PART B - Statistical Method

November 13, 2007

B. COLLECTION OF INFORMATION EMPLOYING STATISTICAL
METHODS
For this study, we will collect information on FSP modernization efforts occurring at the
state and local levels in a manner which will help to reduce the burden on respondents, especially
local FSP agency and partner respondents. The following section describes these information
collection methods.

1.

Potential respondent universe and sampling

The respondent universe for this study is state FSP agencies, local FSP agencies, FSP
partners (faith-based organizations (FBOs), community-based organizations (CBOs), and forprofit contractors), FSP participants, FSP applicants, and eligible FSP nonparticipants. Data will
be collected from respondents through surveys and site visits. Surveys will be administered to
51 state FSP directors, 150 local FSP directors, and 150 representatives of FBOs/CBOs and forprofit contractors. Case studies will be conducted in 14 states and include interviews with staff
at state FSP agencies, local FSP agencies, FBOs/CBOs and contracted organizations, and FSP
participants, nonparticipants, and applicants.
Surveys of State, Local, and Partner FSP Modernization. Three surveys will be
administered during this data collection—the State, Local, and Partner FSP Modernization
Surveys. FSP directors from all 50 states and the District of Columbia will receive the on-line
State FSP Modernization Survey and 150 local FSP directors will receive instructions for
accessing the on-line Local FSP Modernization Survey. The largest county in each state,
counties identified by state FSP administrators as conducting modernization activities, and
additional counties selected at random will also receive the Local FSP Modernization Survey.
The Partner FSP Modernization Survey will be administered to a sample of 150 partners—
FBOs/CBOs and private contractors that work with the state and/or counties—that state and local
FSP directors identify as being involved in some capacity with FSP modernization efforts.
Sampling issues arise in the selection of counties for the survey phase. Because the
study’s objective is to develop an inventory, and not to make any statistical inferences, purposive
sampling techniques will be employed. The FSP director of the largest county (in terms of FSP
participation) in each state will be surveyed and asked to identify counties that are conducting
modernization activities. Local FSP directors of up to two additional counties per state, as
nominated by their state FSP director, will be surveyed. If this yields a sample of less than 150
counties, the sample will be supplemented by a random sample of remaining counties. The
following chart shows how the survey respondents will be selected.

16

November 13, 2007

Survey Sampling Plan
SAMPLES
TAKEN

State FSP
Directors
All states plus
District of
Columbia (51)

Local FSP Directors

Partner Organizations

In each state, county with
largest number of households
reporting FSP participation
(50)

Selected private, for-profit
contractors and nonprofit
community-based and faithbased organizations identified
by state and local FSP
directors that have direct
contact with FSP applicants
and participants (150)

FSP directors in all counties
identified by state FSP
directors and not already
surveyed in the category
above (number to be
determined)

TOTAL
SURVEYED
TOTAL
ESTIMATED
POPULATION

51

Additional counties, selected
at random as needed to reach
target total (100 minus
number of identified
counties)
150

150

51

3,140 5

50,000 6

Several survey procedures will be used to ensure a high response rate, including the
establishment of liaisons for each state, receipt of a packet containing a FNS letter of
introduction and detailed instructions for completing the survey on-line or by hardcopy (see
Appendix E for a sample of these letters), and reminders through email, phone, and postcard at
appropriate intervals. (See B.3 for more information.) It is expected that these steps will
produce response rates of 100 percent for the state FSP agency officials, 95 percent among local
FSP agency officials, and 90 percent among partner organizations.
FSP Modernization Case Studies. A sample of 14 states will be selected for local site
visits. The overall goal is to select states that represent a broad range of variation across these
four factors: 1) state characteristics; 2) type of FSP modernization initiatives; 3) extent of FSP
modernization initiatives; and 4) stage of implementation. States selected will differ by the
following specific measures:

5

This number does not include the District of Columbia.
This is the number of charitable organizations working with America’s Second Harvest. America's Second Harvest
is the nation's largest charitable hunger-relief organization. It supports approximately 50,000 local charitable
agencies operating more than 94,000 programs including food pantries, soup kitchens, emergency shelters, afterschool programs, Kids Cafes, Community Kitchens, and BackPack Programs. See www.secondharvest.org for more
information.
6

17

November 13, 2007

x

State Characteristics/Context
o Size (FSP participants)
o Geographic region
o Demographic characteristics of population
o Application approval rates
o Administrative costs
o FSP error rate

x

Type of FSP Modernization Effort
o Policy changes
o Restructuring administrative functions
o Expanding application of technology
o Partnering with commercial businesses and non-profits

x

Extent of FSP Modernization Initiative
o Statewide
o State initiative piloted in selected counties/regions
o Single county initiative
o Single change versus multiple changes

x

Stage of Implementation
o Planned
o Early implementation
o Fully implemented

Since there are more criteria than the number of states to be selected, choices will have to
be made. FSP data on state characteristics available from FNS for all states, and survey data on
modernization, will be used to categorize states and guide the selection process. Ensuring that
the states selected represent a broad range of modernization activities will be crucial.
It is estimated that, per state, site visit teams will interview approximately: five state
agency staff; one staff from other state agencies (e.g., TANF); 18 local agency staff; five to six
staff of FBOs/CBOs or contractors; and 12 food stamp applicants and participants. Site visit
teams will hold two focus groups per site visit – one group of food stamp participants and one
group of eligible nonparticipants – with approximately six respondents per focus group. At the
state level, interview respondents include the state FSP administrators and staff; state FBO/CBO
partners or vendors responsible for modernization activities providing certification/recertification
services (does not include information technology or management information services); and any
additional state- or local level respondents identified as knowledgeable about and involved in
state modernization efforts.
Local sites that provide illustrative examples of different kinds of FSP modernization
initiatives will be selected based upon the information obtained in the surveys and other
background data collected. Up to two local sites that are involved in modernization activities will
be visited per state. Interviews will be conducted with FSP administrators and staff, and
representatives from local FBO/CBO partner staff and vendors. Due to time and resource

18

November 13, 2007

constraints, it will not be possible to visit all FSP service locations and partner organizations in a
state. Selections of counties will be made on the basis of size and diversity, and in consultation
with the state and local FSP agency staff. For example, FSP service locations and partners that
serve the largest number of FSP participants within the county will be visited. In addition to
selecting sites that represent a range of modernization activities, whether the service location
and/or partner organization focuses on a particular subgroup of the target population will be
considered for site selection purposes.
The local site visits will include focus groups with a small sample of FSP participants and
eligible nonparticipants and intercept interviews with FSP applicants and participants. One
participant focus group and one eligible nonparticipant focus group will be held in each state.
For each focus group, it is expected that approximately 12 respondents will attend. For the
intercept interviews, FSP staff will be asked to help recruit applicants and participants at the
offices. Approximately six FSP applicants and six FSP participants will be interviewed in each
state.
The following chart shows the potential respondents for the case study phase of the data
collection:
Potential Respondents for Case Studies
Categories of Respondents
State Food Stamp Agency
State FSP Interview Discussion
Guide

Estimated Number
Per State
5

Local Food Stamp Office (up to
two counties visited per state)
Local FSP Interview Discussion
Guide

14

Other State
Agencies/Departments
State FSP Interview Discussion
Guide
State or Local
Contractor/Vendor
FSP Partner Discussion Guide

1

FBO/CBO Partner (State or
Local)
FSP Partner Discussion Guide

3

3

Types of Respondent with Category
Director
Policy and Operations Staff
Staff responsible for MIS modernization
activities
Staff responsible for FSP outcomes
reporting
Director
Office supervisor
Eligibility worker(s)
Staff responsible for MIS/reporting
Participants/applicants (brief intercept
interviews)
Participants (focus groups)
Nonparticipants (focus groups)
TANF agency, budget agency, information
technology department

Vendor(s) responsible for implementing
FSP modernization activities (e.g., call
centers, document imaging, eligibility) or a
local vendor representative working
directly with the local agency
Representative of partner organization
working with local FSP office (if
applicable)

19

November 13, 2007

Categories of Respondents
Other Local Respondent
Local FSP Interview Discussion
Guide
Food Stamp Applicants
Applicant/Participant Intercept
Interview
Food Stamp Participants
Applicant/Participant Intercept
Interview and
FSP Participant Focus Group
Discussion Guide
Food Stamp Eligible
Nonparticipants
FSP Eligible Nonparticipant
Focus Group Discussion Guide
ESTIMATED RESPONDENTS
PER STATE

2.

Estimated Number
Per State
4

Types of Respondent with Category
County MIS staff, TANF agency case
managers, representatives of local
advocacy group

6

18

12

66

Procedures for the collection of information

Sampling Procedures. Statistical sampling will be conducted for this study to help
reduce the burden on respondents. The survey of state FSP directors on modernization efforts is
a census of all state Food Stamp Programs, including the District of Columbia. However, the
survey of local FSP directors and the survey of local FSP partners will sample the population of
these respondents. The survey of local FSP partners (i.e., FBOs, CBOs, contractors) will use a
convenience sample based on the recommendations from state and local FSP directors. The
survey of local FSP directors will use a combination of convenience and simple random
sampling to draw from the population. The following describes these sampling methods.
Statistical methodology for stratification and sample selection. Three sampling methods
will be employed for the local FSP directors’ survey. First, as described in B.1, the FSP
directors in each state’s largest county, in terms of the number of food stamp caseload, will be
surveyed. The second subsample of local FSP directors is a convenience sample to be derived
from the recommendations of the state FSP directors. The state respondents will be asked to
provide a list of the counties considered the most active in advancing modernization efforts. The
third subsample will be a simple random sample of the local FSP directors.
Estimation procedures. This survey is intended to develop an inventory of modernization
efforts, not make statistical inferences about these efforts. As this survey methodology for the
local FSP agencies and partners uses convenience sampling (as opposed to random), it is not
possible to establish confidence intervals.
Statistical techniques to ensure accuracy for the purposes described in this justification.
This study is intended to develop an inventory of modernization efforts, not make statistical
inferences about these efforts (see above). The data collected will be used to describe these
20

November 13, 2007

activities in all states and a small sample of local agencies and partners to capture modernization
efforts at a particular point in time. Thus, no statistical techniques will be used to ensure
accuracy.
Specialized sampling procedures to correct unusual problems. No specialized sampling
procedures will be used.
Periodic data collection cycles to reduce burden. This is a one-time data collection effort
and will not require period data collection cycles.
Procedures for Data Collection. This section describes the data collection procedures
developed for this study to provide quality assurance and consistent administration of the surveys
and case studies.
Data collection procedures for the surveys. The survey phase of this data collection
effort will begin upon receipt of OMB clearance. Prior to the surveys being fielded, state
administrators will be contacted and asked to identify and provide contact information on county
FSP agencies that have implemented or are pilot sites for state modernization initiatives and
county FSP agencies that have implemented their own modernization initiatives. State FSP
directors will also be asked to identify partner organizations and vendors involved in
modernization activities at the state and local levels. Once the local FSP offices are selected (see
full sampling procedure above), they will be contacted and asked to identify partner
organizations and vendors involved in modernization initiatives at the local level. The surveys
will be fielded once the full list of respondents is compiled.
A letter from FNS introducing the project and the survey accompanied by a letter from
the Urban Institute explaining how to complete the on-line survey will be sent to state and
selected local FSP directors and partners (see Appendix E for sample letters). A hard copy of the
survey will be mailed or faxed upon request. The same package will also be sent via email to
these respondents.
The survey will be developed for distribution primarily as an on-line survey, with hard
copy and mail-in options provided. The survey will be uploaded into Ultimate Survey, an online survey tool. The surveys will be tested both internally and externally to ensure that they can
be navigated seamlessly and without any malfunctions. Technical support by phone will be
available to individuals completing the surveys from 9:00am-5:00pm, Monday-Friday, while the
survey is being fielded.
Surveys completed electronically will be automatically uploaded into a Microsoft Access
database. Survey team members, trained on the survey instruments and data entry formats, will
manually edit all completed hard copy surveys before data entry within two days of receipt. To
ensure accurate and complete information, programmed edits will detect keying errors for
immediate corrections and flag others for later correction. As each survey is reviewed, follow-up
e-mails and telephone calls will be made to those respondents whose surveys contain errors,
unclear responses, or missing information. If a survey team member is uncertain about how to
code a response or whether follow-up is needed, the study’s survey team leader will review the

21

November 13, 2007

item. All coding decisions made in such cases are documented to assure consistency in coding.
A PC-based tracking system will be used to monitor the receipt of surveys, status of follow-up
reminders, attachments provided by respondents, completion of data entry, and need for further
clarification.
Data collection procedures for the case studies. For the case studies, one senior Urban
Institute staff member with significant and recent FSP experience will lead each site visit team.
An advance letter from the Urban Institute (see sample letter in Appendix E) will be sent to
states letting them know they have been selected as a case study site. Staff will then work
closely with each site to develop a schedule to meet with the appropriate respondents. Below is a
sample site visit schedule that shows how the visits can be scheduled to assure that all needed
respondents are interviewed.
Sample Site Visit Schedule for a Site Visit
Time
9:00 – 11:00
11:00 – 12:30
12:30 – 1:30
1:30 – 3:00
3:00 – 4:30

Day One: State-Level Data Collection
Site Visitor 1
Site Visitor 2
State Food Stamp Director
State Food Stamp Policy & Operations Staff
Lunch
Staff responsible for FSP MIS modernization activities
Staff responsible for FSP modernization outcomes reporting

Time
9:00 – 10:30
10:30 – 12:00
12:00-1:00
1:00 – 4:30

Day Two: State-Level Data Collection
Site Visitor 1
Site Visitor 2
Staff responsible for state-wide MIS modernization
Vendor or other partners responsible for state-wide modernization efforts
Lunch
Additional meetings, as determined

Time
8:00 – 10:00
10:00 – 11:30
11:30 – 12:30
12:00 – 1:00
1:00 – 2:00
2:00 – 3:00
3:00 – 4:30
4:30- 6:30
6:30 – 8:30

Day Three: Local-Level Data Collection
Site Visitor 1
Site Visitor 2
Local FSP Office Administrator
Local Office Supervisor
Eligibility worker
On-site observation
Lunch
Intercept interviews
Eligibility worker
On-site observation
Intercept interviews
Representatives from FBO/CBO Partner
Break
Focus group  FSP Participants

22

November 13, 2007

12:30 – 1:00
1:00 – 2:30

Day Four: Local-Level Data Collection
Site Visitor 1
Site Visitor 2
Eligibility worker
Eligibility worker
Focus group  nonparticipants
Representatives from FBO/CBO
On-site observations/
partner
document and data review
Lunch
Representatives from local vendor
Real-time testing

2:30 – 3:30
3:30- 4:30

Additional meetings, as determined
Intercept interviews
Additional meetings, as determined

Time
8:00 – 9:00
9:00 – 10:30
10:30 – 12:30

Staff training will help assure consistency in data collection procedures and
documentation. All site visitors will participate in a training session that reviews the data
collection instruments, and reviews site visit roles and protocols. The training will also cover
logistics related to arranging focus groups, as well as focus group procedures and methods.
Once the site visits are completed, all staff will use the same outline for internal site summaries,
and site visit information will be coded and entered into a qualitative analysis software package.

3.

Methods to maximize response rates and to deal with issues of non-response

As this is a qualitative study of FSP modernization efforts across the country, no
statistical methods will be used to sample respondent populations. All 50 states, as well as the
District of Columbia, will be systematically surveyed. As all respondents in the universe will be
surveyed, results from the survey will yield a full documentation and analysis of the
modernization activities occurring in all states. The case studies will provide richer detail and
further analysis on particular models of FSP modernization the states are undertaking. Study
results will inform FNS policy discussions, and provide technical and procedurally relevant
information to states. We will also create a centralized source of information that can be used for
assessing ways to improve food stamp certification and responding efficiently to the variety of
stakeholder queries received.
To ensure the full documentation of activities in all states, having the necessary response
rates is vital to the study. The following methods will be utilized to ensure attainment of target
response rates.
Surveys of State, Local, and Partner FSP Modernization. We expect that the steps
outlined below will produce response rates of 100 percent among state officials, 95 percent
among local officials, and 90 percent among partner organizations. These response rate estimates
are based on previous survey work completed at the Urban Institute and by other organizations.
Recent Urban Institute mixed-mode surveys achieved response rates above 98 percent among
state officials in other policy areas. Also, the web-only GAO survey of 2002 Farm Bill options
and the more recent GAO web-based survey on the use of alternative methods to apply for and
maintain benefits obtained responses from all 50 states. 7 In addition, FNS’s strong support for
7

See GAO 2004; and GAO 2007.
23

November 13, 2007

the project and encouragement of participation is likely to have an impact on state and local
officials, and on private contractors and community organizations.
The survey procedures are designed to ensure high response rates among respondents in
all three categories. Respondents for each of the surveys will be contacted by both surface mail
and email. They will receive a packet containing a FNS letter of introduction and an Urban
Institute letter including the link, logname, and password provided to complete the survey online (see Appendix E for a sample of these letters). This letter will also inform respondents that a
paper copy of the survey and prepaid return envelope are available upon request. The content of
the surveys will be the same in either mode. An email reminder will be sent to those who have
not responded within 4-5 days. This should result in an increase in returned surveys. Once the
return rate has again significantly declined (usually 3-4 days), nonrespondents will be emailed
the link to the survey with a shorter cover letter and a hardcopy reminder postcard that
emphasizes the importance of the study. A second email reminder will be sent within 4-5 days to
those who still do not respond, followed by telephone reminders from Urban Institute staff.
Nonrespondents will be contacted at different times and days of the week than in the initial
mailing.
FSP Modernization Case Studies. It is expected that all states and counties within states
selected for the case studies will agree to participate in the study (i.e., a 100 percent response
rate). To ensure that all key respondents are interviewed, site visitors will work closely with a
person assigned to be the primary contact person (i.e., the assigned project liaison at the state
level and someone who is identified to assume this role at the local site level) to help in
scheduling the site visit. One member of the two-person site visit team will take responsibility
for working with the primary contact person to handle the scheduling and logistics of the site
visit. For the site visits, the logistical discussion will include recruitment of focus group
participants and arranging a time and location for the focus groups. Dates for site visits will be
made at least one month ahead of time to permit ample time to schedule interviews. Scheduling
interviews will occur during the month before the site visit, and interviews will be reconfirmed
with the site contact via email or fax 2-3 days prior to the visit. We will request that a quiet
setting that is as private as possible (e.g., a conference room) be made available to interview
those who do not have private offices, in order to encourage respondents to feel they can talk
freely. Based on our experience, following these established field visit protocols leads to an
interview completion rate approaching 100 percent of those scheduled in advance.
For those states or counties using contractors as part of their modernization initiatives, we
will ask the state or local liaison to identify and provide contact information for appropriate
individuals at the contractor organization. If the liaison prefers to make the initial contact, we
will proceed in that manner. While we do not anticipate any difficulty speaking with
contractors, if the liaison advises against an interview with the contractor, we will inquire as to
the reasons (e.g., the state or county may be in the process of contract negotiations).
For the focus groups of FSP participants and nonparticipants, each group will consist of
7-12 individuals. FSP participants will be identified after consultation with local FSP agency
staff. For the focus groups composed of FSP participants, we will rely primarily on lists
provided by the local FNS office of newly approved or recertified participants to recruit

24

November 13, 2007

participants. FSP staff or site visitors, depending on the preference of the site, will conduct
actual recruitment of FSP participants. Eligible nonparticipants will be identified with the
assistance of local programs and organizations that are likely to serve populations who are
eligible for the FSP, such as area agencies on aging, WIC programs, agricultural extension
service programs, and food banks. In essence, this strategy requires recruiting a local-level
organization to assist in the focus group recruitment process. We have found this strategy to be
successful in recruiting hard-to-reach populations (e.g., immigrants) because these organizations
have both the contacts and the trust of the population we want to interview.
To ensure that the desired number of people attends the focus group sessions (i.e., 7-12
people) we will need to over-recruit to allow for the incidence of no-shows. We estimate that we
will need between 16-20 people to agree to attend a focus group session to ensure a group of
sufficient size. An easy to read and colorful one-page flyer will be developed that local agencies
can post in their offices and disseminate to potential focus group candidates. A small monetary
incentive will be offered to potential focus group participants. Once focus group candidates are
identified, they will be contacted by telephone to fully inform them about the purpose of the
focus group and determine whether they are willing to participate. These candidates will be sent
a personalized confirmation letter and then called by telephone 2-3 days prior to the focus group
to remind them about the session and address any outstanding questions or concerns.
Brief intercept interviews with food stamp applicants and clients will be conducted at
local offices, with an expected response rate of 80 percent of those with whom contact is made.
These meetings are not scheduled in advance and are completely voluntary. Several techniques
will be used to maximize the response rate. One site visitor will individually approach the
prospective intercept interview candidate, introduce him/herself and organizational affiliation,
provide a concise description of the study (including an easy-to-read description of the study),
the specific purpose of the interview, and emphasize the brevity and voluntary nature of the
interviews. To allay any potential concerns on the part of the prospective participant that could
lead to lower response rates, an FSP staff person will be designated in advance as an official
agency contact with whom prospective participants can speak to confirm the purpose of the
survey and to answer questions.

4.

Tests of procedures or methods to be undertaken

Pretesting of both the survey and case study instruments was conducted. The following
describes these pretesting activities.
Pretesting of the Survey Instruments. The survey instruments (Appendix A) were
pretested by three states, two counties, and three community-based organizations in August
2007. The states and counties were identified with the assistance of FNS Regional Office staff.
States and counties were selected to include both county- and state-administered Food Stamp
Programs. States, counties, and partner organizations with a range of modernization experiences
were selected.

25

November 13, 2007

The three state agencies participating in the pretest were:
Minnesota Department of Human Services
444 Lafayette Rd.
St. Paul, MN 55155
New York State Office of Temporary and Disability Assistance
40 N. Pearl Street
Albany, NY 12243
Kentucky Cabinet for Health and Family Services
275 East Main St., 3E-I
Frankfort, KY 40621
The two participating county agencies were:
Bibb County Division of Family and Children Services
456 Oglethorpe St.
Macon, GA 31201
Hennepin County Human Service and Public Health
330 South 12th St.
Minneapolis, MN 55404
The three community-based organizations participating in the pretest were:
Food Bank of Central New York
6960 Schuyler Rd.
East Syracuse, NY 13057
Greater Philadelphia Coalition Against Hunger
2901 W. Hunting Park Ave.
Philadelphia, PA 19129
Second Harvest Food Bank of Santa Cruz and San Benito Counties
800 Ohlone Parkway
Watsonville, CA 95076
In addition, staff from the GAO; the Washington, DC Department of Human Services, Income
Maintenance Administration; and the California Association of Food Banks reviewed and
commented on the instruments during the pretest phase.
Initial phone calls soliciting participation in the pretest were made in late July and early
August 2007. Advance letters and surveys were sent via e-mail on August 8th. Respondents
were asked to return completed surveys to the Urban Institute within 10 days. Debriefing calls

26

November 13, 2007

with respondents took place the following week. Results of the pretest and comments were used
to refine the survey instruments and study procedures. In particular, we clarified reference
timeframes, reduced the number of response items for certain questions, and eliminated
redundant questions.
Pretesting of the Case Study Instruments. The case study instruments (Appendix B)
were pretested during spring 2007. Four states were visited and respondents were administered
the semi-structured interview guides, applicant/participant intercept interview guides, and the onsite observation guide. The staff that conducted the site visits provided feedback on the
instruments on aspects such as timing, redundancy, and gaps in content and refined the case
study instruments accordingly. Overall, they found that the semi-structured questions for the
state and local guides worked well. A few areas were identified where some refinements were in
order, mostly having to do with flow and emphasis rather than specific wording of questions. For
example, when using the state and local interview guides, it is important to take the lead from the
respondent with respect to the order of the discussion and modernization initiatives that the
state/local agency respondent believes are most significant or represent the greatest changes in
the way they do business. The discussion guides then can be more effectively used to probe to
make sure the full range of possible modernization efforts about which are asked.
The following state FSP agencies were visited during this initial phase:
Commonwealth of Massachusetts
Department of Transitional Assistance
600 Washington Street
Boston, MA 02111
Department of Workforce Services
140 East 300 South
Salt Lake City, UT 84111
Department of Social and Health Services
Economic Services Administration
Division of Employment and Assistance Programs
1009 College St., SE
P.O. Box 45470
Olympia, WA 98504-5470
Division of Health Care Financing
WI Department of Health and Family Services
1 West Wilson Street, Room 365
Madison, WI 53701

27

November 13, 2007

5.

Individuals consulted on statistical aspects of the design and the contractor

The agency responsible for receiving and approving contract deliverables is:
The Food and Nutrition Service
U.S. Department of Agriculture
3101 Park Center Drive, Room 1014
Alexandria, Virginia 22302
Person Responsible: Rosemarie Downer
(703) 305-2129
[email protected]
All data collection and analysis will be conducted by:
The Urban Institute
2100 M Street, NW
Washington, DC 20037
Person Responsible: Carolyn O’Brien, Principal Investigator
(202) 261-5624
[email protected]

28