WIC income verification study final report

              United States
Department of
Agriculture
Food and
Nutrition
Service
Office of
Analysis and
Evaluation
WIC Income
Verification Study
Final Report
COMPLETED
^ . BEST COPY AVAILABLE
WIC Income Verification Study
Final Report
oV
ACKNOWLEDG-MENTS
On behalf of Quality Planning Corporation and the project staff. I
would like to thank the following individuals and organizations for
their patience and assistance during the course of this project.
Dr. Steven Gale of the Office of Analysis and Evaluation of FNS
served as the Project Officer, saw us through the many changes
of emphasis and direction the project experienced, and wrote the
executive summary. The project advisory panel provided us with
balanced and diverse advice, support, and comments. Members
included leading national figures in WIC—Dennis Bach, Barbara
Blessing, Jane Fouste, Theresa Landau, Charlotte Lowery, Jack
Metz, Philip Petrosky, Carmen Rodriquez, and Debra Stabeno.
Joseph Steinberg, study methodological consultant, aided us in
many technical aspects of the research design and analysis.
Essential clarifications on WIC policy were provided by Clara
French, Ron Vogel, and Philip Cohen of the WIC National Office.
The entire project was made possible by generous cooperation of
the WIC community. We especially wish to thank State WIC
Officials from the 25 sample States who patiently answered our
endless questions on WIC policy and procedure. The hundreds of
hours of time spent by WIC staff in PC88 locals and clinics in
sampling, record abstraction, and interviewing assured that the
results are truly nationally representative. Finally, we wish to
thank the 884 WIV survey respondents who tolerated our intrusion
into their lives and finances.
Richard Beatty and Ester Fleishman of Westat were responsible
for assuring a very high quality of data from the WIV survey. The
Quality Planning project staff were headed by Arthur Burger,
Project Director; Gail Armstrong Taff, Project Manager; Janice
Clinthorne, Survey Director; Janet Holbrook, Programmer; Gail
Simpson, Senior Analyst; and Douglas Burger, Research
Associate. This report was drafted by Gail Armstrong Taff and
Daniel Finnegan with sections and reviews contributed by Arthur
Burger, Gail Simpson, and Douglas Burger.
Daniel Finnegan
President
Quality Planning Corporation
ii
TABLE OF CONTENTS
Chapter Page
ACKNOWLEDGMENTS fj
LIST OF FIGURES v ?
LIST OF TABLES v| |
EXECUTIVE SUMMARY v ? 11
1 INTRODUCTION 1-1
The WIC Program 1-1
Income Eligibility in WIC 1-2
The WIC Income Verification (WIV) Project 1-3
Summary of Prior Research 1-4
Purpose of the Phase II Survey 1-5
Overview of the Report 1-6
2 PHASE I ACTIVITIES AND FINDINGS 2-1
Income Verification Methods Currently Used in WIC 2-1
Summary and Implications for Phase II 2-6
3 SURVEY METHODOLOGY 3-1
Data Collection Methods and Schedule 3-1
In-home Audit Sampling Procedure 3-3
In-home Audits 3-5
Survey Yield 3-6
Survey Limitations 3-11
Sampling Variance 3-11
Measurement Error 3-11
Operational Definitions of Major Variables 3-13
Applicant/Enrollee/Participant 3-13
Certificatton/Recertification v 3-13
Respondent 3-14
Economic Unit/Family/Household 3-14
Income 3-15
Financial Data Reference Period (FDRP) 3-16
Income-eligibility Standard 3-16
Income-eligibility Error 3-17
Income Certification Error Determination Procedure 3-17
4 THE WIC ECONOMIC UNIT 4-1
Federal Regulations and Guidance 4-1
State Policies and Procedures 4-4
General Definitions 4-6
Special Case Definitions 4-7
iii
4
Chapter Page
Examples of State Economic Unit Determination
Procedures 4-8
Example 1. State with Simple Procedures 4-9
Example 2. State with Moderately Complex
Procedures 4-11
Example 3. State with Complex Procedures 4-15
Summary Differences in State Procedures 4-19
Summary and Conclusions 4-24
INCOME OF THE WIC ECONOMIC UNIT 5-1
Countable Income 5-1
State Definitions of Income 5-3
Income-eligibility Limit 5-3
Income Characteristics 5-3
Income by Source 5-5
Income Reporting Error 5-5
Income Change 5-8
Summary and Conclusions 5-10
WIC INCOME-ELIGIBILITY ERROR 6-1
Extent and Nature of Income Certification Error 6-1
Certification Error Rate 6-3
Dollar Error Rate 6-4
Participation Weights 6-4
Comparison with Other Programs' Income-eligibility
Error Rates 6-7
Income Reporting Error and Eligibility Error 6-8
Low Income Distribution of WIC Enrollees 6-8
Continued Eligibility During Certification Period
in Spite of Income Change 6-10
Lack of a Documented Basis for Many
Eligibility Decisions 6-10
Characteristics Associated with Error 6-12
Income Distribution of Ineligible Enrollees 6-12
ENROLLEES' ABILITY TO MEET INCOME
VERIFICATION REQUIREMENTS 7-1
Current State and Local Practices 7-1
Practices Reported by Agencies 7-1
Practices Reported by Enrollees 7-1
Enrollee Capacity to Respond 7-4
r iv
Chapter Page
8 STUDY CONCLUSIONS 8-1
Income-eligibility Error 8-1
Variations in State Policies and Procedures 8-2
Economic Unit 8-4
Countable Income 8-6
Other Findings 8-6
Income Characteristics of WIC Households 8-6
Income Reporting Error 8-6
Appendixes
A Sample Design and Variance A-1
B In-home Audit Questionnaire B-1
C References C-1
D Reconcilation Substudy ,. D-1
E Federal Income-eligibility Regulations and Guidance E-1
F Alternative Definitions of Eligibility Error F-1
G General Definitions of WIC Economic Unit G-1
£>
LIST OF FIGURES
Figure Page
2-1 Primary income verification method 2-4
2-2 Income verification method by type of State agency 2-5
3-1 WIC income verification survey sample States and
local agencies 3-4
4-1 Income-eligibility determination tree 4-5
4-2 State with simple procedures 4-10
4-3 State with moderately complex procedures 4-12
State with complex procedures 4-16
8-1 Percent of States using special case procedures 8-5
8-2 Income distribution of WIC enrollees as a percent of
poverty level 8-7
8-3 Income reporting error 8-8
vi
LIST OF TABLES
Table Page
3-1 Summary of All Stages of the Sample Design 3-7
3-2 Total Survey Response Rates by Participant Category 3-9
4-1 Distribution of Some Procedures for Determining
Economic Unit Membership Across the 25 States
in the WIV Sample 4-20
5-1 State WIC Agencies'Treatment of Income Sources 5-4
5-2 Income Distribution of WIC Enrollees Relative to
Poverty Index 5-6
5-3 WIC Economic Unit Income Sources 5-7
5-4 Income Reporting Error 5-9
5-5 Income Change Among WIC Enrollees 5-11
6-1 WIC Income-eligibility Error 6-2
6-2 Number of Months of Participation During the 6 Months
Following Certification or Recertification 6-5
6-3 Ineligibility Rate by Amount of Income Underreporting 6-9
6-4 Income Distribution of ineligible WIC Enrollees
Relative to Poverty Index 6-13
6-5 Income of Ineligible WIC Enrollees After Certification
Relative to Income-eligibility Standard 6-14
6-6 Certification Error Rates by Enrollee and Household
Characteristics 6-16
6-7 Certification Error Rates by Participation in Other
Social Welfare Programs 6-17
7-1 Percent of WIC Enrollees by State Agency Income
Verification Methods 7-2
7-2 Percent of WIC Enrollees by Local Agency Income
Verification Methods 7-3
7-3 Percent of Enrollees by Reported Income Verification
Procedures Used at Certification 7-5
7-4 Documentation Procedures Used for WIC Enrollees
Who Reported That Documentation Was Requested 7-6
7-5 Percent of Enrollees Capable of Complying with
Alternative Documentation Requirements 7-7
8-1 WIC Income-eligibility Error 8-3
t vll
EXECUTIVE
SUMMARY
SYNOPSIS
BACKGROUND AND
OVERVIEW
The WIC Income Verification (WIV) Study examined the incidence
and magnitude of certification error by conducting independent in-home
audits on a national sample of program participants in late
1988. The income-eligibility determination process was modeled
for each State, and data from the in-home audit were compared to
case-file-derived income and family-size data. Principal findings
reveal a national WIC certification case error and dollar error rate,
respectively, of 5.7 and 5.8 percent. This translates into $84
million spent on ineligibles out of a total of about $1.5 billion in
FY88.1 Participants had relatively low household incomes, but
income misreporting was high—over 43 percent of participants
underreported, while 22 percent overreported, their income.
Almost all households enrolled in the program in 1988 had
incomes at or below the allowable Federal threshold of 185
percent of poverty.
To be eligible to participate in USDA's Special Supplemental Food
Program for Women, Infants, and Children (WIC), applicants must
meet three types of criteria: categorical, nutritional, and income.
Applicants must also reside in the State or local area in which they
are applying for benefits. Findings from this study focus on
income eligibility and describe income certification procedures in
effect and the extent and nature of certification error nationwide
during the period covered by the study.^
In response to the concerns of the U.S. General Accounting Office
(GAO) about the accuracy of income reported by WIC
participants, the WIV Study began in 1986. The study consisted
of two phases. The first phase focused on the analysis of case-file
and other income data from existing national surveys and
studies rather than a collection of new data. In Phase I, the
contractor also collected data on current certification procedures
from all State agencies via telephone interviews with WIC State
directors and their staffs in the fall of 1986. Other Phase I
activities included the design and development of a national
survey to estimate the incidence and magnitude of income
ineligibility.
In Phase II, in-depth income audits were conducted with WIC
participants according to the survey design developed in Phase I.
The audits were conducted at participants' homes, and the
information was compared to case-file data obtained initially at
certification (or recertification). The sample for Phase II audits
was nationally representative and drawn exclusively from
respondents to the Agency's 1988 study of WIC Participant and
Program Characteristics (PC88). The income audits were
conducted in the fall of 1988, and the final WIV sample consisted
of 25 States, 41 local agencies, 72 clinics, and 1,076 participants.
STUDY FINDINGS
WIV Project
Findings—Phase I (Fall
1986)
State Certification
Procedures
Self-declaration, whereby applicants self-report their income
and family size to local staff without providing documentation,
was the primary method in 46 percent of all (geographic,
Indian, and territorial) WIC State agencies.3
In 25 percent of all State agencies, documentation of income
was required at certification. An applicant who failed to
provide income documentation could not be certified for
program participation. In the remaining State agencies, some
variation of self-declaration (16 percent) or a weaker form of
documentation (13 percent) was asked for, but not required, at
certification.
Analysis of certification procedures of the 51 geographic State
agencies (which serve the majority of WIC enrollees) revealed
that in 37 percent of these agencies, the primary certification
procedure was to require applicants to submit income and
household documentation.
In the geographic State agencies, requiring documentation at
certification was followed by local variation (27 percent).
Under this procedure local agencies have the option to use
more rigorous verification procedures. Lastly, statewide self-declaration
(20 percent) and documentation-requested
procedures (16 percent) were utilized, but to a lesser degree.
In Indian State agencies, self-declaration (85 percent) was the
primary procedure used to establish income and family size.3
ix
ID
WIV Project
Findings -Phase II (Fall
1988)
In-Home Audit Survey
Characteristics
State Certification
Procedures: Economic
Unit Determination
State Certification
Procedures: Income
Determination
Data were obtained from participants in 25 State agencies, 41
local agencies, and 72 clinics.
Ninety-six percent of all PC88 participants agreed to participate
in the WIV Study, and over 85 percent of these participants
completed the in-home audit (n-834).
Examination of State policies and procedures in light of
Federal regulations governing the determination of the
applicant's economic unit (i.e., household or family size)
indicated significant State-to-State variation. These definitional
variations affect the basic terms necessary to correctly certify
an applicant as eligible to receive WIC benefits. For example,
such terms as "family membership" and "sharing of income"
differed from State to State.
Further, State-to-State variance in certification of family size
also existed with respect to more specialized definitional terms
and rules regarding foster children, emancipated minors,
institutionalized persons, individuals living apart from the
family, length of family member absence, etc. All States had
provisions for certifying some types of special cases.
Comparisons between surveyed States revealed that the most
common types of special-case categories addressed were
foster child (92 percent), emancipated minor (64 percent), and
student Irving apart from the family (60 percent).
In most States surveyed (84 percent), presumptive eligibility in
WIC certification was allowed. That is, documented
participation in another Federal program with income limits at
or below the State's WIC limits was accepted as evidence of
income eligibility for WIC. The most common basis for
presumptive eligibility used by States was Medicaid (64
percent), followed by Aid to Families with Dependent Children
(48 percent), Food Stamps (40 percent), School Lunch
Program (16 percent), and Supplemental Security Income (8
percent).
Unlike the State-to-State variation in defining the WIC
economic unit or family, examination of State policies on
countable income revealed complete or near complete
agreement on 19 out of 23 common income sources, which, as
a group, constitute 91.6 percent of all countable income.
//
Income Characteristics of
WIC Families
Income Reporting and
Change
State definitions of countable income differed on four major
income sources, namely, the treatment of housing subsidies,
Black Lung benefits, loans, and student financial aid.
However, these sources collectively accounted for only 4.2
percent of all income.
WIC households are predominantly low income. That is, 60
percent of WIC households have verified income under 100
percent of poverty, and almost 95 percent have incomes at or
below the allowable 185 percent Federal poverty income level.
The median income of WIC households is 80 percent of the
Federal poverty income level.
Approximately 63 percent of all WIC households have income
from salaries and wages. Wage and salary income is not only
the most frequent source, but also the highest median monthly
amount. Wage and salary income accounts for almost 75
percent of total WIC income.
The only other major income source received by WIC
households is AFDC (26 percent), which accounts for about 10
percent of total family income. Two categories—wages
salaries and AFDC—account for about 84 percent of total WIC
household income.
Income from child support (9.3 percent), housing subsidies
(6.9 percent), and Social Security (3.8 percent) is less
frequently included in the income of WIC families and
collectively accounts for only 5.2 percent of total countable
income. All other sources of income are received by 3 percent
or fewer of WIC households, and no single source accounts for
more than 1 percent of total household income.
When reported income for certification is compared with
verified in-home audit findings, data reveal that only 36 percent
of enrollees correctly reported their income—with "correct"
defined as reported monthly income within $50 of verified in-home
income for the same time period. Income was
underreported about 43 percent of the time and overrepcrted
about 22 percent of the time. Thus, substantial income
misreporting exists.
In-home audit data indicate that over a 4-to-5 month period
following certification, about 55 percent of WIC households
maintained stable incomes—with "stable" defined as either no
change or a change of less than $100 per month.
\x>
• Over this same time period, approximately 26 percent of WIC
households experienced an income increase (15.4 percent) or
decrease (10.5 percent) of between $100 and $500 per month.
Of the sampled households, 19 percent experienced an
income change—either an increase (10.3 percent) or decrease
(8.7 percent) of greater than $500 per month.
Income Eligibility: Error • The national certification error rate (defined as the percent of
Rates and Characteristics enroltees certified to receive, but not eligible for, WIC benefits)
is 5.7 percent. This translates into a total number of ineligibles
of approximately 214,000 out of a nationwide WIC enrollee
population of about 3,700,000 in FY88.4
• The dollar error (defined as the amount of WIC food funds
spent on income ineligibles) is 5.8 percent. This translates into
$84 million spent on ineligibles out of a total WIC expenditure
of about $1.5 billion in FY88.1
• Ineligible enrollees have, for the most part, relatively low
incomes. That is, about 23 percent have incomes between
186 and 200 percent of poverty, 45 percent between 201 and
250 percent of poverty, 17 percent between 251 and 300
percent of poverty, and less than 5 percent have incomes at or
over 300 percent of poverty.
• While about 55 percent of those determined to be income-ineligible
at certification—based on in-home audit
data—remain ineligible some 3 to 4 months later,
approximately 45 percent experience post-certification income
decreases that would make them income eligible.
• No significant differences in error rates exist across
racial/ethnic groups, household size, and participant category
(i.e., women, infants, or children). On the other hand, for those
who report relatively high income or report wage or salary
income, significantly higher than average ineligibility rates were
observed.
• WIC households that participate in other Federal
programs—Food Stamps, AFDC, Medicaid, and the
Temporary Emergency Food Assistance Program
(TEFAP)—have significantly lower than average ineligibility
rates.
xll
IS
MAJOR STUDY
CONCLUSIONS
National Error Rates. The national WIC certification case error
rate is 5.7 percent. This rate is relatively low compared to
other food assistance programs. The national dollar error rate
is 5.8 percent. This translates into $84 million spent on
inelkjibles out of a total of about $1.5 billion in FY88.1 These
low case error and dollar error rates are due, in large part, to
the very low incomes of WIC households. That is,
considerable underreporting of income exists but does not
seriously affect eligibility. An upward shift in the income
distribution of WIC participants could substantially increase
error rates.
Error Rates Issues. Several factors unassociated with the
error rate tend to keep WIC dollar errors relatively low. These
programmatic factors include the concept of a "single level"
benefit award whereby participants receive a fixed WIC benefit
independent of family income (i.e., no errors are associated
with benefit over-issuance). Also, because there is no Federal
requirement that participants report income changes, eligibles
whose incomes rise above the Federal or State poverty
threshold level may continue to receive benefits. Finally, the
relatively low discovered error rate obtained in this study was
influenced, in part, by the general methodological framework in
which error was defined. That is, errors were only declared
when there was a clear and documentable violation of State
procedures.
Income Characteristics of WIC Families. WIC households are
primarily low income families with more than one-half below
100 percent of the Federal income poverty guidelines. Almost
all households now enrolled in the program have incomes at or
below the allowable Federal income level (i.e., 185 percent of
poverty). Almost two-thirds of all WIC households have
income from wages and salaries which accounts for three-quarters
of their total income.
State Income Determination. Almost all States agree on the
income sources that should be counted toward WIC eligibility.
These commonly agreed-upon income screes account for the
overwhelming proportion of total WIC household income.
Changes in WIC Household Income. Nearly two-thirds of all
WIC households have verified income that differed from the
income reported at certification by at least $50—a criterion
used frequently in other food assistance programs. Once
certified income-eligible, about one-half of those households
maintain a relatively stable income over the next few months,
XMI
/<*
while the others experience considerable increases or
decreases in countable income. However, these major income
shifts, as well as the substantial differences between reported
and verified income, do not affect program ineligibility rates
anywhere to the same degree.
State Certification Procedures. Self-declaration, a method
used to establish family size and income, was used by a
majority of all WIC State agencies (geographic, Indian, and
territorial) in the fall of 1986.3 Only one-fourth of all State
agencies required documentation at certification. In
geographic WIC State agencies, where the majority of WIC
participants are enrolled, just over one-third of all applicants
must submit income documentation in order to be determined
income-eligible.
State Economic Unit Determination. Federal regulations and
guidance allow for some State discretion in defining "economic
unit." States, in turn, have defined the term in a variety of
ways—legal, economic, biological, social, etc. In part, this is
intended to allow for State coordination among several
assistance programs. The extent of variation from State-to-
State is such that the members constituting a WIC family or
household in one State might not be considered an eligible
family or household in an adjoining State. Furthermore,
specialized definitions and rules regarding eligibility for foster
children, emancipated minors, institutionalized individuals, etc.,
vary considerably from State to State.
if
xlv
Executive Summary Notes
1 Total available FY88 food expenditures were $1,435,363,000.
2P.L 101-147 (10 Nov. 1989) altered program income eligibility. See: Sec. 123(a).
3lndian State agencies account for 28 of the 40 WIC State agencies that use self-declaration as the
primary method to determine income and family size. The large number of Indian State agencies
using self-declaration is due to the fact that WIC Regulations allow these agencies, where the
majority of the Indian households in the agency's service area have incomes at or below the State
agency's income guideline, to obtain FNS approval to use an abbreviated method of self-declaration.
Under this method, an Indian State agency may determine income eligibility by
simply asking each Indian applicant household to sign a statement that certifies that the
applicant's family income does not exceed the State agency's maximum income guideline.
4Based on the preliminary FNS enrollment estimate for October FY88 of 3,777,283.
xv
Ifo
1.
INTRODUCTION
THE WIC PROGRAM
This chapter presents background information on the WIC
program and income eligibility in WIC and introduces the WIC
Income Verification (WIV) project. Data on income eligibility errors
in WIC from prior studies, which were reviewed as part of Phase I
of the WIV project, are briefly summarized. The purpose of Phase
II, the WIV survey, is then discussed. The chapter ends with an
overview of the remainder of the report.
USDA's Special Supplemental Food Program for Women, Infants,
and Children (WIC) is a program for low-income infants, preschool
children, and pregnant, postpartum, and breastfeeding women at
nutritional or medical risk. The program's objective is to improve
the health of participants by providing nutritious supplemental
food, nutrition education, and referral to health care services.
Legislation authorizing the WIC program was passed in 1972,
following the recommendation of the 1969 White House Con-ference
on Food, Nutrition and Health that special attention be
given to the nutritional needs of pregnant women and preschool
children. Originally authorized as a 2-year pilot project, the
program has since become nationwide in scope. In 1974, WIC
served approximately 88,000 participants at an annual cost of
$10.4 million. In 1988, the program served over 3.4 million
participants at an annual cost of $1.7 billion.
WIC is administered by the USDA's Food and Nutrition Service
through FNS headquarters and seven regional offices. The
program operates through cash grants to State health
departments or comparable State agencies. State WIC agencies
distribute funds to participating local agencies, which in turn
operate through local clinics. State agencies bear the
responsibility for developing specific applicant eligibility criteria
and for ensuring that these criteria are applied properly at local
WIC sites. Currently, 1,600 local WIC agencies and 7,500 WIC
clinics are responsible for certifying applicants as eligible for
program benefits.
WIC applicants must meet three types of eligibility criteria:
categorical, nutritional, and income. Categorical eligibility is
limited to pregnant women up to 6 weeks postpartum,
breastfeeding women up to 1 year after childbirth, nonbreastfeed-ing
postpartum women up to 6 months after delivery, and children
from birth until 5 years of age. Within broad Federal guidelines,
each State sets its own nutritional risk criteria based on State
nutrition and health policies. States also have a certain amount of
discretion in establishing income eligibility criteria. Moreover,
1-1
n
applicants must reside within the State (or for Indian agencies,
within the agency's jurisdiction). States may also require
applicants to reside within a local agency's service area.
INCOME ELIGIBILITY IN Although income-eligibility determination always has been a
WIC component of the WIC application process, formal Federal
guidelines were not introduced until July of 1981. Historically,
greater emphasis has been placed on categorical status and
nutritional risk. Even with the introduction of formal guidelines, the
Federal Government has allowed some degree of discretion on
income eligibility procedures for States and local WIC agencies,
and practices at these agencies vary.
Federal regulations in place for Fiscal Year 1989 required the
State agency to set income guidelines either "equaling the income
guidelines ... for reduced-price school meals or identical to the
State or local guidelines for reduced-price health care" (CFR
246.7(c)(1), July 1988). However, if the State chooses to use the
reduced-price health care guidelines for WIC, the WIC income
guidelines must fall between 100 and 185 percent (inclusive) of
the Federal poverty income guidelines.
Both sets of guidelines specify maximum allowable incomes that
vary with family size. Because of this variation, two types of data
are required to determine income eligibility:
Economic unit size: How many persons are in the
applicant's family (also called the household or "economic
unit")?1 and
Economic unit income: What is the total countable
income** of those persons?
Income-eligibility determination in WIC involves identifying all
potential members of the applicant's economic unit. Next, each
person is determined to be a member or a non-member of the
applicant's economic unit. The countable income of each member
is then determined. Finally, the income of the applicant's
economic unit is compared to the income guideline for families of
that size. If the economic unit's income exceeds the maximum for
that size economic unit, the applicant is ineligible for WIC. If the
economic unit's income is equal to or less than the maximum for
that size economic unit, the applicant meets the income-eligibility
criteria for WIC. It is worth emphasizing that knowing the
economic unit's income is not enough to determine whether or not
an applicant meets the income guidelines. The size uf the
economic unit must also be known. This issue is examined in
greater detail in chapter 4.
1-2
11
Within approved boundaries, Federal regulations allow
considerable discretion to State and local agencies in determining
and verifying income and household information. States are
required to design an application form or approve a locally
developed application form for income determination. States may
use self-declaration, in which the State accepts the amount
reported by the applicant as the income of his or her economic
unit, or they may request or require the applicant to provide
documents substantiating the reported income. State and local
WIC agencies may contact third parties to verify income
information but are not required to do so. States may also use a
procedure called "presumptive eligibility." In this procedure, the
State accepts proof of participation in another means-tested
program as evidence that an applicant meets the WIC income-eligibility
guidelines. The requirements are that the other program
is State-administered, has income-eligibility guidelines at or below
the State's WIC income-eligibility guidelines, and routinely verifies
income. In the quality assurance literature, this is known as
"piggybacking" on another program's verification process.
The Administration, Members of Congress, and others with
program oversight responsibilities have expressed their concern
over possible income-eligibility determination error in the WIC
program. A major concern is that income-eligibility determinations
based on WIC applicants' self-declaration may result in
unintentional or deliberate misreporting of income or household
size (or both). In a 1985 report, the General Accounting Office
(GAO) recommended that the Secretary of Agriculture pay greater
attention to income-eligibility determination in the WIC program
and promulgate regulations to require income documentations
The WIV project was USDA's response to these concerns.
THE WIC INCOME
VERIFICATION (WIV)
PROJECT
The WIV project was a two-phase evaluation of participant income
eligibility determination procedures in the WIC program.4 The
study had two objectives:
to inform the Agency of the status of currert income
certification procedures and
to estimate current income certification error on a
nationwide basis, both as a percentage of WIC
enrollees and in total program dollars.
The study was conducted in two phases. Phase I consisted
of several major activities:
10
1-3
the identification and assessment of exemplary
income verification procedures through interviews,
review of written materials (forms, reports, and a
sample of case files), and on-site observation at 10
local agencies in five States;
• a State census of current income-eligibility
determination procedures in all 87 State agencies
(State agencies include Indian agencies and
territories) and
design of a national survey to determine the overall
income eligibility error rate for WIC.
Phase II involved collecting, analyzing, and reporting data on
current income certification error nationwide. This was
accomplished through the WIV survey. Survey interviews
were conducted in September through November of 1988.
The survey employed a nationally representative sample of
884 WIC enrollees selected from 25 States, 42 local WIC
agencies, and 71 WIC clinics. The core of the survey
interview was an in-home audit in which enrollees were asked
for detailed information and documentation relating to their
WIC income eligibility.
This report presents the results of those in-home audits.
SUMMARY OF PRIOR No formal prior research has attempted to develop valid estimates
RESEARCH of income-eligibility determination error rates in the WIC program.
However, there are several sources of data on the self-reported
income of WIC participants. One of the first activities of Phase I
was a review of extant data on income certification error in WIC.
The Census Bureau's Survey of Income and Program Participants
(SIPP), a national in-home survey of 20,000 households, obtains
income data on households that report receiving WIC benefits.
Data for the period from July to September 1983 indicate that
between 9 and 14 percent of the families had incomes above 185
percen: of the poverty threshold.
USDA/FNS'S 1984 Study of WIC Participant and Program
Chaacteristics (PC84), which involved a review of 6,444 WIC
cas'3 files, indicated that very few participants had incomes that
were either above or just below 185 percent of the poverty
threshold.5 This finding is comparable to that of the USDA/FNS
National WIC Evaluation, which found that approximately 80
percent of WIC mothers reported household incomes below
$13,000.6 A re-analysis of these data in Phase I of the WIV
1-4
elO
project found that 78 percent of pregnant women on WIC reported
incomes below the poverty threshold, with 97 percent below 185
percent of the poverty threshold.7
All three of these national studies had methodological limitations
with regard to estimating WIC income certification error:
Income data was from unverified participant interviews or
abstracted from case records.
Across-State variations in income-eligibility criteria,
including differences in the poverty level percentage for
eligibility and the definitions of economic unit and
countable income, were not reflected in the analysis.
The income data were not linked to the period during
which the respondent actually participated in WIC.
The PC84 survey also collected self-reported income data on WIC
participants. The survey estimated that 3.2 percent of participants
have incomes over 185 percent of the poverty level; this is the
income standard used in most States. There are several reasons
to think that this estimate is low. First, because the data were
collected in WIC clinics by WIC staff, enrollees may have felt that
reporting higher income would threaten their benefits. Second,
few steps were used to ensure full and complete income
reporting. Third, no documents were requested to support the
reported income amount.
Furthermore, there is no reason to expect a one-to-one
correspondence between income data obtained by PC84 and that
obtained by WIV from the same respondents. PC84 used uniform
definitions of economic unit, reporting periods, and income for all
survey respondents. In contrast, the WIV survey based its
definitions of economic unit, reporting period, and income on
prevailing State and local eligibility standards and procedures.
Two very limited studies of WIC income-eligibility determination
error rates were conducted during 1985 and 1986. The studies,
which took place in three States, indicated error rates ranging
from 1 to 9 percent. However, these studies were not
methodologically sound, nor were the sample sizes sufficient for
valid statistical inference.8
PURPOSE OF THE The WIV survey estimated the national case income certification
PHASE II SURVEY error rate (excluding Indian and territorial State agencies because
of substantial differences in income-eligibility determination
procedures). The national case income certification error rate is
1-5
the proportion of all cases certified or recertified as eligible for
WIC benefits (in the 50 States and the District of Columbia) that
were ineligible because of income exceeding the allowable
amount as defined by the State program.
Secondary objectives included estimating other income
certification error rates, such as the error rate if the Federal
standard of 185 percent of the poverty level were applied to all
cases, and the error rate partway through the certification period.
The respondents' estimate of their ability to comply with
requirements for income documents and other information as a
condition for participation in WIC was also investigated.
OVERVIEW OF THE Chapter 2 presents a brief overview of Phase I study findings on
REPORT State agency policy and procedures relating to income eligibility
verification and documentation.
Chapter 3 reviews the methodology of the Phase II WIV survey.
Chapters 4 and 5 discuss the key concepts in income certification
error, economic unit size and economic unit income, and they
present the results of the survey relating to the WIC economic unit
and its income.
Chapter 6 gives the main findings of the WIV survey. Income
certification case and dollar error are presented, and correlates of
error, error in relation to other Federal programs, and the effects
of changes in income are discussed.
Chapter 7 covers the reported ability of WIC enrollees to conform
to alternative procedures for income certification. Chapter 8
summarizes the study's conclusions. The report concludes with
technical appendices on sampling, the WIV survey questionnaire,
references, a substudy that attempted to reconcile information
gathered at certification with information from the WIV survey, the
Federal regulations and guidance relevant to income certification,
alternate definitions of income-eligibility error, and general
definitions of WIC Economic Unit.
1-6
Chapter 1 Notes
1 "Economic unit" was defined in FNS Instruction 803-3 as "a person or group of persons who usually
(although not necessarily) live together, and whose production of income and consumption of
goods or services are related" (10/22/82, p. 3). The instruction suggests that the terms "economic
unit" and "family" be used interchangeably. The term "household" is also used interchangeably
with the other two terms in this report.
2Note that, in the WIC program, countable income does not include assets or in-kind income.
3U.S. General Accounting Office, Need to Foster Optimal Use of Resources in the Special
Supplemental Food program for Women, Infants and Children (WIC), September 27,1985,
GAO/RCED-85-105, pp. 67-68.
4lncome-eligibility determination, as part of the process in WIC by which eligibility for benefits is
certified, is also known as "income certification."
5Ebon Research Systems, Inc., Study of WIC Participant and Program Characteristics, 1986.
6Research Triangle Institute, Evaluation of the Special Supplemental Food Program for Women,
Infants, and Children (WIC), 1986.
7Review of Current Income Eligibility Determination Procedures and Reported Participant Income
Levels, Quality Planning Corporation, March 1987.
8Each study was limited to a single State, did not employ a statistically representative sample, and
failed to provide an independent and valid method for verification of total household income. For
details, see WIC Income Verification Project: Final Literature Review, Quality Planning
Corporation, January 1987; The New Enrollees'Income, Department of Health and Rehabilitative
Services, State of Florida, 1986; and The Client Summary by Financial Category and Priority,
Department of Health and Environmental Sciences, State of Montana, 1986.
1-7
^3
2.
PHASE I
ACTIVITIES AND
FINDINGS
INCOME VERIFICATION
METHODS CURRENTLY
USED IN WIC
A primary Phase I activity was the WIC State Agency Census of
income verification methods currently in use. The census
involved telephone interviews with the WIC Director or other staff
at each of the 87 WIC State agencies.1 A copy of the State's
WIC income certification forms and the section dealing with
income and household size from each State agency's procedure
manual was also requested.
As explained in chapter 1, the income-eligibility determination
process in WIC includes determination in WIC clinics of economic
unit membership (and thus economic unit size) and economic unit
income, and comparison of the applicant's economic unit size and
income to the income-eligibility standard. The State Census,
however, focused on another aspect of income-eligibility
determination, namely income verification. Income verification
refers to procedures for confirming that the income reported by
the applicant is correct.
The results reported in this chapter are at the State level. That is,
the percentages and numbers presented represent WIC State
agencies, not WIC enrollees. Income verification procedures are
covered on the enrollee level in chapter 7, while the States'
procedures for determining economic unit membership and
income are covered in chapters 4 and 5, respectively.
The process by which a WIC local agency or clinic confirms that a
household's reported income meets its eligibility standard is its
income verification procedure. Each State agency was asked to
describe its policy regarding the procedures to be used by WIC
clinic staff in its jurisdiction. Based on the responses, four main
policies regarding income verification were identified. In order of
increasing rigor, these are:
Self-declaration. Applicants report their income and
household size to a clinic staff member. The applicant
then signs a form showing the reported information and
stating the consequences of misreporting. All local
agencies and clinics throughout the State use this
procedure; the State does not allow the use of other
procedures.
Local variation. The State policy is self-declaration.
Local agencies have the option of using more rigorous
techniques, and in some cases the State encourages
them to do so.
JM
2-1
Documentation requested. Applicants are asked to
provide documents showing their household's income,
such as a paycheck stub or an Aid to Families with
Dependent Children (AFDC) award letter. If an applicant
fails to bring income documents to the certification
interview, self-declaration is used. In some States, the
applicant is asked to bring the documents within 30 days
after certification; the applicant is provided with benefits
for 30 days, and, in some cases, benefits are then
terminated if no documents have been produced. In other
States, the applicant is asked to bring documents later,
with no deadline given; and in still others, the clinic does
not pursue documentation once the applicant is certified.
Documentation required. Applicants are told to bring
income documents to the certification interview. An
applicant who does not bring documents is not certified.
The documentation-required category can be divided into
two subcategories: application pended and application
not pended. "Application pended" refers to the practice of
completing all certification requirements other than
documentation, and placing the application in a pending
status. The applicant has only to bring in the documents
to be certified. If the application is not pended, the
applicant must make another appointment for the
certification interview; all certification procedures are
carried out at the second appointment.
According to the State Census conducted in 1986, then, income
verification in WIC primarily involves two techniques: self-declaration
and documentation. Local agencies also use a third
technique, contact with a third party (such as the employer or
social worker) when incorrect reporting of income by the applicant
is suspected; however, States reported that the time required by
this method restricted its use to suspect cases. Other techniques,
such as computer matching of WIC enrollment records with wage
records or those of other agencies, are occasionally used, but not
on a regular basis.
Variations occur within the four categories identified above. For
example, many Indian agencies used an abbreviated method of
self-declaration. The abbreviated method is a special procedure
allowable under WIC regulations only in Indian agencies that
serve a predominantly poor population. Under the abbreviated
method, a WIC agency may determine income eligibility by simply
informing applicants of the income cutoff point for their size of
household and asking whether their household income is below
that point; specific income information is not required.
2-2 If
Another variation is the application of more rigorous techniques to
suspect cases. Only documentation-required States insist on
documentation for all applicants; however, in any State, a
participant may be required to provide documentation of
household income in specific cases where clinic staff have reason
to suspect misreporting. Third-party contacts, such as calling the
Food Stamp office to verify the applicant's household income, are
occasionally used in suspect cases.
Finally, documentation-requested States pursue documentation to
different degrees; some make a strong effort to obtain
documentation in all but the most difficult cases, while others use
self-declaration routinely after an initial attempt to document
income.
The State census revealed that the predominant income
verification practice is the use of self-declaration, followed by
documentation required. Among the geographic State agencies,
documentation required is the most common method of income-eligibility
determination.
Figure 2-1 shows the distribution of the primary income
verification methods across the 87 WIC State agencies. As can
be seen from the figure, self-declaration was the primary income
verification method for 40 of the 87 WIC State agencies. The next
most common method, used by 22 State agencies, was to require
documentation. Local variation and documentation requested
accounted for 14 and 11 State agencies, respectively.
The income verification method varied with the type of State
agency (see figure 2-2). Indian agencies accounted for 28 of the
40 WIC State agencies th?. used self-declaration as the primary
method.
Two of the territories used self-declaration, and one required
documentation.
Among the geographic State agencies, which serve the
overwhelming majority of WIC participants, required documenta-tion
was the most common method. Documentation of income
was required in 19 of the 51 geographic State agencies. Local
variation was found in 14 geographic State agencies, self-declaration
in 10, and documentation requested in 8.2
None of the Indian agencies or territories showed local variation in
the primary income verification method; only the geographic
2-3
4fc
IED Method
Figure 2-1. Primary income verification method.
£7
2-4
Geographic
States
D Self-declaration
Local variation
Documentation requested
Documentation required
Indian
Agencies
rm±-\
Territories
Figure 2-2. Income verification method by type of State agency.
JL<d
2-5
States had some iocal agencies that used self-declaration and
some tha* requested or required documentation. Of the 14 local
variation States, 7 provided estimates of how many of their local
agenc s used self-declaration. These estimates, which ranged
from 5 percent to 90 percent, indicate the degree of variation
within the local variation category. This group of seven States
was considered to have less rigorous income verification practices
than the States that request documentation statewide; in five of
the seven, half or more of the local agencies used self-declaration
only.
SUMMARY AND
IMPLICATIONS FOR
PHASE II
Phase I of the study found substantial across-State variation in
income verification procedures. While self-declaration
predominated as the primary method in the 87 State agencies,
documentation required was the most common method used in
the 51 "geographic" State agencies (those representing the 50
States and the District of Columbia) that serve the overwhelming
majority of WIC enrollees. It was not possible in the course of this
study to directly examine case-file data nor to "reconstruct" the
decision process at the clinic level where eligibility was actually
determined. Therefore, Phase II of the study was only able to
verify total size and income of the economic unit. As a result, it
was not possible to provide data on misreporting^ by income
source nor on the local agency's grounds for determining
household size.
2-6
Ad
Chapter 2 Notes
1The 87 State agencies include the 50 States plus the District of Columbia for a total of 51
"geographic" State agencies, plus 3 territories and 33 Indian agencies.
2By the time of the WIV survey, one of the local variation States had switched to requiring
documentation. While this project did not collect 1989 data on State policies, we recently learned
that another State had changed from local variation to documentation required.
3Misreporting is the incorrect reporting of income or household size by the applicant; it may be either
accidental or deliberate.
2-7
Jo
3.
SURVEY
METHODOLOGY
DATA COLLECTION
METHODS AND
SCHEDULE
During the latter part of Phase I of the WIV project, the design for
Phase II was modified to coordinate data collection efforts with
those for the 1988 Study of WIC Participant and Program
Characteristics (known as PC88).1 PC88 sampled and collected
data on WIC State agencies, local agencies, and participants;
WIV added questions to the PC88 State and local agency
questionnaires and the participant case-file abstraction form, and
conducted in-home audits of a subsample of the PC88 participant
sample. These in-home audits comprised the WIV survey.
The WIV survey was an in-home survey of persons certified as
eligible for WIC during April to July 1988. The purpose of the
survey was to estimate the national error rate in income-eligibility
determinations, known as the "income certification error rate."
The survey was designed to overcome the deficiencies noted in
the review of prior research. Specifically, the WIV survey used:
A nationally representative sample of WIC enrollees.
A detailed in-home interview, with documents requested
for all income sources. This in-home audit approach has
been shown to produce more accurate and complete
income data than a simple self-report procedure.
The same income time period used by the WIC clinic to
determine income eligibility.
The same set of rules on countable income and
household size used by the WIC clinic to determine
income eligibility, as mandated at the State level. This
was crucial, since State agencies' rules vary considerably
within the limits set by Federal regulations.
Data from both PC88 and the WIV survey are presented in this
report. This section lists the various data sources and time
periods in which data were collected. PC88 consisted of five
primary data collection efforts:
1. SAQ (State Agency Questionnaire), a mail survey of 87
State agency WIC directors.
01/88
2. LAQ (Local Agency Questionnaire), a mail survey of
220 local agency WIC directors.
04/88
31 3-1
3. PQ (Participant Questionnaire), interviews of 6,618 WIC
enrollees conducted by local agency staff immediately
after certification.
04/01/88 to 07/27/88 (with data collection complete at
most sites by 06/30/88)
4. PRAF (Participant Record Abstraction Form), initial
and follow-up record abstractions of the 6,618 enrollees'
files by local agency staff. PRAF provided the data used
in the WIV Project on reported income and household
size.
04/01/88 to 07/30/88
5. Benefits Issuance Data, data from the local agency on
benefits issued to the 6,618 sampled enrollees over the 6
months following certification or recertification. Benefits
were defined as food vouchers issued, though not
necessarily redeemed.
10/03/88 to 12/30/88
In addition to the five primary data collection efforts, PC88
generated a wide variety of other data files. These include
sample weight files at the State, local, clinic, and enrollee levels;
response probability flies at the local level by category of
participation; and total enrollment data at the State level by
participant category and priority level. These files were used by
WIV to develop sample weights.
The WIV project undertook five main data collection efforts:
1. Census of State Income-eligibility Determination
Procedures, described in chapter 1.
11/01/86 to 12/18/86
2. WIV In-home Audits. In-home audits of 884 of the 6,618
PC88 sampled enrollees were conducted. These in-home
audits were the main income verification technique
employed. The in-home audits verified income data
collected on the PRAF for the same time period, and also
collected income data for the month preceding the in-home
audit.
10/05/88 to 11/15/88
3. State Verification of Income-eligibility Determination
Standards and Procedures. Sample States reviewed
and corrected summaries of their income-eligibility
determination standards and procedures.
10/17/88 to 12/05/88
3-2
3^
4. Income Reconciliation Substudy. When errors in
income-eligibility determinations were found, enrollees
were contacted by telephone to reconcile discrepant data.
01/16/89 to 02/14/89
5. Contamination Substudy. To evaluate the possibility
that linking with PC88 could bias WIV in-home audit
results, a special substudy was conducted. The substudy
conducted in-home audits of a matched sample of
enrollees who had not been involved in PC88, and
compared the results to those for the main WIV sample.
01/24/89 to 02/24/89
Several elements of the study time-line require comment. First,
the census of State income-eligibility procedures conducted in late
1986 is now a few years old. We know of two States that have
changed their policies in the intervening period; there may have
been more. Second, in-home audits took place an average of
about 5 months after enrollment in WIC. The delay was
necessitated by conditions of the merger of the two studies. We
have no information on how much the delay resulted in
deterioration of the quality of the in-home audit data nor on our
ability to locate potential respondents. Memory problems may
have masked or distorted some results.
IN-HOME AUDIT Figure 3-1 shows a map of the State and local agencies selected
SAMPLING PROCEDURE to participate in the WIV survey. The WIV sample consisted of a
subsample of respondents to PC88.
The PC88 sample was constructed in four stages using a nested
design. A first-stage sample of 28 State agencies (25 non-Indian
and 3 Indian) was selected. Within the sampled States, a sample
of 220 local agencies (217 non-Indian and 3 Indian) was selected.
One or more certification sites was then selected from each
participating local agency. Finally, a sample of 7,040 WIC
enrollees was selected, recruited, and interviewed by local agency
staff.
With minor variations, the PC88 enrol.ee sample consisted of
individuals certified or recertified in sampled clinics from April
through June 1988; at a few sites, data collection started later and
continued into July. The sampled enrollees were interviewed, and
data were abstracted from their case files2
The WIV survey subsample was exclusively drawn from PC88
respondents. All 26 geographic PC88 State agencies were
3-3
Figure 3-1. WIC Income verification survey sample States
and local agencies
3q
included in the WIV subsample. However, the Indian agencies
were excluded because of substantial differences on income-eligibility
issues. Within the sampled States, WIV subsampled 41
of the 220 PC88 local agencies. Within each of the WIV local
agencies, one or two PC88 clinics were sampled for a total of 72
clinics. Finally, within the sampled clinics, all PC88 respondents
who agreed to allow follow-up data collection were sampled for
WIV in-home audits.3 The final WIV sample contained 25 States,
41 local agencies, 72 clinics, and 1,076 participants.
IN-HOME AUDITS In-home audits are personal interviews combined with income
documentation reviews. The in-home audits were conducted in
the program recipients' homes by professional interviewers who
had experience in conducting income studies. The interviews
covered four primary topics: income, household composition,
demographic characteristics of participants, and ability to respond
to income verification requirements. The data from the in-home
audits were used to determine each sampled enrollee's income
eligibility (see the section titled "Operational Definitions of Major
Variables" in this chapter for details).
To help ensure a high response rate, a variety of data collection
steps were takon. First, an introductory letter requesting an
interview and explaining the study was mailed to each
respondent. The letter was designed to acquaint the respondents
with the significance of the study, to assure them that confiden-tiality
would be maintained, and to inform them that their
participation in WIC would not be affected by information obtained
during the interview. Accompanying this letter was a list of the
types of income-related documents that the respondent would be
asked to show the interviewer during the in-home audit.
WIC enrollees were then contacted by telephone to schedule
appointments for personal in-home interviews. Up to five attempts
were made to contact respondents. To maximize the coverage of
the calls, one call was made during each of the following periods:
weekday evening, weekday afternoon, weekend, and weekday
morning. If telephone contact could not be achieved, up to three
visits were made to the respondents' homes to schedule an
appointment. Like the telephone calls, the visit periods were
staggered. Once contact was made, an appointment was
scheduled at the respondent's convenience.
The interview included a number of questions designed to identify
and characterize all potential members of the enrollee's economic
unit. During the interview, respondents were also asked to
answer a detailed set of questions that required that they report
^
3-5
the presence or absence of income from a long list of potential
income sources, and the amount of income from each source for
each adult household member. For every income source
mentioned, the respondent was requested to supply documentary
support. Examples of documentary evidence requested included
check stubs and program eligibility certificates. Income
information was collected for two time periods. The first was the
time period used to determine eligibility at the time of certification
(the financial data reference period or TDRP" income). The
second period was the month before the in-home audit; this was
referred to as "current income."
Requiring respondents to report separately for each adult
household member and for each potential income source typically
reveals income that would not otherwise be reported. For
example, when asked to report all household income, many
respondents do not include such income as their grandmother's
pension; they may not consider this part of the household's
income because it is not the major source of income and may not
be shared. Detailed income reporting prevents such omissions.
Further, there is evidence that respondents may be more willing to
commit sins of omission and simply not mention an income
source, but are often unwilling to commit sins of commission and
lie when asked if an individual has a particular source of income.
Finally, the in-home audit method prevents respondents from
using their own understanding of which types of income are
considered countable by the WIC program and should be
reported.
After data on all potential income were collected, he interpretation
of what income to count was made by a senior project staff
member familiar with program regulations in the participant's
State. This same individual determined which persons were
actually members of the enrollee's economic unit, in accordance
with State and Federal policies.
SURVEY YIELD Sampling weights were used in analyzing the WIV survey data to
account for the varying probabilities of selecting a State, a local
agency, or an enrollee into the sample. Generally, a sampling
weight is equal to the reciprocal value of the probability that the
sampling unit is included in the sample. The WIV sampling weight
may be thought of as the number of enrollees in the population
that the sampled enrollee represents. The weights allow us to
make valid inferences about the entire WIC enrollee population.
Table 3-1 presents the basic sample design (further details are
provided in Appendix A).
3-6
3h
Table 3-1. Summary of All Stages of the Sample Design
Symbol Selection Level Random
Selection
Mechanisms
Data
Source
PL|S
PWL|L
PA|WL
PWA|A
PE|WA
*R|i
WR|R
AV|WR
IHA|AV
WEP
*SR
Probability of Selection
of State Agency
Probability of Selection
of Local Agency Within a
Selected State
Probability of Selection of
Local Agency by WIV Given
Selection of the Agency by
PC88
Probability of Selection
of a Site by PC88 Within a
WIV Selected Agency
Probability of Selection
of a Site by WIV Given the
Site was Selected by PC88
Probability of Selection
of Enrollee in Selected
Site
Probability of Completed
Response to PC88 Given
Enrollee Selected by PC88
Probability of Completed
Response Is Forwarded to
WIV by PC88
Probability That PC88
Respondent Could Be Lo-cated
by Time of WIV
Probability of Completed
In-home Audit Given WR
Weight of Enrollee to
Reflect Level of
Participation
Ratio Adjustment to State
and Regional Participation
Totals
N
N
N
N
NA
NA
Probability Propor-tional
to Enrollment
PC88 Sample
Weight File
Probability Propor-tional
to Enrollment
PC88 Sample
Weight File
Constant Probability
and Random Selection
WIV Sample
File
Probability Propor-tional
to Enrollment
PC88 Sample
File
Constant Probability
and Random Subselec-tion
WIV Sample
FHe
Random Probability
Conditioned on
Enrollment Cate-gory
and Sequence in
Selection Period
PC88 Site
Probability
File
Unknown; Probabil-ities
Determined at
the Local Agency
by Participant
Category
PC88 Local
Agency
Reports
Unknown; Sources of
Losses Include Re-fusal
to Allow
Follow-up, State and
PC88 Failure to
Forward; Probabilities
Determined at Local
Agency Level
PC88 Case
Tracking
Files
Unknown; Determined
by Individual
Respondent Life
Conditions
WIV Case
Tracking
File
Unknown; Nonresponse
Analysis Found no
Systematic Biases
WIV Case
Tracking
File
Based on Observed
Number of Months
Sampled Enrollee
Participated
PC88 Fol-low-
up Files
Ratio Adjustment by
Participant Category
Federal
Partici-pation
Files
37
3-7
Final sample weights were the cross-product of table elements.
Four elements of the table (PRJE, PWR|R, PAV|WR, and PIHA|AV)
represent nonrandom probabilities of survey response. Because
the values associated with these probabilities are not random,
issues of nonresponse bias must be addressed. The response
rates for each of these values were generally high. The overall
response rate for PC88 was PR.E . 0.86. Of those responding to
PC88, over 90 percent were included in the WIV sample base,
PWRIR " °-91 • 0f the w,v samP|e Dase>over 9° P^cent could be
located for the WIC survey, PAv|WR m°92- RnaNy.tne WIV
survey itself had an 85-percent response rate for P!HA|AV ■ °-86
While the response rates were high at all sample stages, there is
reason for concern about the cumulative effect of the multiple
points of sample loss. The cumulative e' ct was a low total
response rate. The total response rate Cc*n be defined as the
product of the four terms: 0.86 * 0.91 * 0.92 * 0.86 - 0.62.4
Table 3-2 presents total response rates by WIC participant
category.
Sample weights were developed in four stages: (1) analysis of
response bias, (2) development of sampling probability weights,
(3) development of ratio adjustments, and (4) correction for length
of participation.
First, a response bias analysis was conducted. We were
concerned that systematic patterns of nonresponse might bias the
survey results. For example, individuals who misreported their
income to WIC might be more likely to refuse to cooperate in the
survey than individuals who correctly reported their income. If this
were the case, the survey results could underestimate the level of
income certification error. To check for such potential biases we
conducted four tests.
First, we analyzed reasons for nonresponse. Overall, 15 percent
of those sampled did not respond. However, only 3.7 percent of
those sampled did not respond because they refused to be
interviewed. Another 2.1 percent of those sampled avoided being
interviewed without refusing.5 The remaining 9.2 percent of
nonrespondents were not interviewed because they could not be
located (6.6 percent), were i!i or otherwise unavailable for the
interviewing period (0.7 percent), were not at home after repeated
calls (1.0 percent), or for other reasons (0.7 percent). This
analysis showed a very low rate of refusal or avoidance of the
interview.
3-8
2i
Table 3-2. Total Survey Response
Rates by Participant Category
Category Response Rate
Pregnant Woman 0.52
Breastfeeding Woman 0.70
Postpartum Woman 0.61
Infant 0.66
Child 0.58
Total 0.62
3-9
^3
Second, we compared respondents with nonrespondents on data
from WIC files that were available for all individuals sampled.
These variables included participant category, reported income,
reported household size, local agency, and date of certification.
We could detect no statistically or substantively significant
difference between respondents and nonrespondents on available
data.
Third, using WIV in-home audit data, we constructed an equation
that predicted the probability of income inedibility. The equation
was based on predictor variables obtained from the case-file
abstracts, which were available for all sample members.6 Using
this equation, we then predicted the ineligibility rate for both
respondents and nonrespondents. Again, no difference was
detected.
Fourth, we compared the survey response rate for clinics with
high ineligibility rates to those with low ineligibility rates. If
ineligible enrollees were more likely to refuse interviews, we would
expect a higher nonrepsonse rate in clinics with high ineligibility
rates. We found no correlation between clinic ineligibility rates
and survey response rates.
These four tests led us to conclude that survey nonresponse is
very unlikely to have biased our estimates of WIC income-ineligibility
rates.
The second step in weighting the sample was to develop survey
design weights. These weights assigned to each respondent a
weight equal to that respondent's probability of se'action into the
sample. Basically, these weights are equal to the inverse of the
probability of selection of a State times the probability of selection
of a local agency within that State times the probability of
selection of a clinic within the local agency times the probability of
selection of an enrollee within the clinic. Within this basic
structure, additional adjustments were made to account for
nonresponse by weighting to total selected sample size by clinic
and participant category (see table 3-1).
The third step in developing sample weights was to adjust the
total weights so that the national enrollment estimates were
consistent with the best available data. Such adjustments, known
as ratio estimators, help improve the quality of the estimates and
reduce sample enor.' Using ratio estimators, we conected the
sample weights to reflect FNS regional WIC enrollment reports
from March 1988.
3-10
The fourth step in developing sample weights was to correct for
period of participation. As noted above, the basic sample design
selects WIC enrollees. However, not all enrollees participate for
the same period of time. Therefore, for the sample to adequately
mirror WIC participants, it is necessary to weight it so that an
enrollee who participates for 6 months, for example, is weighted
twice as high as an enrollee who participates only for 3 months.
Estimates of participation are needed for calculating the dollar
value of benefits provided to enrollees who were erroneously
certified as income-eligible. To develop these participation
weights, PC88 obtained data on the number of months for which
sampled enrollees participated in the WIC program during their
certification period. These data, in turn, were used to develop
weights that converted enrollment estimates into participation
estimates.
The use of sampling weights introduces the possibility that a few
highly weighted cases may significantly influence survey results.
Two measures were taken to detect and correct this potential
problem. First, an analysis was conducted of the influence of
weights on the survey results. On the average, weights
accounted for less than 5 percent of the variance on survey
variables. Second, all analyses were conducted with both
weighted and unweighted data. When any findings were
discovered to depend on a few highly weighted cases, categories
were collapsed to aid in stabilizing results. When statistically
stable results were not obtainable, no finding is presented in this
report.8
SURVEY LIMITATIONS
Sampling Variance The results achieved from all sample surveys are subject to
sampling error. Sampling error is defined as the difference
between the results that would have been obtained had the entire
WIC enrollee population been surveyed and the actual results
obtained from the sample. Appendix A contains a detailed
discussion of this source of error.
Measurement Error In addition to sampling error, survey results are potentially subject
to a variety of measurement errors. The in-home audit procedure
is a major source of potential error. A respondent can either
refuse to cooperate or deliberately underreport income. Extensive
efforts were made to minimize the number of refusals. Also,
rather elaborate measures were undertaken to minimize the
incentives and opportunities for underreporting A major incentive
for underreporting was the threat that the income information
would reach the WIC clinic and result in the loss of WIC benefits.
We removed this incentive for underreporting income by
3-11
hi
guaranteeing confidentiality and that survey results would not
affect WIC benefits; this guarantee was given both verbally and in
writing. Respondents were also requested to sign an agreement
form, which committed them in writing to providing full and
accurate information. The agreement said, "I understand that the
information that I provide must be accurate in order to be useful. I
agree to give responses that are complete to the best of my
knowledge. I understand that any information I give will be kept
private and confidential." The interviewer then signed the second
part of the agreement, which stated that she would keep all
information reported by the respondent confidential. This sort of
informal contract gives the respondent an incentive for correct
reporting of income. Research has shown that such agreements
substantially increase the complete ness and accuracy of survey
data.10 Ninety-five percent of tne respondents signed the
agreement form. The 5 percent who did not sign was too small a
subgroup to permit meaningful statistical comparisons with the
total sample.
Finally, the in-home audit questionnaire was designed to minimize
the opportunity for underreporting. Income information was
collected separately for all adult household members. For each
adult, respondents were asked if that individual had any of 28
specific sources of income. This data collection technique
minimizes forgotten income sources and varying understandings
of what constitutes income and whose income should be reported.
Further, with this method any misreporting must be an act of direct
commission and not simply omission. Extensive probes were
used to discover income in cases where individuals were reported
to have no means of support.
A second source of potential measurement error arises from the
link between the WIV study and PC88. Because the WIV in-home
audits were based on a subsample of PC88 clinics, clinic
workers knew well in advance that income information might be
verified. This knowledge could have caused them to be more
diligent or careful in collecting and recording income information
than they might have been otherwise. We know that one State
WIC program reviewed all PC88 clinic certifications before the in-home
audits began. As a precaution against the potentially large
biasing effects of the link with PC88, a special substudy was
conducted. In this substudy, we conducted in-home audits with a
sample of 44 enrollees not subject to PC88 interviews. The
substudy sample was drawn from two PC88 clinics and matched
to the PC88 sample in those clinics. Statistical analysis of the
matched samples found no statistically nor substantively
significant differences between respondents subject to PC88
interviews and respondents not included in PC88.11
3-12
4^
The third possible source of measurement error concerns coding
of eligibility by WIV project staff. Multiple measures were taken to
prevent such errors. Briefly, each case was coded using the State
rules in place at the time of enrollment; our interpretation of the
rules was verified by State WIC staff; and all error cases were
independently validated in three ways: a computerized eligibility
cross-check, a reconciliation with abstracted WIC case-file data,
and a reconciliation with the enrollee.12
OPERATIONAL DEFINI-TIONS
OF MAJOR VARI-ABLES
Applicant/Enrollee/
Participant
Certification/
Recertification
An applicant for WIC is a person who is applying to receive WIC
benefits. If the applicant meets the eligibility requirements and the
program's budget is sufficient to allow the provision of benefits to
that applicant, the applicant is enrolled in the program and
becomes an enroiiee. A participant is an enrollee who receives
nutrition education, referrals to other health programs, and/or
supplemental food.
In this study, we generally use the term "enrollee" to refer to
persons in the WIV sample. By this term, we mean any sample
member who was certified as eligible to receive WIC benefits,
regardless of whether or not he or she actually received benefits.
When we use the term "participant," we refer to an individual who
was both certified for WIC and received benefits. While there are
often other WIC enrollees or participants in the household, the
terms will refer specifically to that person selected for the WIV
survey.
Certification is the process by which the applicant for WIC benefits
becomes an enrollee. Applicants must meet categorical,
nutritional, and income-eligibility criteria in order to be enrolled in
the WIC program. In addition, applicants are required to reside
within the jurisdiction of the State agency (or, for Indian State
agencies, within the agency's jurisdiction).
The certification process varies across State and local agencies.
Although each State operates a WIC program, there are still a few
areas within States that are not served by a local WIC agency.
Persons residing in a WIC service area may make an appointment
at a local WIC clinic to be screened for program eligibility at no
cost to the applicant. At the clinic, the applicant completes an
application form and undergoes a brief interview in which income
eligibility is determined. Categorical eligibility, residence in the
agency's jurisdiction, and nutritional risk are determined during the
2(3
3-13
same appointment. If the applicant meets all three criteria and the
clinic has the capacity to enroll all eligible applicants, he or she is
certified. Eligible applicants in categories to which a low priority
has been assigned (e.g., postpartum nonbreastfeeding women)
may not be certified, if the clinic lacks the capacity to serve all
eligible applicants.
WIC participants are usually certified for a period of 6 months.
However, the intervals used for WIC certification vary somewhat.
For example, a pregnant woman is certified for the duration of her
pregnancy and up to 6 weeks postpartum. At the State agency's
option, infants may be certified for a 1-year period extending up to
the first birthday, provided that the quality and accessibility of
health care services are not diminished.
Respondent
At the end of the certification period, the enrollee may be eligible
for continuing benefits. To receive those benefits, the enrollee
must again be certified as eligible on all three eligibility criteria;
this process is called recertification, and usually involves the same
procedures as does certification.
The respondent to a survey is the person who is interviewed. In
the case of woman enrollees, the selected enrollee is also the
respondent. However, many WIC enrollees are infants and
children and cannot answer the survey questions. An adult
respondent was selected to answer for these enrollees. This
respondent was in most cases the person who brought the infant
or child to the WIC clinic for certification or recertification (usually
the mother). In the case of infant and child enrollees, the
respondent was not automatically part of the enrollee's economic
unit. For example, if the economic unit consisted solely of a foster
child and the respondent was the foster mother, the respondent
was not in the child's economic unit.
Some of the variables measured by the survey apply to the
enrollee, such as type of health insurance. Some are qualities of
the enrollee's household, such as household size and income.
Other variables apply to the respondent, such as how he or she
heard about the WIC program and whether he or she was asked
for income documentation.
Economic
Unlt/Family/Household
The terms "economic unit," "family," and "household" are used
interchangeably in WIC. The Federal regulations define a family
as a "group of related or nonrelated individuals who are not
residents of an institution but who are living together as one
economic unit."1^ An economic unit is generally defined as a
group of persons whose production of income and consumption of
goods and services is related.14
3-14
Uu
Defining the applicant's household is an essential step in
determining income eligibility. The number of household
members defines which maximum income applies, since the
Federal poverty guidelines (on which the WIC income guidelines
are based) set different maximum incomes for different family
sizes. Also, the household's income is the sum of the income of
its members; therefore, including or excluding a particular person
can affect both household size and household income.
Many States have developed detailed policies to deal with
questions of what persons should be included in the applicant's
household. In addition to general policies, many States have
policies that apply to special cases, such as pregnant teenagers,
institutionalized family members, foster children, children under
joint custody arrangements, and college students living away from
home.
The definition of an economic unit is discussed in detail in
chapter 4.
Income The household's income for income-eligibility determination
purposes is current gross cash income. Certain types of income
are considered countable and included when determining income,
while others are excluded. In addition to in-kind income, Federal
regulations specifically exclude "payments or benefits provided
under certain Federal programs or acts by legislative
prohibition."15 Many States also have specific policies listing
excluded and included forms of income. Thus, for each
household member, income from each countable source must be
determined. To make this a bit more complex, in some States
part of the income from certain sources is countable and part is
not.
Federal regulations also provide the State and local agencies with
discretion in deciding whether current income or income during
the past 12 months best represents the family's status.16
Current income is not easily defined. The period of time that is
thought to best represent current income varies across State and
local agencies as well as across applicants. Perhaps the most
common time period used is the calendar month prior to the date
of certification. States were encouraged to use this period in the
FNS Instruction 803-3, Revision 1, of April 1,1988. Since
applicants' household income may vary over time, it is important
to know what time period was used by the local agency when
income eligibility was determined.
3-15
Financial Data Reference
Period (FDRP)
Income-eligibility
Standard
The definitions of income used in the WIC program are expanded
upon in chapter 5.
"Financial Data Reference Period" is a term that was developed
for use in this survey. It is defined as the time period for which
income was assessed in the income-eligibility determination
process. In other words, the FDRP is either the 12 months prior
to certification or the local agency's operational definition of
"current" for determining the household's current income.
We used the FDRP established by the local agency for a
particular case wlierever possible. When abstracting the case file
for PC88, local agencies were to record the FDRP used.
However, because the record abstraction took place after the
certification and the FDRP is not generally recorded in the case
file, the FDRP was not available in 68 percent of the cases. In
these cases, the local agency's policy for assigning an FDRP, as
reported in the PC88 local agency survey, was followed.
When an FDRP was based on local agency policy and an error
was found, the FDRP was verified with the respondent through an
individual telephone follow-up call. For details on the procedures
used, see appendix D.
WIC State agencies are of three types: geographic (i.e., the 50
States and the District of Columbia), Indian, and territorial.
Because of significant differences in populations served, income-eligibility
determination procedures, and other characteristics
across these three groups, a decision was made to limit the WIV
survey to the geographic States.
Each State agency sets its own income-eligibility standard;
however, Federal regulations require the standards to be between
100 and 185 percent of the Federal poverty guidelines set by the
Department of Health and Human Services (DHHS).17 Poverty-level
income varies with the size of the household. To allow for
changes in the cost of living, the poverty level is revised yearly
effective July 1.
In 1986, 75 percent of the geographic State agencies used 185
percent of the poverty level as the income-eligibility standard. A
few States allow local agencies to set their own income-eligibility
levels within the Federal limits. States also establish policies on
how income and household size are to be defined. Revised
policies are issued as needed.
Most local agencies use a procedure known as "piggybacking" or
"presumptive eligibility" for some applicants. In this procedure,
Ub
3-16
Income-ellglblllty Error
INCOME CERTIFICATION
ERROR DETERMINATION
PROCEDURE
the income-eligibility determination procedure, followed by a
program with a standard equal to or lower than that for WIC, is
assumed to be accurate. WIC then "piggybacks" on that
program's eligibility determination. The clinic does not determine
household size and income, then, for an applicant who shows
proof of participation in a program such as Medicaid or AFDC; the
applicant is certified based on participation in the other program.
We defined two primary types of eligibility error. First is
certification error. The certification error rate is the percent of WIC
enrollees who were enrolled in WIC in violation of the prevailing
income-eligibility standards.18 Second is dollar error. The dollar
error rate is the percent of WIC food dollars that are spent on
enrollees who are ineligible.
In appendix F, other types of case and dollar error definitions are
considered: error rates using Federal rather than State
standards, error rates based on case-file abstraction data from the
PC88, and error rates based on income changes during the
certification period (using income from the calendar month before
the in-home audit to estimate income change).19 These have
been compared with the main error rates.
Determining income eligibility for the enrollees in the WIV sample
required constructing operational and measurement models. The
income-eligibility determination process was modeled for each of
the State agencies in the sample, and any focal agency variations
were also noted. The model was then applied to each enrollee
sampled from that State or local agency to produce an
independent determination of income eligibility based on data
from the WIV in-home audit. An income certification error was
assessed for each enrollee for whom this independent
determination concluded that the enrollee was not eligible.
The independent income-eligibility determination followed a 10-
step procedure:
1. Identification of Variables Used In Income eligibility
Determination. Based on WIC State plans and
procedure manuals, project staff identified all variables
currently used by WIV sampled States in the income-eligibility
determination process. These variables were
incorporated in the in-home interview.
2. Chart and Annotate State Income-ellglblllty
Operational and Measurement Models. In this step,
we produced detailed flowcharts of State procedures.
These flowcharts specified operational flows and
3-17
*0
measurement tests that permitted assignment of any
given applicant to either the income-eligible or the
income-ineligible category.
3. Provide Flowcharts and Annotations to WIV State
and Local Agencies for Confirmation. The models
were based upon the State plans and procedure
manuals. However, while some of these plans and
manuals provide detailed policy on all conceivable
issues, others describe only the general State policy.
Some States have augmented their State plans and
procedure manuals with additional guidance to local
agencies on eligibility procedures. In many cases,
States have given discretion to local agencies on some
elements of eligibility determination. These
considerations made it essential that WIV State and
local agencies be given an opportunity to correct or
supplement the models of their procedures. The State
and local agencies annotated the models with any
necessary changes.
4. Adjust Income-eligibility Operational and
Measurement Models on the Basis of State and
Local Feedback. The flowcharts developed in step 2
were modified to reflect the clarifications and additions
provided by WIV State and local agencies.
5. Abstract Eligibility Determination Records. The WIV
sample was based on a subsample drawn for PC88. As
part of PC88, enrollee records were abstracted. These
record abstracts included reported income and
household size.
6. Conduct In-home Audits. In-home audits were
conducted on a subsample of PC88 survey respondents.
During the in-home audits, respondents were requested
to provide detailed information on all income sources for
all household members. Supporting documentation was
requested. Respondents were also asked about all
persons who were potentially part of the enrollee's
economic unit.
7. Conduct Independent Determination of Income
Eligibility. Results of the in-home audits provided the
basis for an independent determination of WIC income
eligibility. As described above, the determinations were
based on applicable State and local income-eligibility
models. Errors were declared only when the in-home
3-18
audits produced unambiguous evidence of rule
violations. When the models allowed discretion on the
part of the professional determining eligibility, the choice
was always made in favor of the applicant. This
decision rule was used for two reasons: First, our case
reviewers were not in the position to overrule profes-sional
judgments made by WIC certification workers who
had had direct contact with the prospective enrollee.
Second, because most States explicitly allow various
areas of eligibility worker discretion, use of this
discretion should not be interpreted as error.
The independent determination classified each case as
eligible, ineligible, or unable to determine eligibility
(usually because of insufficient information).
8. Perform Computer Edit Review. Computer routines
were written that compared State eligibility standards,
enrollee income, and enrollee household size to produce
an eligibility determination. Computerized determina-tions
were used to detect errors in the eligibility
determinations made during step 7. In all, less than
one-half of 1 percent of the determinations were found to
be incorrect. These determinations were corrected.
9. Conduct Case Reconciliation. For all cases initially
identified as having errors in step 7, a rereview was
conducted. A second coder, who was trained and
experienced in WIC eligibility procedures, independently
reviewed the WIV questionnaire data for that case.
Whereas the first coder relied exclusively on in-home
audit data, the rereview also included the relevant data
abstracted from the case file fo PC88. The rereview
was designed to detect any errors in the original
determination and to determine whether the original
eligibility determination was allowable given the in-home
audit data. Approximately 11 percent of the cases found
ineligible in step 7 were reversed (to either eligible or
unable to determine eligibility status) by the rereview.
Cases for which the first reviewer had been unable to
determine eligibility were also reviewed, and several
were assigned to the eligible category on rereview.
10. Perform Enrollee Reconciliation. An attempt was
made to contact by telephone all cases found in error in
step 9, as well as all cases where inadequate data
precluded an eligibility determination. Respondents
were asked to aid us in understanding discrepancies
3-19
lit
between income and household size data from their WIC
case file and data obtained during the in-home audit. In
approximately 22 percent of the error cases, new
evidence was found to reverse the determination and
declare the enrollee eligible. In all, 13 error determina-tions
were reversed: 4 because of interviewer error, 6
because an incorrect income period had been used for
the independent eligibility determination, 2 because the
questionnaire was not sufficiently detailed to allow
correct determination of eligibility, and 1 because of
respondent recall error during the WIV interview; of the
cases with inadequate data, 1 was determined to be
ineligible and 4 were determined to be eligible.
3-20
TO
Chapter 3 Notes
I Study of WIV Participant and Program Characteristics, 1988 Final Report (Volumes 1 and 2).
2WIV did not have direct access to case-files, but obtained case-file data indirectly from the PC88
PRAF.
396 percent of all PC88 respondents in the sampled clinics agreed to follow up.
4The equation does not replicate exactly because of rounding effects.
5Avoiding being interviewed without actually refusing can involve a wide variety of creative
behaviors such as never having a free moment for the interview, repeatedly telling the interviewer
to please call back later, and repeatedly being away from home at the time scheduled for the
interview.
^Discriminant function procedures were employed. The primary predictors in the equation were
reported household income, median income reported in local agency, presence of wage income,
and absence of AFDC income. The fit of the model was statistically significant at the 0.001 level.
7Ratio estimates are the ratio of the unadjusted estimate to the actual value as observed from an
independent source. See Leslie Kish, Survey Sampling, New York: Wiley and Sons, 1965, pp.
433-34. Ratio estimators were necessary because the PC88 sample was based on the incidence
of enrollment in WIC, not the prevalence of participation.
8As a rule of thumb, we excluded estimates with a relative variance above 0.5. We also did not
present any error rate estimates for population subgroups containing fewer than 100 sample
cases.
^Despite these measures, one potential respondent refused to cooperate, called her local agency,
and withdrew from the WIC program.
10C. F. Cannel, P. V. Miller, L. Oksenberg, "Research on Interviewing Techniques," Sociological
Methodology, S. Leinhardt, ed., Jossey Bass, San Francisco, 1981.
II For additional information on the substudy, see appendix D.
12For details, see the error determination procedures presented on page 3-17.
13Federal Register, Vol. 50, No. 30, Feb. 13,1985, p. 6122.
14FNS Instruction 803-3, Oct. 22, 1982, p. 3.
^Federal Register, Vol. 50, No. 30, Feb. 13,1985, p. 6127. See also Fee/era/ Register, Vol. 52, No.
107, June 4,1987, p. 21234, for recent exclusion of certain Federal student assistance grants and
scholarships.
^Federal Register, Vol. 50, No. 30, Feb. 13,1985, p. 6127.
3-21
SI
177 CFR, Section 246.7(c). The income guidelines used by State agencies must equal either those
for reduced-price school meals (185 percent of the poverty level) or those for free or reduced-price
health care, with the proviso that the WIC income guidelines must fall between 100 and 185
percent of the poverty level.
18We include in the certification error rate errors that occur either at initial certification or at
recertification. The definition of prevailing income-eligibility standards was made on a State-by-
State basis using State plans, procedure manuals, and other materials provided by States. States
individually verified our definitions of prevailing standards. No implication should be drawn that
these other materials have been reviewed or approved at the Federal level.
19State agencies have discretion in ruling whether or not income changes during the certification
period affect eligibility. Each State's policy was followed in calculating the error rate due to
changes in income.
3-22
4.
THEWIC
ECONOMIC UNIT
FEDERAL REGULATIONS
AND GUIDANCE
Determination of income eligibility for WIC is a complex process.
Two variables are required for determining whether an applicant's
economic unit's income falls within the income-eligibility
guidelines: the number of persons in the economic unit, and the
total countable income of those persons. This chapter discusses
the first variable, determination of the economic unit membership.
Once the economic unit is defined and its membership is
established, the members' collective income can be enumerated.
Chapter 5 reviews the second variable, economic unit income.
Federal regulations allow States a certain level of discretion in the
definition and determination of economic unit membership. An in-depth
review of the income-eligibility policies and procedures for
the 25 sampled State agencies revealed that each State
employed a unique set of policies and procedures.
The process by which the economic unit is determined varies
across States at the most fundamental levels: Basic definitions of
economic unit vary, definitions for special cases vary, and
procedures for operationalizing the definitions vary. The degree
of variation in these policies and procedures was a major finding
of the WIV project.
WIC income certification error consists of enrolling applicants in
violation of established eligibility standards. Therefore,
understanding the variation in State economic unit membership
standards is an essential prerequisite to understanding
certification error. There is no across-State standard; State
variations exist at the center of the income-eligibility determination
process. Many applicants eligible in one State would not be
eligible in another.
This chapter addresses the existing policies and procedures for
identifying the members of an economic unit, as promulgated at
the Federal level and implemented by State agencies. The
chapter starts with a review of Federal regulations and
instructions. This is followed by a discussion of State agency
policies and procedures, including three examples of State
economic unit determination models.
State agencies have been provided by the Federal Government
with two types of guidance in defining the economic unit. These
two types are regulations and instructions.
A consolidation of the WIC regulations was issued by the USDA's
Food and Nutrition Service in July 1988. This consolidation
4-1
5-3
includes regulations issued June 4,1987; July 2,1987;
January 27,1988; and July 6,1988. The consolidation, earlier
Federal regulations, and instructions issued by FNS all help to
define economic unit membership.
According to the consolidation,
"'Family' means a group of related or nonrelated individuals
who are not residents of an institution but who are living
together as one economic unit"
The Federal regulations provide no further information on how
membership in the applicant's family or economic unit can be
determined.
In 1982, FNS issued Instruction 803-3 (10/22/82), titled WIC
Program-Certification: Income Eligibility, to supplement the
regulations. Revision 1 to FNS Instruction 803-3 was issued
April 1,1988; this revision had limited influence on this study,
since it was issued on the date that data collection for the study
began and most State agencies did not have time to respond to it
before data collection ended. According to FNS, these
instructions are binding on States.
The 1982 version of FNS Instruction 803-3 reiterated the
discretion allowed the State and local agencies in economic unit
determinations, provided a definition of an economic unit, and
followed this with specific policies developed for special cases.
The full text of this section is relevant because it illustrates the
Federal view on certain types of special cases as well as the
general definition of an economic unit. The text is as follows:
The philosophy regarding 'economic unit' and related issues is
offered as guidance for State agencies using the WIC
definition of income as outlined in the regulations. It is not the
intent of this instruction to make income determination kor WIC
a complicated and lengthy procedure, nor is an answer readily
available for every question on a specific case. Therefore,
there will be times when State and local agency workers will
need to use their discretion in determining income eligibility
within the general framework of regulatory requirements and
basic program policy.
"A. For the purposes of the WIC Program, the terms
'economic unit' and 'family' can be used interchangeably. We
perceive a family to be a household or an economic unit
composed of a person or group of persons who usually
(although not necessarily) live .ogether, and whose
rt
4-2
production of income and consumption of goods or
services are related. [Emphasis added.]
"B. A child is counted in the family size of the parent or
guardian with whom the child lives. For example, the mother
and child live together and receive child support payments
from the father. The father has remarried and lives with his
new wife who is expecting a baby. The new wife and the ex-wife
apply for WIC benefits. The ex-wife and her child are a
family of two, and the child support payments are counted as
income. The father and the new wife are also a family size of
two until the new wife has her baby. The child support
payments cannot be deducted from the father's income and
the child living with the ex-wife cannot be counted in the
father's family.
"C. If a child resides in a school or institution and the child's
support is being paid for by the parent or a guardian, the child
may be counted in the family size of that parent or guardian.
Even though this child is living apart from his parent or
guardian for the majority of the time, the child may be counted
as part of the family, since the family continues to provide the
economic support for the child.
"D. If the child is a foster child who is living with a family but
who remains the legal responsibility of a welfare or other
agency, the foster child shal' be considered a family of one.
The payments made by the welfare agency or from any other
source for the care of that child shall be considered to be the
income of that foster child. If the annual income for the foster
child is at or below the income criterion, the foster child is
income eligible for WIC benefits.
"E. When a family has an adopted child or a child for whom
the family has accepted legal responsibility, the child is
counted in the family size of the family. The size and total
income of third family shall be used to determine the child's
income eligibility for WIC." (pp. 3-4)
The 1982 version of Instruction 803-3 also provides a definition of
an emancipated minor:
The determinant of whether a minor is emancipated and thus
a separate economic unit for our program purposes is whether
or not the minor is living without economic support form (sic]
other persons. If the minor receives any support for which she
does not pay, such as shelter or meals for example, she
should not bo considered a separate economic unit. If ihe
ST
minor pays all expenses for her own support, it is possible that
the minor may then be considered a separate household. It is
entirely possible for two separate economic units to reside
under the same roof, although the determination of this is
usually not a clear cut process." (p. 4)
This instruction also advises that pregnant women not be counted
as two persons in determining the size of the economic unit.
STATE POLICIES AND Because Federal regulations are quite general and allow State
PROCEDURES agencies considerable discretion in defining economic unit
membership, State agencies have developed their own policies
and procedures. Therefore, to understand WIC economic unit
definitions, one must be familiar with the State policies and
procedures. The WIV project had three sources of information
about these policies and procedures:
The State Census conducted in Phase I.
State plans and procedure manuals. The regulations
require each State agency to submit a State plan to FNS
for approval by August 15 of each year. The State plan
must include a copy of the procedure manual issued by
the State agency for local agency use, including a section
on certification procedures (CFR 246.4 (a)(11)(i), 7/88).
The sections of these State plans and procedure manuals
treating economic unit determination were requested from
the State agencies during the State Census. Later, more
complete information was requested for the 25 State
agencies in the WIV sample; this information was used to
construct a flowchart of each sampled State's procedures.
Contacts with State agencies by mail. Each of the 25
sampled State agencies reviewed its flowchart and
corrected it as necessary.
The initial review included policy and procedure manual sections
from 44 of the 51 geographic State agencies for which information
was obtainable from State and Federal sources. The purpose of
this review was not to provide a complete and detailed review of
State procedures, but to simply identify variables used in the
procedures. Figure 4-1 was based on that initial review. It shows
the main variables required to determine income eligibility for an
applicant. In order to make a determination of income eligibility,
information on the specific percentage of the poverty income
guidelines used by that State is required. Next, the number of
persons in the economic unit must be determined. Finally, the
TJb
Determination of
Income Eligibility
Percent of
Poverty Income Guidelines
Number of Persons
in Economic Unit
Criteria Defining
Economic Unit
Amount of Income
Period of Time in
Which Income
Was Received
I
Criteria Defining
Income
Special Case Definitions
Share
Income
Share
Goods and
Services
Share
Living
Quarters
Persons Whose
Income Is
Counted
Definitions of Included Income Definitions of Excluded Income
Emancipated
Minors
Child Under
Joint Custody
Parent Who
Takes the Tax
Deduction
Household in
Which Child
Spends Major-ity
of Time
Institutionalized
Family Member
Relationship of
Members
— Blood
— Law
— Other Foster Child
I
Who Has Legal
Responsibility?
Extent of
Family's
Financial
| Support
Salary and
Wages
Self-Support
— Nature
— Extent
— Source
Cash Public
Assistance
(AFDC, SSI)
Age Criteria
Pensions,
Social Security
Payments from
Federal
Programs That
Are Prohibited
by Law from
Inclusions,
e.g., VISTA,
RSVP, etc.
Living
Quarters
Dividends, Interest,
Income from Estates
and Trusts
Marital
Status
Unemployment
Insurance
Benefits
Student
Financial Aid
Funded by
Higher
Education Act
of 1965,
Title IV
Pregnancy
Status
Alimony/Child
Support
Duration
of Time in
Institution
Withdrawals from
Savings, Investments,
Trust Accounts
In-kind
Benefits, e.g.,
Value of Food
Stamps
Figure 4-1. Income-eligibility determination tree.
income of those persons must be measured and compared to the
income guidelines.
This chapter covers the left side of figure 4-1, or economic unit
determination. Most State agencies' procedure manuals give a
general definition of an economic unit, followed by definitions
applying only to special cases. Under "General Definitions," the
figure lists some of the variables that the States in the WIV
sample included. For example, membership in the economic unit
may require being related to the applicant, residing with the
applicant, sharing goods and services with the applicant, or
sharing income with the applicant.
Some States have more specific general requirements; for
example, in some States the only acceptable relatives are the
applicant's spouse, parents, and siblings; another State specifies
that economic unit members must purchase food together, not
just any goods or services.
Most States also have special case definitions; in some cases, the
procedure manual has only special case procedures and lacks a
general definition. Figure 4-1 shows some of the types of special
cases that were identified during the initial review.
General Definitions On the most basic level, family or economic unit can be defined in
terms of legal relationships, economic relationships, residency,
biological relationships, or social relationships. Different State
WIC agencies have adopted various combinations of all these
approaches.
A number of State agencies quote FNS Instruction 803-3 as their
general definition of the economic unit. A few cite the Federal
regulations as their definition. Others have unique general
definitions, usually quite similar to that in FNS Instruction 803-3.
Examples of these unique general definitions from State agencies'
manuals reveal the fundamental differences that exist:
"The household, family or economic unit is defined as follows:
A person or group of persons who usually, although not
necessarily, live together, and whose production of income and
consumption of goods or services are related A key factor
in determining an economic unit is establishing if income and
consumption of goods and services are shared. A group of
persons, whether living together or separately, are an
economic unit by virtue of sharing income and consumption of
goods and services." (1/87)
4-6
"Definition of Family. Persons living in one household who are
related by blood, marriage, law or conception."
"Family size includes all related or non-related individuals who
live under one roof and share financial resources for the
purchase of food."
The three examples present entirely different criteria for
determining household membership. The first definition explicitly
excludes residency as a criteria; the second and third require it.
The first and third definitions include economic criteria, while the
second makes no mention of economic relations. The second
definition relies on relationships of "blood, marriage, law or
conception," criteria absent from the first and third definitions.
Appendix G presents additional examples of general definitions of
households used by State agencies.
Special Case Definitions As fundamentally different as are the basic definitions of
economic unit used by States, these basic definitions reveal only
a portion of the total across-State variance in definitions of
economic unit.
For most applicants, membership in the economic unit can be
determined by the general definitions. For others, the special
case definitions may be invoked. Some States have only one to
two special case definitions, while others include procedures for
many types of special cases in their manuals. Special cases
defined by different States include households with individuals
such as pregnant minors, foster children, college students who
live apart from their parents but whose permanent home is with
their parents, institutionalized family members, and children under
joint custody arrangements.
In some States, special case definitions always include or exclude
certain individuals from the economic unit. In other States, such
individuals may be either included or excluded, depending on
whether they meet one or more criteria. Some of these criteria
are shown in figure 4-1 in boxes below the type of special case.
For instance, in many States a foster child is a one-person
economic unit if an agency (such as a court or welfare agency)
has legal responsibility for that child; the key criterion is who has
legal responsibility. To determine whether a teenage girl is an
emancipated minor (i.e., a separate economic unit from her
parents), States may consider, to list some of the more common
criteria, whether the teenager is pregnant or has children; whether
she supports herself, and if so the extent, nature, and source of
her self-support; her current age or age at conception; whether
4-7
Examples of State
Economic Unit
Determination
Procedures
the teenager lives with her parents, other relatives, nonrelatives,
or alone; and whether she is now or has ever been married.
Once the State agency's criteria have been applied to each
potential member of the applicant's economic unit, the identity and
number of persons in the economic unit is known. This
information is crucial for determining the applicant's eligibility,
since the income of persons who are not members of the
economic unit is not counted. For example, a pregnant minor
living with her parents may be eligible for WIC benefits if she lives
in a State that excludes her parents from her economic unit, and
ineligible if she lives in a State that includes her parents in her
economic unit. Economic unit definitions, therefore, can
significantly affect eligibility.
Variations in definitions of economic unit only reveal the surface of
across-State variations in economic unit determination
procedures. More fundamental differences exist in how these
definitions are operationalized in practice.
Economic unit determination practice is an integrated and unique
system in each State. Flowcharts of these systems were
developed for each sampled State in the WIV study. To illustrate
the variations in State agency policies, three examples of the
State economic unit determination flowcharts are presented and
explained in this section.
The "General Definitions" of economic unit discussed above and
illustrated in figure 4-1 are represented on the flowcharts as "main
paths"; these paths show the process that most applicants
experience. States also have "side paths" that apply only to the
special cases (see "Special Case Definitions" above and in figure
4-1).
The boxes on the flowcharts represent the steps in determining
economic unit membership: actions, decision points, and
outcomes. Paths, or the sequence of steps followed, are
indicated by lines connecting the boxes. All flowcharts in this
section are read beginning at the box in the upper left corner of
the page. The main path starts from the bottom of that box and
continues down the left side of the page; it generally turns at the
bottom of the page and ends in the center of the right side of the
page. The main path always ends at the final decision box, which
asks whether the applicant's economic unit income is under the
maximum for that size of family.
The diamond-shaped boxes represent decision points; if the
question in the box is answered "yes," one path is followed, and if
4-8
60
it is answered "no," another path is followed. The parallelograms
represent outcomes, and the rectangular boxes are actions to be
taken. For a particular case, a series of paths connecting certain
boxes is followed until the final outcome box (either "Income
Eligible for WIC" or "Not Income Eligible for WIC") is reached; the
paths and boxes represent the procedures followed by the WIC
staff member certifying an applicant.
Example 1. State with The first example flowchart (figure 4-2) is for a State with a very
Simple Procedures simple procedure. This State has only one type of special case
and does not use the presumptive eligibility procedure. Judgment
is to be applied at only one step: deciding whether to apply a
higher income limit to the case than the limit used for most cases.
This flowchart begins with a decision box asking whether the
applicant has a VOC (Verification of Certification) card from
another WIC agency. Any applicant who possesses a valid VOC
card is considered to be income-eligible for WIC in this State and
need not go through the rest of the procedure. (The VOC card
can be expired, as long as income eligibility was determined
within the last 12 months.)
If the applicant does not have a valid VOC card, the next box asks
whether the applicant is a foster child for whom the State has
legal responsibility. If the answer is "yes," the side path to the
right is followed. The initial outcome is that the applicant is an
economic unit of one person. The WIC staff then takes an action
(determining the foster child's income). The next decision box
asks whether that income is at or below the maximum allowed for
a one-person family; if the answer is "yes," the applicant is
income-eligible for WIC, and if "no" the applicant is not income-eligible
for WIC.
Note the discretion allowed in the final decision box in this
example. In exceptional cases, a different income standard can
be applied. (Discretion is found at various points in the
procedure for different States, as can be seen from the three
examples in this section.)
An applicant with no VOC card and who is not a foster child for
whom the State has legal responsibility goes to the third decision
box on the main path: "Reside in household?" This means that,
for each person who is potentially a member of the economic unit,
the WIC staff must determine whether he or she resides in the
applicant's household. If the answer is "no," that person is not a
member of the applicant's economic unit (see outcome box).
■6/ 4"9
Economic unit >
Determine
income of the
economic unit
Not included
in the
economic unit
Not included
in the
economic unit
Figure 4-2. State with simple procedures.
4-10
b£s
Example 2. State with
Moderately Complex
Procedures
Once all those who reside in the applicant's household are
identified, it must be determined which of these persons share
economic resources. Any who do not share with the others are
not members of the economic unit. The fourth decision box on
the main path shows this decision.
At this point, the identity and number of economic unit members is
known. The next step is to determine the income of the economic
unit (by determining the countable income of each member and
summing). These applicants then reach the final decision box: Is
the economic unit's income under the eligibility limit for that size of
economic unit? As for a foster child, if "yes" the applicant is
income-eligible for WIC, and if "no" the applicant is not income-eligible
for WIC. Again, the discretionary higher eligibility limit
may be applied.
A foster child who is the legal responsibility of the State is the only
special case in this State. These applicants are the only ones for
whom the procedure follows the side path. All other applicants
follow the main path.
The second example (figure 4-3) shows a State with more
complex procedures. This state has numerous types of special
cases (foster children, children living in schools or institutions,
persons in prison, self-supporting minors, and children of divorced
or separated parents). This State accepts enrollment in either of
two means-tested programs (Medicaid and Food Stamps) as
evidence of income eligibility for WIC. Judgment is to be applied
at two steps: determining whether certain minors are essentially
self-supporting, and making the final determination of economic
unit membership.
The first apparent difference between this State and the one in
the previous example appears in the first decision box (in the
upper left corner of the chart). In addition to valid VOC cards, this
State accepts enrollment in either Medicaid or Food Stamps as
presumptive evidence of income eligibility. Persons who are
enrolled in those programs are considered to be income-eligible in
this State and need not go through the rest of the procedure.
If the applicant does not have a valid VOC card and is not
enrolled in Medicaid or Food Stamps, the next decision box asks
whether any child in the household resides at a school or insti-tution.
If the answer is "yes," the sidepath to the right is followed.
If that child's parents pay for support, the child is not excluded
from the economic unit. If not, that child is excluded from the
63
4-11
6H
4-12
Child not
included in the
noncustodial
household's
economic unit
e Do \
individuals in\
household
share y
Vincome/^
\ ? /
vNO
/
/ Not included in /
/ the economic /
unit /
Size of
(economic unit and,
ID of members
(use judgment),
Determine
income of the
economic unit
Income-eligible
forWIC
Not
income-eligible
lorWIC
Figure 4-3. (continued).
4-13 cr
applicant's economic unit. Note that if a child is excluded from the
economic unit for this reason, the procedure for determining
economic unit membership for the other members of the house-hold
continues on the main path.
The next decision box asks whether any member of the potential
economic unit is in prison. If so, that person is not included in the
economic unit.
The fourth decision box asks whether the applicant is a foster
child. If so, the sidepath to the right is followed: If an agency has
legal responsibility for the foster child, the child is a one-person
economic unit and the money paid for the child's support is the
economic unit's income. If that income is 185 percent or less of
the poverty level for a one-person household, the child is income-eligible
for WIC; if not, the child is not income-eligible for WIC.
If the applicant is not a foster child (or is a foster child for whom
an agency does not have legal responsibility), the main path
continues with the fifth decision box: Is the applicant a minor who
lives in a separate household with no support from others? If the
answer is "yes," the minor (and her children, if any) is a separate
economic unit. The income and then the income eligibility of that
economic unit are determined.
The sixth decision box asks: Is the applicant a minor who lives
with her parents and is not claimed by them as a dependent for
income tax purposes? If so, the sidepath to the right is followed.
The decision box to the right asks whether the minor essentially
supports herself. If the minor is determined to be self-supporting,
the minor (and her children, if any) are a separate economic unit
from her parents' economic unit. Several States allow a minor
living with her parents to be considered a separate economic unit
under certain circumstances. Note that judgment is to be used in
making this decision.
If the applicant does not meet all of the criteria in the fifth decision
box, the main path continues with an action box: enumeration of
all household members who are potential members of the
economic unit. This box is repeated on the next page to show the
point from which the flowchart is continuing.
The first decision box on the second page asks whether the
household includes any divorced or separated parents. If "yes,"
the sidepath to the right is followed. The purpose of this sidepath
is to determine economic unit membership for children of parents
who are divorced or separated. If a child is under a joint-custody
4-14
arrangement, the child is counted as a member of both parents'
economic units. If not, the next decision box on this sidepath asks
whether the noncustodial household pays child support. If "yes,"
the child is counted as a member of both parents' economic units;
if "no," the child is a member only of the custodial parent's
economic unit.
The second decision box on the second page of figure 4-3 asks
whether the applicant's household pays child support for a child
living in another household. If "yes," the child is counted as a
member of the applicant's economic unit (as well as a member of
the custodial household's economic unit).
The main path continues with the third decision box on the second
page: Do individuals in the household share income? Any
individual who has income and does not share any of it with other
household members is not included in the economic unit.
The outcome of this process is the size of the economic unit and
the identity of its members. Note that, again, WIC staff are
instructed to use their judgment in making the final decision on
economic unit membership.
Once the economic unit has been determined, an action box is
reached. The countable income of each member is determined
and summed; this is the economic unit's income. The final
decision box asks whether this income is less than or equal to 185
percent of the poverty level income for that size of economic unit.
If "yes," the applicant is income-eligible for WIC; if "no," the
applicant is not income eligible for WIC.
Example 3. State with The third example (figure 4-4) is a State with. complex
Complex Procedures procedure. This State has a large number of .ypes of special
cases (foster children, separated couples, persons who are
temporarily absent from the household, households including
persons other than parents and their children, children who are
eligible for Medicaid, married minors, and children over age 18
who receive only food and shelter from their parents). The State
accepts proof of enrollment in or eligibility for any of several
programs (AFDC, Medicaid, Food Stamps, free/reduced price
School Lunch), or enrollment as a county health department
patient with a recent eligibility review as evidence of WIC income
eligibility. This State also differs from those in the other examples
in that the use of discretion or judgment is not specified at any
point in the procedure.
67 4-15
Figure 4-4. State with complex procedures.
6?
4-16
••Parent" includes a cohabiting partner included in the economic unit.
Also includes a step-parent
Figure 4-4. (continued)
&e 4-17
As in the other States, the procedure begins with a determination
of whether the applicant possesses a valid VOC card. If so, the
applicant is automatically considered to be income-eligible for
WIC.
The next three decision boxes ask about eligibility for AFDC,
Medicaid, Food Stamps, and reduced price/free School Lunch,
and enrollment as a health department patient in certain income
categories. If the applicant is a health department patient, a
sidepath is followed; if the applicant's last eligibility review was
less than 6 months ago, income eligibility for WIC is presumed
and the applicant need not go through the rest of the procedure.
Note that for the other programs, eligibility for the program is
specified rather than enrollment. Thus, if a WIC applicant
presents recent evidence that AFDC has determined her to be
eligible, that can be used as evidence of income eligibility for WIC
even though the applicant has not yet started to receive AFDC
benefits.
The last decision box on this page covers the special case of a
foster child applicant. This is handled in the same way as in the
other example States, except that this State does not specify that
an agency must have legal responsibility for the child.
As in Example 2, the first page of the flowchart ends with an
action box: enumeration of all members of the household. The
second page begins with the same box to show the continuity of
the procedure.
The first decision box on the second page asks whether all
household members reside together. If "no," this is a special case
and the sidepath to the right is followed.
If all household members reside together, the procedure
continues along the main path with the second decision box on
this page. This asks whether the household includes only parents
and their children (with the definition of "parent" including
cohabiting partners and step-parents). If the household includes
anyone else, this is a special case and the sidepath to the right is
followed.
If the household is composed only of parents and their children,
the procedure continues along the main path with the next
decision box. This box asks whether any of the children are
eligible for Medicaid. If so, those children are separate economic
units from their parents. The parents and any remaining children
continue along the main path. The next decision box asks
4-18
10
whether all of the children are unmarried. If "no," this is a special
case and the married minor, his or her spouse, and their children
are a separate economic unit from the rest of the household.
If all the children are unmarried, the next decision box asks
whether all the children are under age 18. If not, this is a special
case and the sidepath to the right is followed.
If all the children in the household are under 18, the outcome box
at the bottom of the page is reached: The economic unit is the
parents and their children. The next step in the procedure is
represented by an action box; the income of the economic unit is
determined. The final decision box, as in the other examples,
asks whether that income is below the eligibility limit for that size
of economic unit. If "yes," the applicant is income-eligible for WIC.
If "no," the applicant is not income-eligible for WIC.
Summary Differences In Although the determination of economic unit membership in each
State Procedures State is a unique and integrated process, some insight into the
types of approaches used by States can be gained by examining
the distribution of