2021 NSCH Sampling Frame and Sampling Flags Creation

Appendix B

National Survey of Children’s Health
Sample Frame and Sampling Flags Creation

This document was approved by the Disclosure Review Board on Jan 14, 2021 (DRB Approval
No. CBDRB-FY21-POP001-0052).

2020 National Survey of Children’s Health sample
frame
Garret Christensen and John Voorheis
Center for Economic Studies
US Census Bureau
April 2, 2020
This document describes using administrative records to build a sample frame for the
National Survey of Children’s Health (NSCH) for 2020.

Click Here to Enter DRB Delegated Authority Approval Number

Population of interest
The population of interest is all children residing in housing units in the US on the date of the
survey.

A sample frame for all households with children
The sample frame identifies three mutually exclusive strata:
• [1] Households with explicit links to children in administrative data.
• [2a] Households without explicit links to children in administrative data, but predicted to be
likely to have children conditional on administrative data.
• [2b] Households without explicit links to children in administrative data, but predicted to be
unlikely to have children conditional on administrative data.
This document first explains the construction of the Stratum 1 flag, and then documents the
separation of Strata 2a and 2b.

Stratum 1: identifying explicit links from children to addresses
The Stratum 1 flag for all households with explicit links to children comes from three data
sources: (1) the Numident, (2) a list of Social Security Number applicants with data updated
from various administrative records, and (3) the Census Household Composition Key (CHCK,
formerly called CARRA kidlink) file, a prototype linkage between children and parents based on
Census and administrative records. Household addresses are updated with the Master Address
Auxiliary Reference File, a file that links person identifiers with the latest location updates from
a variety of administrative data.
Using the Numident to identify children
The Numident is based on all individuals who have been assigned Social Security Numbers.
Demographic data from the Numident is updated from federal tax data and various
administrative records. There are 83,650,000 children in the 2019 Numident who will be aged 0–
17 years on June 1, 2020. Figure 1 shows the distribution of date of birth for these children.
Figure 1: Distribution of date of birth, aged 0–17 years as of June 1, 2020 (2019 Numident)

Identifying the households containing the children in the Numident
To sample households with children, we must connect the children in the Numident to the
households in which they live. We do this with the CHCK file.

Census Household Composition Key File
The CHCK uses data from Census surveys and federal administrative records to link children
PIKs to parent PIKs. We can use this file to identify the parents of children in the Numident.

The source data for the CHCK are: the Census Numident, the 2010 Census Unedited File, the
IRS 1040 and 1099 files, the Medicare Enrollment Database (MEDB), Indian HealthService
database (IHS), Selective Service System (SSS), and Public and Indian Housing (PIC) and
Tenant Rental Assistance Certification System (TRACS) data from the Department of Housing
and Urban Development. Of these, the IRS 1040 provides the most significant information.
In the CHCK file generated March 2019, there are 64,440,000 unique records for children
who will be aged 0–17 years on June 1, 2020.
In addition to the links between parents and children available in the CHCK, we will also
utilize the links between household members which can be measured in the American
Community Survey, which is not an underlying data source for the CHCK. For each child in the
Numident aged 0-17 on June 1, 2020, we harvest relationships with the head of household and
the spouse of the head of household. We then use these links to supplement the links in the
CHCK.
Let us consider how many children from the Numident have been linked to a parent in the
CHCK file or to a parent in the ACS. Table 1 shows the number of children linked with both a
mother and a father, linked with a mother only, linked with a father only, linked with a parent in
the ACS or not linked with any parent.
Table 1: Child-parent links in the CHCK file relative to the Numident population, aged 0–17
years as of 2020, March 2019 CHCK file and ACS
Type of link
Frequency
Percent
Mother and father
Mother only
Father only
ACS link
No link
All children in
Numident

58,140,000
13,800,000
2,423,000
97,600
9,185,000
83,650,000

70%
17%
2.9%
0.1%
11%
100%

Figure 2 compares the distributions of date of birth for these children against the distribution
shown in Figure 1.

Figure 2: Frequency distributions of date of birth, Numident vs. CHCK entries, aged 0–17 years
as of June 1, 2019

The CHCK file was updated in March 2019 for NSCH sample frame production.
Updating household location using the MAF-ARF
In order to update household location, we use a Census dataset called the Master Address
Auxiliary Reference File (MAF-ARF). The MAF-ARF links person identifiers to address
identifiers using Census survey data and federal administrative data. The source data for the
MAF-ARF file are: the Census Numident, the 2010 Census Unedited File, the IRS 1040 and
1099 files, the Medicare Enrollment Database (MEDB), Indian Health Service database (IHS),
Selective Service System (SSS), and Public and Indian Housing (PIC) and Tenant Rental
Assistance Certification System (TRACS) data from the Department of Housing and Urban
Development, and National Change of Address data from the US Postal Service. Of these, the
IRS 1040 provides the most significant information.
Out of 83,650,000 1 children in the Numident, 68,390,000 are matched directly to a MAFID.
Out of 71,940,000 CHCK-matched mothers, about 66,400,000 are matched to a MAFID. Out of

1

All unweighted counts and estimates in this document are rounded to no more than four significant figures in
accordance with Census Disclosure Review Board rules on rounding.

60,560,000 CHCK-matched fathers, about 56,010,000 are matched to a MAFID. Additionally,
out of 9,433,000 ACS-matched parents, 8,790,000 are matched to a MAFID.
For each child observation from the Numident, we now have multiple possible MAFIDs: the
child-to-MAF-ARF MAFID, the child-to-CHCK-to-mother-to-MAF-ARF MAFID, the child-toCHCK-to-father-to-MAF-ARF MAFID, and the child-to-ACS parent-to-MAF-ARF MAFID. We
allocate a single MAFID to each child using that order. First, we assign the directly identified
child MAFID (65,750,000 cases). If the MAFID is missing, we assign the mother MAFID
(4,893,000 cases). Then, if the MAFID is still missing, we assign the father MAFID (2,032,000
cases). Finally, if the child, CHCK mother and CHCK father MAFIDs are missing, we assign the
ACS parent MAFID (43,000 cases). That leaves 10,930,000 children from the Numident not
assigned MAFIDs (a MAFID match rate of 86.9%).
There are some MAFIDs associated with a great number of children. As an example, out of
72,720,000 associated with a MAFID, 7,440,000 children are associated with a MAFID with
more than 20 child-MAFID links.
The 77,600,000 children associated with a MAFID are then collapsed down to 38,160,000
unique MAFIDS. This implies 1.91 children per household for households assigned a flag.
For 2020, we apply one additional step in the construction of stratum 1. We use
administrative HUD PIC and TRACS data, which contain flags for the number of children
present at the household level for all public housing and voucher households, to enhance the
existing stratum 1 process. We merge all MAFIDs not assigned a stratum 1 flag using the above
CHCK-MAF-ARF process with the most recent data on all public housing and voucher
households in the PIC-TRACS data. We will then assign a stratum 1 flag to all households which
have a child present flag in the HUD data. This adds 185,000 households to stratum 1.
We then need to scale up the MAFID list to the universe of MAFIDs to allow sampling of
unflagged households. A merge of the 38,160,000 unique child-flagged MAFIDS with the
January 2019 ACS MAF-X file matches 38,160,000 MAFIDS with child flags, removes
171,000,000 MAFIDS with child flags, and adds 400 MAFIDs without child flags. The sample
frame file now has about 209 million valid MAFIDS. Compare this with the 2011 ACS, in which
about 37 million out of 115 million households included related children. 2

Stratum 1 construction visualization
Figure 3 shows a visualization of the sample frame construction.

Figure 3: Stratum 1 construction
2

http://www.census.gov/prod/2013pubs/p20-570.pdf

Strata 2a and 2b: identifying probabilistic links from children to
addresses
In 2016, the Stratum 1 flag performed well. That is, the surveyed sample contained
approximately the same rate of children as had been predicted before the survey. The survey
team would like to further increase the sampling efficiency of the survey by adding more
information to the second stratum. By definition, Stratum 2 does not have explicit links from
children to households in the administrative data. In 2020 as in previous years, we further
bifurcate Stratum 2 into those households more likely to have children and those households less
likely to have children.

Households are assigned to Stratum 2a based on a model of child presence as a function of
variables available in administrative data for all households in the MAF. The model is estimated
with data from the most recent year of the ACS, in which child presence can be observed. Then
parameter estimates from that model can be used to predict the likelihood of child presence for
all households. These models are estimated separately for each state, and the threshold for
bifurcation is based on an objective of minimizing the size of Stratum 2a while also maintaining
95% coverage of children in Strata 1 and 2a.
Definitions
Population or sample concepts
• 2018 ACS sample, edited and swapped
– unit of observation is the household, unless noted otherwise
– sample includes sampled vacant dwellings, unless noted otherwise
• MAF
– population but restricted to MAFIDs marked as valid for ACS
Sample frame notation
•
•
•
•

h indexes household
s indexes states
C equals 1 if a household has any children, 0 otherwise
Strata:
– S1: household with children
– S2a: household likely to have children – S2b: household unlikely to have children

• Strata sizes:
– p(S1)
– p(S2a)
– p(S2b)

• Strata child rates:
– p(C|S1)
– p(C|S2a)
– p(C|S2b)

• Coverage with unsampled S2b:
– p(S1 ∪ S2a|C)

Model
Our goal is a scalar measure of the likelihood of a child being associated with a MAFID. This
measure must be available for all ACS-valid MAFIDs in the MAF. Using a sample in which the
presence of children is observable, we will estimate a model of child presence. The regressors
used to make the index prediction must be observable for all MAFIDs (i.e., to predict outside of
the estimation sample to the entire MAF).
The general model is:
Ch = f(Xh;θ),
where C is equal to one if a household includes any children and zero otherwise, X is a vector of
characteristics available for all households, and θ is an unknown vector of parameters.
We estimate the model using the most recent ACS 1-year sample:
E[Ch|Xh] = f(Xh;βˆACS) for households h in the ACS.

With parameter estimates from the ACS, we make predictions for the entire MAF:
Cˆh = f(Xh;βˆACS) for households h in the MAF.

In practice, we estimate models separately for each state. We do this to account for
systematic differences in administrative records coverage and MAF quality across states. The
model can now be specified as:
E[Chs|Xhs] = f(Xhs;βˆs,ACS) for households h in state s in the ACS,

where s is the MAFID’s state and the parameters βˆs,ACS now vary across states. The state-specific
predictions become:
Cˆhs = f(Xhs;βˆs,ACS) for households h in state s in the MAF.

Estimation
The model above is estimated as a linear probability model separately for each state using the
edited and swapped 2018 ACS sample. The outcome is child_present, a flag for whether a child
is present at the sampled MAFID.
The following covariates are included (with associated data sources) and are available for
each MAFID (except where a missingness flag is used):
• 2018 ACS 5-year published aggregate data
– acs_blkgrp_childrate_lvout: proportion of residents of block group who are children,
excluding the own-observation child counts from the numerator and denominator

• MAF-ARF
– female2050: flag for female between ages 20 and 50 at MAFID
– adult2050: flag for adults between ages 20 and 50 at MAFID
– coresid_sexdiff: flag for coresidence of men and women between ages 20 and 50 at
MAFID
– miss_adult2050: flag for missingness from MAF-ARF
• IRS 1040 filings, tax year 2018
– any_kid_deduct_max: does any tax form associated with this MAFID have any
deduction related to children? 3
– itemized_max: does any tax form associated with this MAFID use itemized
deductions?
– miss_any_kid_deduct_max: flag for MAFIDs without associated tax forms
• VSGI NAR commercial data
– vsgi_nar_homeowner_max: does any observation associated with this MAFID record
it as homeowener-occupied?
– miss_vsgi_nar_homeowner_max: flag for MAFIDs without associated VSGI data
• VSGI CRD commercial data
– homeowner_crd: is this address owner occupied in the VSGI CRD data
– any_child_crd: are there any children at this address
– num_children_crd: number of children at this address
– child_age_*: flags for child age ranges
– miss_*_crd: flags for missingness in the CRD data
• Targus commercial data
–
–
–
–
–
–
–

3

targus_homeowner_0: various flags for homeowner-occupied MAFID
targus_homeowner_A: various flags for homeowner-occupied MAFID
targus_homeowner_B: various flags for homeowner-occupied MAFID
targus_homeowner_C: various flags for homeowner-occupied MAFID
targus_homeowner_D: various flags for homeowner-occupied MAFID
targus_homeowner_E: various flags for homeowner-occupied MAFID
targus_homeowner_F: various flags for homeowner-occupied MAFID
miss_targus_homeowner: flag for MAFIDs without associated Targus data

The following IRS variable were used to make this variable: child exemptions and EITC qualifying children.

–

Parameter estimates are stored in the file frame2018_child_present_bystate.csv. In general
these models provide a reasonable fit, explaining between 40 and 50 percent of the in-sample
variation for most states.

Sample frame objective function
In order to choose an optimal Stratum 2a, we use the following objective function:
• Minimize the size of Stratum 2a while maintaining coverage of at least 95%
Stratum 2a is defined as:
S2a = {households in the MAF with Cˆh > C¯ but not in S1}.

Stratum 2b is defined as

S2b = {households in the MAF but not in S1 or S2a}.

With state-specific modeling, the objective function and coverage constraint also becomes
state specific:
• Minimize the size of Stratum 2a in each state while maintaining coverage of at least 95% in
each state
State-specific Stratum 2a is defined as:
S2a = {households in the MAF with Cˆhs > C¯s but not in S1}.

Stratum 2b is defined as before.

Optimization algorithm
The optimization parameter is a threshold on the child-present prediction probability, such that
MAFIDs with values above the threshold are assigned to Stratum 2a. Starting at a low threshold
(C¯) 4, follow this algorithm:

1. Under the current threshold C¯, calculate the proportion of MAFIDs in Stratum 2a, p(S2a),
and the coverage of Strata 1 and 2a under no sampling of Stratum 2b, (p(S1 ∪ S2a|C)).

4

The most conservative starting threshold would be at p(S1), where p(S2b) = 0.

2. If p(S2a) > 0 and p(S1 ∪ S2a|C) ≥ 0.95, then increase the child prediction threshold C¯ one
step (e.g., 0.01) and return to (1). If p(S1 ∪ S2a|C) < 0.95, then the previous threshold C¯ is
the optimal cutoff for S2a.

Under state-specific modeling, this algorithm is applied separately to each state.
Optimal strata
Table 2 shows the optimal strata under a 95% coverage constraint for Strata 1 and 2a. The
coverage constraint assumes non-sampling of Stratum 2b. The notation is as defined above. The
strata were optimized separately for each state using parameter estimates from separate state
regressions of child presence in the 2018 ACS microdata.

Table 2: 5 NSCH optimal Strata
State
US
AL
AK
AZ
AR
CA
CO
CT
DE
DC
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
ME
MD
MA

N
p(S1) p(S2) p(S3) p(C|S1) p(C|S2) p(C|S3) p(C|!S1) p(!S3|C) q
2143000
0.22 0.446 0.334
0.759
0.149
0.043
0.108
0.953
35000 0.208 0.535 0.257
0.697
0.133
0.053
0.11
0.95
8700 0.141
0.54 0.319
0.719
0.144
0.197
0.155
0.88
41500 0.205 0.473 0.322
0.75
0.16
0.046
0.119
0.951
20500 0.215
0.54 0.246
0.723
0.137
0.06
0.115
0.95
201000 0.268 0.367 0.364
0.765
0.189
0.044
0.122
0.952
35500 0.227
0.41 0.363
0.789
0.16
0.039
0.106
0.95
21500 0.228 0.374 0.398
0.79
0.157
0.035
0.099
0.952
6800
0.19 0.361 0.448
0.745
0.139
0.029
0.085
0.952
4300 0.174 0.594 0.232
0.655
0.076
0.038
0.066
0.951
113000 0.196 0.401 0.404
0.681
0.142
0.028
0.088
0.952
51500 0.243 0.453 0.304
0.732
0.168
0.052
0.126
0.953
9200 0.139 0.613 0.248
0.696
0.241
0.053
0.189
0.951
11000 0.215 0.455
0.33
0.78
0.161
0.044
0.113
0.951
89000 0.226 0.407 0.367
0.768
0.16
0.041
0.109
0.952
44000 0.228 0.422
0.35
0.76
0.152
0.043
0.107
0.951
32000 0.197 0.645 0.158
0.796
0.086
0.206
0.098
0.942
24500 0.221 0.394 0.385
0.776
0.157
0.04
0.106
0.953
31000 0.222 0.569 0.209
0.767
0.135
0.072
0.119
0.95
27500 0.224 0.454 0.322
0.682
0.149
0.045
0.11
0.951
16000 0.141 0.483 0.376
0.77
0.093
0.03
0.07
0.95
35500 0.245
0.38 0.374
0.785
0.162
0.041
0.106
0.95
39500 0.213 0.414 0.373
0.803
0.143
0.036
0.095
0.952
5

National Survey of Children’s Health sample frame

31
21
-1
28
20
37
36
40
43
21
40
28
21
30
34
33
-1
34
17
31
28
37
37

C_hat_S2
0.032
0.034
-0.199
0.06
0.035
0.11
0.08
0.083
0.066
0.014
0.069
0.071
0.101
0.084
0.072
0.062
-0.164
0.068
0.012
0.078
0.03
0.081
0.075

MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY

94500
69000
17000
46500
10500
19500
18000
10500
50000
15000
124500
64000
8800
83000
43500
25500
110000
5900
30500
9000
40500
137000
18000
8100
50500
45500
13500
71500
4100

0.2
0.209
0.222
0.21
0.15
0.206
0.227
0.174
0.233
0.166
0.206
0.215
0.177
0.219
0.215
0.209
0.2
0.2
0.212
0.186
0.23
0.256
0.306
0.149
0.24
0.23
0.151
0.199
0.172

0.35
0.363
0.55
0.446
0.698
0.585
0.451
0.491
0.373
0.689
0.476
0.438
0.682
0.406
0.658
0.448
0.378
0.371
0.441
0.69
0.415
0.46
0.4
0.618
0.394
0.419
0.684
0.364
0.754

0.45
0.428
0.228
0.344
0.151
0.209
0.322
0.334
0.394
0.146
0.318
0.348
0.141
0.376
0.128
0.342
0.422
0.429
0.347
0.124
0.355
0.284
0.294
0.233
0.365
0.351
0.165
0.437
0.074

0.787
0.828
0.702
0.761
0.77
0.8
0.72
0.807
0.793
0.677
0.753
0.746
0.776
0.772
0.735
0.782
0.795
0.758
0.719
0.782
0.747
0.752
0.825
0.797
0.783
0.796
0.73
0.807
0.777

0.141
0.14
0.136
0.134
0.098
0.106
0.153
0.1
0.183
0.119
0.15
0.155
0.081
0.148
0.131
0.134
0.137
0.152
0.133
0.102
0.152
0.18
0.189
0.082
0.158
0.16
0.104
0.145
0.106

0.028
0.033
0.065
0.039
0.151
0.1
0.046
0.035
0.038
0.176
0.043
0.041
0.169
0.034
0.139
0.041
0.032
0.031
0.039
0.201
0.039
0.064
0.066
0.044
0.041
0.042
0.171
0.031
0.146

0.083
0.087
0.118
0.096
0.105
0.104
0.113
0.076
0.114
0.126
0.11
0.109
0.091
0.097
0.132
0.097
0.087
0.092
0.096
0.113
0.104
0.141
0.14
0.072
0.106
0.109
0.117
0.088
0.11

0.953
0.952
0.951
0.952
0.926
0.95
0.951
0.952
0.952
0.934
0.951
0.951
0.935
0.954
0.95
0.951
0.952
0.95
0.95
0.934
0.952
0.951
0.95
0.952
0.95
0.95
0.872
0.951
0.951

43
40
18
31
-1
9
31
31
38
-1
30
33
-1
37
3
31
40
43
33
-1
34
25
25
17
36
34
-1
41
1

0.075
0.071
0.024
0.052
-0.188
-0.034
0.066
0.046
0.101
-0.153
0.068
0.073
-0.152
0.07
-0.046
0.055
0.067
0.086
0.06
-0.191
0.076
0.073
0.097
0.004
0.082
0.078
-0.185
0.07
-0.106

Auditing the sample frame against the ACS
To examine the performance of the administrative records used to build the sample frame, we
merge the list of MAFIDs constructed above with the American Community Survey housing-unit
sample from 2018. This is an in-sample audit, and as such it will by design meet the 95%
threshold.

All estimates are weighted with the housing-unit-level weights, which include weight for
vacant units (about 200,000 vacant housing units in the 2018 ACS). In vacant housing units, we
assign zero children. These estimates should reflect the NSCH survey production process.
State-specific performance
In 2020, the smallest oversample strata were in Hawaii, Maine, Vermont, and West Virginia. The
largest oversample strata are in California, Texas, and Utah. The highest rates of Type 1 error are
in DC, Florida, Louisiana, Mississippi, Nevada, and South Carolina. The highest rates of Type 2
error were in Alaska, Hawaii, New Mexico, Texas, and Utah.

Table 3: 6 NSCH strata, ACS, all addresses audit
State N
p(S1)
p(S2)
p(S3)
p(C|S1) p(C|S2) p(C|S3) p(C|!S1) p(!S3|C)
US
2143000
0.22
0.437
0.343
0.759
0.152
0.043
0.108
0.95
AL
35000
0.208
0.535
0.257
0.697
0.133
0.053
0.11
0.95
AK
8700
0.141
0.54
0.319
0.719
0.144
0.197
0.155
0.88
AZ
41500
0.205
0.471
0.324
0.75
0.16
0.046
0.119
0.951
AR
20500
0.215
0.54
0.246
0.723
0.137
0.06
0.115
0.95
CA
201000
0.268
0.367
0.365
0.765
0.189
0.045
0.122
0.951
CO
35500
0.227
0.411
0.362
0.789
0.16
0.039
0.106
0.95
CT
21500
0.228
0.374
0.398
0.79
0.157
0.035
0.099
0.952
DE
6800
0.19
0.361
0.448
0.745
0.139
0.029
0.085
0.952
DC
4300
0.174
0.595
0.231
0.655
0.076
0.038
0.066
0.951
FL
113000
0.196
0.401
0.403
0.681
0.142
0.028
0.088
0.952
GA
51500
0.243
0.452
0.305
0.732
0.168
0.052
0.126
0.953
HI
9200
0.139
0.613
0.248
0.696
0.241
0.053
0.189
0.951
ID
11000
0.215
0.454
0.331
0.78
0.161
0.044
0.113
0.951
IL
89000
0.226
0.407
0.368
0.768
0.16
0.041
0.109
0.952
IN
44000
0.228
0.422
0.35
0.76
0.152
0.043
0.107
0.951
IA
32000
0.197
0.645
0.158
0.796
0.086
0.206
0.098
0.942
KS
24500
0.221
0.393
0.386
0.776
0.158
0.04
0.106
0.953
KY
31000
0.222
0.569
0.209
0.767
0.135
0.072
0.119
0.95
LA
27500
0.224
0.454
0.322
0.682
0.149
0.045
0.11
0.951
ME
16000
0.141
0.486
0.373
0.77
0.093
0.03
0.07
0.95
MD
35500
0.245
0.38
0.375
0.785
0.162
0.042
0.106
0.95
MA
39500
0.213
0.414
0.373
0.803
0.143
0.036
0.095
0.952
MI
94500
0.2
0.351
0.449
0.787
0.141
0.028
0.083
0.953
MN
69000
0.209
0.363
0.428
0.828
0.14
0.033
0.087
0.952
MS
17000
0.222
0.549
0.229
0.702
0.136
0.065
0.118
0.951
MO
46500
0.21
0.446
0.344
0.761
0.134
0.039
0.096
0.952
MT
10500
0.15
0.698
0.151
0.77
0.098
0.151
0.105
0.926
NE
19500
0.206
0.586
0.209
0.8
0.105
0.1
0.104
0.95
NV
18000
0.227
0.452
0.321
0.72
0.153
0.046
0.113
0.951
NH
10500
0.174
0.492
0.333
0.807
0.101
0.034
0.076
0.954
NJ
50000
0.233
0.372
0.394
0.793
0.183
0.038
0.114
0.952
NM
15000
0.166
0.689
0.146
0.677
0.119
0.176
0.126
0.934
NY
124500
0.206
0.476
0.318
0.753
0.15
0.043
0.11
0.951
NC
64000
0.215
0.438
0.348
0.746
0.155
0.041
0.109
0.95

6

National Survey of Children’s Health sample frame

ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY

8800
83000
43500
25500
110000
5900
30500
9000
40500
137000
18000
8100
50500
45500
13500
71500
4100

0.177
0.219
0.215
0.209
0.2
0.2
0.212
0.186
0.23
0.256
0.306
0.149
0.24
0.23
0.151
0.199
0.172

0.682
0.405
0.658
0.448
0.379
0.371
0.442
0.69
0.415
0.459
0.4
0.62
0.394
0.419
0.684
0.364
0.753

0.141
0.376
0.128
0.343
0.421
0.429
0.346
0.124
0.355
0.284
0.294
0.231
0.366
0.35
0.165
0.437
0.075

0.776
0.772
0.735
0.782
0.795
0.758
0.719
0.782
0.747
0.752
0.825
0.797
0.783
0.796
0.73
0.807
0.777

0.081
0.148
0.131
0.135
0.137
0.152
0.133
0.102
0.152
0.18
0.19
0.082
0.157
0.16
0.104
0.145
0.106

0.169
0.034
0.139
0.041
0.032
0.031
0.039
0.201
0.039
0.064
0.066
0.044
0.042
0.042
0.171
0.031
0.147

0.091
0.097
0.132
0.097
0.087
0.092
0.096
0.113
0.104
0.141
0.14
0.072
0.106
0.109
0.117
0.088
0.11

0.935
0.954
0.95
0.951
0.952
0.95
0.95
0.934
0.952
0.951
0.95
0.952
0.95
0.95
0.872
0.951
0.95

We additionally audit the frame out of sample against an early release file of 2019 ACS
microdata, as shown in table 4. Currently, this audit uses unedited ACS data (i.e., item
nonresponse are left as missing and are not imputed including children’s age). If item
nonresponse is random with respect to the presence of children in the household, this should not
cause any systematic bias in the audit. Performance is slightly lower than the in-sample audit in
table 3, but is in line with the results in previous sample years.
Table 4: 7 NSCH strata, ACS2019, all addresses audit
State
US
AL
AK
AZ
AR
CA
CO
CT
DE
DC
FL
GA
7

N
1889000
29000
5800
36000
17500
181000
32500
19000
5700
3800
97000
45000

p(C|!S1 p(!S3|C
p(S1)
p(S2)
p(S3)
p(C|S1) p(C|S2) p(C|S3) )
)
0.233
0.416
0.351
0.834
0.127
0.046
0.09
0.938
0.232
0.516
0.252
0.785
0.116
0.045
0.092
0.955
0.183
0.598
0.219
0.748
0.176
0.435
0.246
0.718
0.221
0.457
0.322
0.815
0.131
0.054
0.099
0.933
0.231
0.531
0.239
0.804
0.125
0.062
0.106
0.945
0.275
0.35
0.374
0.826
0.16
0.038
0.097
0.953
0.236
0.399
0.364
0.858
0.126
0.031
0.081
0.957
0.223
0.367
0.411
0.87
0.129
0.025
0.074
0.96
0.209
0.363
0.428
0.815
0.106
0.028
0.064
0.946
0.169
0.572
0.259
0.749
0.067
0.032
0.056
0.952
0.209
0.39
0.402
0.778
0.117
0.028
0.072
0.949
0.263
0.432
0.306
0.801
0.133
0.047
0.097
0.949

National Survey of Children’s Health sample frame

HI
ID
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN
MS
MO
MT
NE
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SD
TN
TX
UT
VT
VA
WA
WV
WI
WY

7700
9700
80500
40000
29500
22000
27500
22500
12000
31500
35500
80500
61500
14000
41000
8600
17500
16000
8900
45000
11500
108000
55000
7300
75000
34000
23000
95500
5200
26500
7700
36500
119000
16000
6400
46000
41000
11000
62500
3400

0.146
0.257
0.229
0.234
0.204
0.233
0.231
0.243
0.177
0.263
0.222
0.219
0.228
0.242
0.225
0.172
0.225
0.239
0.212
0.25
0.193
0.222
0.231
0.205
0.227
0.236
0.215
0.211
0.208
0.228
0.217
0.239
0.273
0.331
0.182
0.253
0.242
0.18
0.214
0.208

0.606
0.411
0.386
0.404
0.643
0.375
0.566
0.427
0.497
0.359
0.397
0.333
0.332
0.532
0.431
0.736
0.582
0.446
0.442
0.355
0.718
0.453
0.413
0.694
0.382
0.664
0.444
0.359
0.352
0.421
0.698
0.401
0.439
0.393
0.587
0.377
0.404
0.717
0.342
0.735

0.248
0.333
0.385
0.362
0.153
0.392
0.203
0.33
0.326
0.378
0.381
0.448
0.44
0.226
0.345
0.093
0.193
0.315
0.346
0.395
0.088
0.325
0.356
0.102
0.391
0.1
0.341
0.429
0.44
0.35
0.085
0.36
0.288
0.277
0.231
0.37
0.354
0.103
0.444
0.056

0.744
0.849
0.841
0.842
0.876
0.862
0.834
0.754
0.838
0.852
0.869
0.857
0.886
0.756
0.841
0.816
0.872
0.799
0.867
0.859
0.75
0.82
0.827
0.846
0.855
0.779
0.847
0.873
0.834
0.796
0.863
0.818
0.816
0.872
0.862
0.863
0.853
0.822
0.868
0.802

0.252
0.137
0.13
0.13
0.063
0.137
0.108
0.135
0.087
0.126
0.12
0.119
0.12
0.134
0.12
0.096
0.078
0.128
0.095
0.155
0.111
0.138
0.129
0.081
0.119
0.126
0.109
0.113
0.125
0.112
0.098
0.129
0.152
0.173
0.088
0.123
0.134
0.084
0.125
0.108

0.062
0.042
0.049
0.042
0.236
0.058
0.057
0.043
0.033
0.033
0.027
0.03
0.034
0.078
0.043
0.253
0.158
0.034
0.028
0.032
0.289
0.036
0.035
0.245
0.031
0.186
0.041
0.029
0.036
0.035
0.266
0.037
0.052
0.068
0.039
0.027
0.034
0.226
0.029
0.238

0.197
0.095
0.09
0.088
0.096
0.097
0.095
0.095
0.066
0.078
0.075
0.068
0.071
0.117
0.086
0.113
0.098
0.089
0.066
0.09
0.131
0.095
0.086
0.102
0.075
0.133
0.079
0.067
0.076
0.077
0.116
0.085
0.113
0.129
0.074
0.076
0.087
0.102
0.071
0.117

0.944
0.951
0.927
0.942
0.859
0.917
0.956
0.944
0.947
0.956
0.959
0.944
0.943
0.935
0.942
0.9
0.888
0.959
0.959
0.955
0.898
0.955
0.951
0.902
0.952
0.935
0.943
0.948
0.931
0.949
0.919
0.949
0.95
0.95
0.958
0.963
0.956
0.9
0.947
0.949

Local-area Internet-accessibility
Here we describe the construction of a tract-varying Internet-accessible household flag.
Since 2012, ACS respondents have been able to submit survey forms over the Internet. ACS
paradata record whether a respondent chose the online option. The ACS paradata has been
summarized at the tract level. Our Internet-accessible household measure is equal to a weighted
proportion of the respondents that chose to submit the ACS survey over the Internet if given the
option to do so. Figure 4 shows the kernel-smoothed distribution of tract-level Internet response
for the 2013–2014 ACS survey years.
Figure 4: Kernel-smoothed probability distribution function of tract-level ACS Internet response
rate, ACS paradata, 2013–2014 survey years

0

.2
.4
.6
.8
ACS Internet response rate, weighted, by tract

1

To construct an Internet-access flag, we use the first tritile for a cut-off. A block is considered
to have low Internet access if the Internet accessibility index is below the first tritile of the blocklevel distribution. For low-population blocks, we replace missing values of the block-varying
low-Internet flag with the modal value from the corresponding block group. For very new
housing units without assigned Census blocks, we assign a value of zero for this binary variable
(i.e., the default for these new households is high Internet accessibility.)

Local-area household income relative to the poverty rate
The frame has a set of poverty variables from the 2018 5-year American Community Survey file.
These variables measure the proportion of households with household income in an interval
defined by the poverty rate. Figure 5 shows the kernel-smoothed probability distribution function
of the proportion of households in the block group that have household income less than 150% of
the poverty rate.
Figure 5: Kernel-smoothed probability distribution function of block-group-level 150% poverty
rate, ACS, 2018 5-year file

0
.2
.4
.6
.8
1
Proportion of individuals below 150% of poverty line, weighted, by block group

Final sample frame data layout
The component data files are merged together based on MAFID. The data layout for this
combined file is given in Table 2.
Table 2: NSCH population data file layout
Variable name

Label

Level of
variation

Type

Any
missing?

MAFID

long

no

State
County
Tract
Blocck
group
Block
MAFID
MAFID

str2
str3
str6

no
no
yes

str1

yes

str4
byte
byte

yes
no
no

maf_curstate
maf_curcounty
maf_curblktract

Master Address File
ID
State
County
Tract

maf_curblkgrp

Block group

maf_curblk
stratum1
stratum2a

Block
Stratum 1 identifier
Stratum 2a identifier

stratum2b
kids_00_02

Stratum 2b identifier
Number of children
aged 0–2 years
Number of children
aged 3–5 years

MAFID
MAFID

byte
integer

No
no

MAFID

integer

no

kids_06_08

Number of children
aged 6–8 years

MAFID

integer

no

kids_09_11

Number of children
aged 9–11 years

MAFID

integer

no

mafid

kids_03_05

kids_12_14

Number of children
aged 12–14 years

MAFID

integer

no

kids_15_17

Number of children
aged 15–17 years

MAFID

integer

no

blkgrp_185_200_povrate

Pr. HH w/ inc. 185–
200% poverty rate

Block
group

float

yes

blkgrp_gt_200_povrate

Pr. HH w/ inc. > 200%
poverty rate

Block
group

float

yes

blkgrp_lt_150_povrate

Pr. HH w/ inc. <
150% poverty rate

Block
group

float

yes

MAFID

byte

yes

mailvaldf
Valid mailing address
Filename: nsch_pop_file.sas7bdat
Population: all MAFIDs in 2019 MAF-X
Unit of observation: household (MAFID)