complete list of their CCCs participating in CACFP as
reported to FNS on Form 44 in March 2015. The State agencies were also asked to provide a list of
all their sponsoring organizations and the number of meals claimed in the month of March 2015.
We provided the State agencies with guidelines for the compilation and submission of their data
files, including data file requirements and instructions for uploading the file to the EPICCS’ secure
website. Upon submission of their data files, the Kokopelli team and Westat conducted reviews of
the data, and followed up with the State as needed. During the course of multiple tiered reviews,
data edit checks, additional follow up with States, data reviews, and other quality assurance tasks
were conducted. Updates and/or corrections to the data files were applied as needed and
documented.

2.1

Exclusion of Out-of-Scope Centers

The CCCs on the lists provided by the States were classified into three types for sampling purposes:
(1) independent or self-sponsored (ICCC), (2) sponsored (SCCC), and (3) head start (HSC). Certain
types of centers such as emergency shelters, adult centers, outside school hours (OSH) centers, and
At-Risk (AR) centers are out-of-scope of EPICCS and were removed from the frame. “Mixed
centers,” defined as centers with both in-scope and out-of-scope components, were kept in the
frame and given a chance of selection for the study sample. In those situations where the in-scope
and out-of-scope components of a center were listed as separate entities in the frame, the out-ofscope component was excluded.

2.2

Imputation of Average Daily Attendance (ADA)

The average daily attendance (ADA) of enrolled children was requested for each center listed in the
statewide sampling frames. ADA was used to derive the PSU and Center measure of sizes (MOSs)

Page 2 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

for sample selection; hence it was necessary to get a strictly positive value of ADA for all CCCs in
the target population. Some CCCs did not have a valid ADA because they did not submit claims for
March 2015. These CCCs were kept in the frame because there is a chance that they will be active
during the time period for analysis, July 2016 to June 2017. For the purpose of sample selection,
these CCCs were assigned (imputed) an ADA equal to the average ADA of those centers with
nonmissing ADA under the same sponsoring organization. If the average value was not valid (i.e.,
the ADA for all centers from the same sponsoring organization was zero or missing) then the
average ADA in the State for the given type-of-center was assigned to the center. The use of the
statewide average to impute ADA was most commonly used for ICCCs which by definition do not
have a sponsoring organization (i.e. self-sponsored).
Four States (California, Florida, New Jersey, and Pennsylvania) could not provide the requested
ADA information and instead submitted licensed capacity as a proxy for ADA. Virginia did not
provide ADA nor licensed capacity. The Westat team was able to obtain from the Virginia
Department of Social Services website the licensed capacity for about two thirds of the CCCs in the
target population (462 out of 721 centers). The CCCs with zeros or missing values of licensed
capacity in Virginia and in the other four States were assigned an average licensed capacity following
the rules described in the previous paragraph. Table 2.1 summarizes the number of centers for
which either ADA or licensed capacity was imputed for the 16 States where imputation was
required.
Table 2.1 Number of centers for which ADA or licensed capacity was imputed, by State
State

Type of Data

CO
CT
FL
GA
IL
KS
LA
MO
NC
NY
OH
PA
SC
TX
VA
WI

ADA
ADA
LICENSED CAPACITY
ADA
ADA
ADA
ADA
ADA
ADA
ADA
ADA
LICENSED CAPACITY
ADA
ADA
LICENSED CAPACITY
ADA

Total Number of
Eligible Centers
502
250
3,240
1,306
1,685
503
604
976
1,994
2,061
1,522
2,886
413
4,167
721
805

Number of Centers with
Zero or Missing Value
4
2
1
3
56
1
5
5
1
68
1
858
6
1
259
2

Page 3 of 15

Percent of Centers with
Zero or Missing Value
0.80%
0.80%
0.03%
0.23%
3.32%
0.20%
0.83%
0.51%
0.05%
3.30%
0.07%
29.73%
1.45%
0.02%
35.92%
0.25%

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Virginia, with 35.9%, had the highest proportion of centers with zero or a missing value for ADA or
licensed capacity followed by Pennsylvania with 29.7%. For Illinois and New York, about 3.3% of
the CCCs had a zero or a missing value for ADA or licensed capacity. The remaining States in Table
2.1 had only trivial numbers of CCCs with a non-positive value for ADA or licensed capacity.
Licensed capacity and ADA are not the same and were expected to be different, although they are
correlated. Initial evaluations indicated that the reported licensed capacities for two large States,
Florida and Pennsylvania were considerably different from the expected total ADAs. In these two
States, the reported licensed capacity was much higher than the corresponding total ADA based on
the 2015 FNS NDB Data table shown in Appendix A. If these unadjusted licensed capacity values
were to be used to derive the measure of size for sampling, they would lead to disproportionately
large samples of PSUs and centers from these States.
Hence, we adjusted the licensed capacities for these States to reflect the level of ADA we would
expect from the 2015 FNS table in Appendix A. Ratio adjustments were applied to the licensed
capacities so that, for each of the five relevant States (California, Florida, New Jersey, Pennsylvania,
and Virginia), the licensed capacities sum to the expected total ADA of the in-scope CCCs. The
expected total ADA for the in-scope CCCs used in this adjustment was obtained by multiplying the
total ADA for all child care centers in the State from the NDB data (which included in-scope and
out-of-scope) by the estimated proportion of At-Risk centers shown in the last column of the table
in Appendix A. We also adjusted the ADAs for New York because the data provided by New York
did not separate out the At-Risk centers. We, therefore, adjusted the ADA values for New York to
account for the inclusion of At-Risk centers in their reported ADA data.

2.3

Creation of Primary Sampling Units (PSUs)

In total, we identified 31,226 eligible CCCs across the 25 selected States. Among them, 11,945 (38%)
were classified as independent child care centers (ICCCs); 11,245 (36%) were classified as sponsored
child care centers (SCCC); and 8,036 (26%) were classified as head start centers (HSC). When a
center was identified as mixed (more than one type) from the information received from the States,
the following priority order was used to assign the center to the appropriate stratum for sampling:
(1) HSC followed by (2) ICCC and finally (3) SCCC. This rule was implemented to ensure that each
center is classified in one category and one only, and in such a way that reduces extraneous variation
in weights due to misclassification.

Page 4 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

All 31,226 CCCs in the final frame were geocoded (i.e., assigned to OMB-designated metropolitan
areas or counties) using the address of the center provided by the State. In some instances, the
Westat team conducted online investigations to enhance or to determine the center addresses that
were missing or incomplete. Geocoding allowed us to place every CCC into the appropriate Core
Based Statistical Area (CBSA) or a non-CBSA county. CBSAs are metropolitan areas consisting of a
group of counties around at least one urban center with a population of 10,000 or more
(http://www.census.gov/geo/reference/gtc/gtc_cbsa.html).
The 31,226 centers in the EPICCS State sampling frames were initially mapped into a total of 1,287
CBSAs and non-CBSA counties. Among them about half had three CCCs or less. We combined the
CBSAs and the non-CBSAs counties with a small number of CCCs to ensure a minimum number of
CCCs per PSU. Selecting remote or isolated geographical areas with only few centers can have a
significant impact on the data collection costs. These geographic areas were combined with other
areas to maintain them in the sample frame. The basic rule used for combining the geographical
areas was to make sure that all combined areas had a minimum of 6 CCCs. We targeted a minimum
of 9 CCCs per combined area but went as low as 6 to accommodate special cases such as States with
a small number of counties or to avoid forming areas with too many counties. Two other
requirements for the combined areas were (1) to form a contiguous geographical area and (2) to
keep the combined area as small as possible. The latter requirement was implemented manually by
visual inspection of maps. CBSAs crossing State boundaries were split as necessary to define PSUs
that were fully contained within States.

2.4

Illustration of Method Used to Combine CBSAs and Counties

To combine CBSAs with non-CBSA counties where necessary, we used a Westat application called
WebSumars. The application plots the counties appearing in the EPICCS sampling frame for a given
State with their boundaries and associated number of centers as shown in Figure 2.1 below. For
example, the county with the code 53047 has 13 centers in the EPICCS frame. Using this
application, we visually identified CBSAs or non-CBSA counties with a small number of centers and
combined them with neighboring counties. The coloring scheme shows the final combinations
performed for the counties in Washington State. This process was repeated for all States.

Page 5 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Figure 2.1 Counties from Washington State identified in the EPICCS sample.

2.5

The Final PSU Sampling Frame

The final combined geographical areas are the primary sampling units (PSUs) for the EPICCS.
These PSUs can be a single CBSA, a set of CBSAs, a combination of a CBSA and non-CBSA
counties, a set of non-CBSA counties, or a single non-CBSA county. Each PSU had at least 6 CCCs
in the PSU based on geocoding. The final EPICCS PSU sampling frame included 593 PSUs with an
average of about 3.1 counties per PSU. Note that most of the PSUs were individual or parts of
CBSAs. There were a total of 31,226 centers in the frame; hence on average there were about 52.7
centers per PSU. However the number of centers varied widely, ranging from 6 in the smallest PSUs
to 1,297 in the New York-Northern New Jersey-Long Island, NY-NJ-PA CBSA (New York part).
On average, there were 20.1 ICCCs per PSU, 19.0 SCCCs per PSU and 13.6 HSCs per PSU.

Page 6 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

PSU Selection

From the frame of PSUs created as described in the previous section, 50 PSUs were selected with
probability proportional to size (PPS). To reflect the fact that the PSUs are usually composed of the
three types of centers, a composite measure of size (MOS) was computed and used for sample
selection.

3.1

Assignment of Composite Measure of Size
(𝑖)

Let 𝑀𝑂𝑆𝑠ℎ = the total ADA (after imputation and adjustment) for the PSU ℎ in State 𝑠 for the
center type 𝑖. The superscript 𝑖 can take the values 1 for independent, 2 for sponsored, and 3 for
head start. We can estimate the total (adjusted) ADA for the type 𝑖 centers within the 48 contiguous
States and the District of Columbia by
̂ (𝑖)

𝑀𝑂𝑆

(𝑖)

𝑀𝑂𝑆𝑠ℎ
=∑ ∑
,
𝑃𝑠
𝑠=1
ℎ=1
25

𝑚𝑠

where 𝑚𝑠 is the number of PSUs from State 𝑠 and 𝑃𝑠 is the probability of selecting State 𝑠. The
composite PSU measure of size is given by
𝑃𝑆𝑈
𝑀𝑂𝑆𝑠ℎ
=

3
1
(𝑖)
∑ 𝑓𝑖 𝑀𝑂𝑆𝑠ℎ
𝑃𝑠
𝑖=1

where the 𝑓𝑖 ′ s are proportional to the desired overall sampling rate to be used to select applications
̂ (𝑖) , where 𝑛(𝑖) = the number of applications to be
within the type 𝑖 of centers; i.e., 𝑓𝑖 = 𝑛(𝑖) / 𝑀𝑂𝑆
selected from the type i of centers. In this study, 𝑛(𝑖) was set to be 3,0001. The number 3,000 was
obtained by assuming an average of 10 applications for each of the 300 centers to be included in the
initial sample of centers. For each type of center, the target number of final participating CCCs is
150. Hence, the extra CCCs selected for each type will serve as a reserve sample to compensate for
sample losses due to non-response and ineligibility; more details about the allocation of the initial
sample of centers to primary and reserve samples are provided in Section 4.

Note that because 𝑛(𝑖) = 3,000 is a constant across the PSUs, it will not impact the relative importance of the PSUs
and therefore will not affect the probability of selecting a PSU. Any other constant will result in the same PSU
probabilities of selection.

Page 7 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

The use of the above composite MOS for Maryland and Virginia would have led to expected sample
sizes below 1 for both States. This occurs because the center-level ADA data reported by these
States (which is presumed to be accurate) was considerably less that the corresponding estimates
used to select the State sample. Consequently, there would be a non-zero probability that no PSUs
would selected from one of these States if the MOS as computed above is used. To ensure that both
Maryland and Virginia are allocated at least one sample PSU, we multiplied the reported ADAs for
these States by factors of 2.4 and 1.4, respectively. The ADA for Arkansas was also adjusted slightly,
by a factor of 1.05, to ensure that at least one PSU would be sampled.
Table 3.1 compares the expected sample sizes as presented in the EPICCS Study Plan (see Table 2-1
from EPICCS 3rd Revised Study Plan) with those based on the composite MOS described above.
The two sets of expected sample sizes are generally close. Note that we do not expect the two sets
of numbers to be the same because we used updated information from the States to compute the
new expected sample sizes, and the study plan measure of size included the At-Risk ADA
components. The actual numbers of sampled PSUs (see Section 3.2 below) are shown in the last
column of Table 3.1.
Table 3.1 Expected PSU sample sizes
STATE

CERTAINTY

STUDY PLAN EXPECTED SAMPLE SIZE

EXPECTED SAMPLE SIZE
BASED ON COMPOSITE MOS

NUMBER OF
SELECTED PSUS

YES

6.46

5.37

YES

3.93

3.90

YES

2.11

2.17

YES

1.80

2.34

YES

3.82

2.71

YES

1.84

2.33

YES

1.68

2.08

YES

2.05

1.69

YES

5.79

5.62

1.28

1.69

1.28

1.02

1.28

1.94

1.28

1.57

1.28

1.47

Page 8 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

STATE

CERTAINTY

STUDY PLAN EXPECTED SAMPLE SIZE

EXPECTED SAMPLE SIZE
BASED ON COMPOSITE MOS

NUMBER OF
SELECTED PSUS

1.28

1.54

1.28

1.01

1.28

1.04

1.28

1.41

1.28

1.33

1.28

1.36

1.28

1.62

1.28

1.15

1.28

1.13

1.28

1.42

1.28

1.31

3.2

Selection of PSUs

The selection process consisted of stratifying the States in two groups, the certainty and the noncertainty States (refer to the April 29, 2015 State Selection memorandum). From the certainty States,
28 PSUs were selected with probabilities proportionate to the composite MOS, and similarly 22
PSUs were selected with probabilities proportionate the composite MOS from the non-certainty
States. Within the certainty States, four PSUs (namely Los Angeles CBSA, Manhattan CBSA,
Chicago CBSA, and Houston CBSA), were so large in terms of the composite measure of size that
each in effect were the equivalent of multiple sample PSUs. To ensure that 50 distinct PSUs were
selected, we selected these four large PSUs as certainties, and then selected 24 PSUs from the
remaining list of PSUs in the certainty States. The last column of Table 3.1 provides the number of
selected PSUs per State.
Note that the number of PSUs selected in the certainty States is somewhat different to the expected
sample sizes especially for California and Florida. This was due to the fact that the four largest PSUs
had a much higher than 1 expected chance of being in the sample and were selected once. For
example, Los Angeles CBSA had an expected number of selections equal to 2.0 and is therefore
equivalent to 2 PSUs. The Manhattan CBSA had an expected number of selections equal to 1.8, and
this is also equivalent to almost 2 PSUs. Taking this into account the sizes of the PSUs, California
and New York have roughly an equivalent of 5 and 2, respectively. Appendix B provides the final

Page 9 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

list of the selected PSUs with the CBSAs and/or the non-CBSA counties belonging to each PSU as
well as the total number of centers found eligible per PSU.

Center Selection

The centers across the 50 sampled PSUs were organized into three sub-frames, one for each type of
center (ICCC, SCCC, and HSC). For each type, we selected a total of 300 center-equivalents. The
number of distinct centers per type was less than 300 because some centers were selected multiple
times. The centers were selected with probabilities proportionate to size (PPS) where the measure of
size was equal to
𝑀𝑂𝑆𝑠ℎ𝑗 = 𝐴𝐷𝐴𝑠ℎ𝑗 ⁄𝑃𝑠 𝑃𝑠ℎ
and where 𝐴𝐷𝐴𝑠ℎ𝑗 is the final adjusted ADA as described in Section 2 assigned to the center 𝑗 in
PSU ℎ in State 𝑠, 𝑃𝑠 = the probability of selecting State 𝑠 and 𝑃𝑠ℎ = the (conditional) probability of
selecting PSU ℎ in State 𝑠.
In total, 825 distinct centers were selected across 24 States excluding Tennessee. In Tennessee, we
selected 10 sponsoring organizations, among which 6 were self-sponsored. Table 4.1 provides the
counts by center type and certainty versus non-certainty PSUs. The four non ICCC sponsoring
organizations selected in Tennessee will be contacted and asked to provide the list of their
sponsored CCCs. The final sample of sponsored CCCs for Tennessee will then be obtained in a
subsequent sampling stage using the lists from the sponsoring organizations as the Tennessee center
frame.
Table 4.1 Number of centers by type and certainty status of PSU
Center Type
ICCC

Certainty PSU
23

Non-certainty PSU
256

All centers
279

SCCC
HSC

25
35

251
235

276
270

All Types

742

825

The goal of the study is to have 150 respondent centers for each type. To achieve that goal, 150
centers of each type will be fielded initially. The remaining sample will be treated as a reserve to
Page 10 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

compensate for sample losses in the initial sample. Note that the reserve sample is intended to
augment the initial sample in order to achieve the desired target sample sizes. The reserve sample
will be released in phases upon regular evaluations of the shortage of centers necessary to achieve
the desired 150 participating centers by type. Figure 4.1 shows the geographic distribution of the full
sample of 835 centers (or sponsoring organizations for Tennessee) across the country.
Figure 4.1 Distribution of the full sample of 835 centers across the country

We assessed how well the (weighted) distribution of the center sample matched the corresponding
distribution of the centers in the frame with respect to two available sponsor-level characteristics. As
shown in Table 4.2 below, the results for the sample closely match that of the centers in the frame.
Table 4.2 Comparison of the sample versus the frame for two sponsor characteristics
Source
Sample
(Weighted Estimates)
Frame
(25 selected States)

Percentage of Out-of-State Sponsors

Percentage of Centers in CBSAs

3.07 percent

92.57 percent

3.21 percent

93.08 percent

Page 11 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Appendix A
NATIONAL DATA BANK VERSION 8.2 – ANALYSIS
FY 2015
Substate/Region

CN
Outlets
All
Child
Care
Centers

CN Avg Daily
Attendance
Child Care
Centers

CN
Outlets
After Sch
At-Risk

% of At Risk
Centers

0191501 AL STATE DEPT OF EDUCATION

927

45,007

188

20.28%

0291501 AK DEPARTMENT OF EDUCATION & EARLY DEVELOPMENT

183

8,543

48.63%

0491501 AZ DEPT OF EDUCATION

638

23,628

152

23.82%

0592901 AR DEPT OF HUMAN SERVICES

1,239

62,398

434

35.03%

0691501 CA DEPARTMENT OF EDUCATION

6,559

380,167

2,772

42.26%

0891701 CO DEPT OF HEALTH

597

26,858

292

48.91%

0991501 CT DEPARTMENT OF EDUCATION

349

16,360

16.62%

1091501 DE DEPT. OF EDUCATION

330

15,413

128

38.79%

4,804

31.25%

1291701 FL DEPT OF HEALTH

4,192

218,025

792

18.89%

1391502 GA DEPT OF EARLY CARE AND LEARNING

2,021

101,662

455

22.51%

818

1591501 HI STATE DEPT OF EDUCATION

247

10,528

2.83%

1691501 ID DEPT OF EDUCATION

158

6,541

26.58%

1191501 DC OFFICE OF THE STATE SUPERINTENDENT OF EDUCATION

1491501 GU DEPARTMENT OF EDUCATION

1791501 IL STATE BOARD OF EDUCATION

0.00%

2,337

95,398

737

31.54%

1891501 IN DEPT OF EDUCATION

820

53,901

261

31.83%

1991501 IA DEPARTMENT OF EDUCATION

630

28,676

5.87%

2091501 KS STATE BOARD OF EDUCATION

541

22,238

153

28.28%

2191501 KY DEPT OF EDUCATION

1,064

52,531

212

19.92%

2291501 LA DEPT OF EDUCATION

994

57,472

371

37.32%

2392901 ME DEPT OF HEALTH AND HUMAN SERVICES

164

4,247

14.02%

2491501 MD DEPT OF EDUCATION

656

34,868

225

34.30%

2591501 MA DEPT OF EDUCATION

638

28,201

189

29.62%

2691501 MI DEPT OF EDUCATION

913

43,172

6.02%

2791501 MN DEPT OF EDUCATION

441

26,976

126

28.57%

2891501 MS STATE DEPT OF EDUCATION

727

41,965

0.69%

2991701 MO DEPT OF HEALTH

958

40,598

301

31.42%

3092101 MT DEPT OF PUBLIC HEALTH AND HUMAN SERVICES

237

8,616

9.28%

3191501 NE DEPARTMENT OF EDUCATION

466

21,259

19.74%

3291101 NV DEPARTMENT OF AGRICULTURE

339

14,360

8.55%

3391501 NH DEPT OF EDUCATION

180

7,394

20.00%

1,209

65,219

188

15.55%

520

22,907

12.69%

3691701 NY DEPT OF HEALTH

3,768

210,394

1,726

45.81%

3792101 NC DEPT. OF HEALTH

10.57%

3491101 NJ DEPT OF AGRICULTURE
3593301 NM CHILDREN YOUTH & FAMILIES

1,817

90,733

192

3891501 ND DEPT OF PUBLIC INSTRUCTION

141

6,164

3.55%

3991501 OH DEPARTMENT OF EDUCATION

794

42,702

439

55.29%

4091501 OK DEPT OF EDUCATION

852

36,081

138

16.20%

4191501 OR DEPARTMENT OF EDUCATION

608

29,620

367

60.36%

Page 12 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Substate/Region

CN
Outlets
All
Child
Care
Centers

CN Avg Daily
Attendance
Child Care
Centers

CN
Outlets
After Sch
At-Risk

% of At Risk
Centers

4291501 PA DEPARTMENT OF EDUCATION

2,128

82,987

808

37.97%

4391501 PR DEPT OF EDUCATION

1,277

27,150

0.16%

4491501 RI DEPT OF ELEM & SEC ED

179

6,924

20.67%

4592501 SC DEPT OF SOCIAL SERVICES

457

25,130

108

23.63%

4691501 SD DEPT OF EDUCATION

243

9,962

9.47%

4792901 TN DEPT OF HUMAN SERVICES

1,462

48,202

751

51.37%

4891101 TX DEPT. OF AGRICULTURE

5,608

365,276

2,125

37.89%

4991501 UT STATE OFC OF EDUCATION

244

15,565

21.31%

5091501 VT DEPT OF EDUCATION

192

6,860

29.69%

5191701 VA DEPT. OF HEALTH

904

34,725

482

53.32%

1,384

8.33%

1,072

42,291

274

25.56%

5491501 WV DEPT. OF Education

651

19,810

301

46.24%

5591501 WI DEPT OF PUBLIC INSTRUCTION

695

35,819

128

18.42%

5691501 WY DEPT OF EDUCATION

149

6,166

0.67%

5291501 VI DEPARTMENT OF EDUCATION
5391501 WA OFFICE OF SUPERINTENDENT OF PUBLIC INSTRUCTION

Page 13 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

Appendix B
Primary Sampling Units (PSUs) for EPICCS
PSU

STATE

GEOGRAPHICAL AREA

TYPE

Birmingham-Hoover, AL

CBSA Only

134

Montgomery, AL

CBSA Only

Little Rock-North Little Rock-Conway, AR

CBSA Only

282

Los Angeles-Long Beach-Santa Ana, CA

CBSA Only

972

Riverside-San Bernardino-Ontario, CA

CBSA Only

173

Sacramento--Arden-Arcade--Roseville, CA

CBSA Only

170

San Francisco-Oakland-Fremont, CA

CBSA Only

321

Denver-Aurora-Broomfield, CO

CBSA Only

177

Greeley, CO

CBSA Only

Hartford-West Hartford-East Hartford, CT

CBSA Only

Crestview-Fort Walton Beach-Destin, FL
Walton Co., FL

CBSA and non-CBSA

Miami-Fort Lauderdale-Pompano Beach, FL

CBSA Only

865

Naples-Marco Island, FL

CBSA Only

280

Palm Bay-Melbourne-Titusville, FL

CBSA Only

Tampa-St. Petersburg-Clearwater, FL

CBSA Only

420

Atlanta-Sandy Springs-Marietta, GA

Carroll

585

CBSA and non-CBSA

Dalton, GA
Fannin Co. and Gilmer Co., ID (non-CBSA)
Boise City-Nampa, ID
Ontario, OR-ID (ID part)
Washington Co. (non-CBSA)
Pocatello, ID
Caribou Co. and Oneida Co., ID (non-CBSA)
Chicago-Joliet-Naperville, IL-IN-WI

CBSA

Cape Girardeau-Jackson, MO-IL (IL part)
Paducah, KY-IL (IL part)
Johnson Co. Pulaski Co., and Union Co. (non-CBSA)
McPherson, KS
Massac Co., Ellsworth Co., Marion Co., and Rice Co.
(non-CBSA)

Number of Centers

1,084

CBSA and non-CBSA

Baton Rouge, LA

CBSA Only

109

Baltimore-Towson, MD

CBSA Only

128

CBSA and non-CBSA

St. Louis, MO-IL (MO part)

CBSA Only

317

Dunn, NC

CBSA Only

Duluth, MN-WI
Cook Co., MN (non-CBSA)
Duluth, MN-WI (MN part)
Owatonna, MN
Waseca Co., MN (non-CBSA)

Page 14 of 15

APPENDIX K: EPICCS CENTER SAMPLE SELECTION MEMO#1

PSU

STATE

GEOGRAPHICAL AREA

TYPE

Lumberton, NC

CBSA Only

Winston-Salem, NC

CBSA Only

587

CBSA Only

1,297

New York-Northern New Jersey-Long Island, NY-NJ-PA
(NJ part)
New York-Northern New Jersey-Long Island, NY-NJ-PA
(NY part)
Watertown-Fort Drum, NY
Lewis Co., NY (non-CBSA)

Number of Centers

CBSA and non-CBSA

Cleveland-Elyria-Mentor, OH

CBSA Only

277

Springfield, OH

CBSA Only

Philadelphia-Camden-Wilmington, PA-NJ-DE-MD
(PA part)

CBSA Only

1,167

Scranton--Wilkes-Barre, PA

CBSA Only

154

Augusta-Richmond County, GA-SC (SC part)
Greenwood, SC
McCormick Co., SC (non-CBSA)

CBSA and non-CBSA

Hilton Head Island-Beaufort, SC

CBSA Only

Greeneville, TN
Johnson City, TN
Johnson Co., TN (non-CBSA)

CBSA and non-CBSA

Brownsville-Harlingen, TX

CBSA Only

185

Dallas-Fort Worth-Arlington, TX

CBSA Only

766

Houston-Sugar Land-Baytown, TX

CBSA Only

888

Killeen-Temple-Fort Hood, TX

CBSA Only

115

Midland, TX
Odessa, TX
Pecos, TX
Pecos Co. TX

CBSA and non-CBSA

Waco, TX

CBSA Only

Richmond, VA

CBSA Only

126

Kennewick-Pasco-Richland, WA

CBSA Only

Spokane, WA

CBSA Only

101

Milwaukee-Waukesha-West Allis, WI

CBSA Only

216

TN only provided list of sponsors. The 19 cases in the TN PSU are composed of 14 self-sponsored centers (ICCCs)
and 5 sponsors.

Page 15 of 15

File Type	application/pdf
File Title	Date:
Author	wilson_e2
File Modified	2016-05-11
File Created	2016-02-22