PART A.final

Contents

Mathematica Policy Research

CONTENTS
PART A: SUPPORTING STATEMENT FOR PAPERWORK ACT SUBMISSION .......................... 1
1.

Circumstances Necessitating the Data Collection ..................................... 1

2.

How, by Whom, and for What Purpose the Information Is to Be Used........ 2

3.

Use of Improved Technology to Reduce Burden ....................................... 8

4.

Efforts to Identify Duplication .................................................................. 9

5.

Methods to Minimize Burden on Small Businesses or Entities ................... 9

6.

Consequences of Not Collecting the Data ................................................ 9

7.

Special Data Collection Circumstances ................................................... 10

8.

Federal Register Notice .......................................................................... 10

9.

Respondent Payments ........................................................................... 11

10. Privacy................................................................................................... 13
11. Questions of a Sensitive Nature ............................................................. 17
12. Hour Burden of the Collection of Information ........................................ 18
13. Estimated Total Cost Burden to Respondents and Record Keepers ......... 18
14. Estimated Annualized Cost to the Federal Government .......................... 19
15. Changes in Burden ................................................................................ 19
16. Publication Plans and Project Schedule .................................................. 19
17. Reasons for Not Displaying Expiration Date of OMB Approval ................ 20
18. Exception to the Certification Statement ................................................ 20
REFERENCES ............................................................................................................... 21

iii

Contents

Mathematica Policy Research

TABLES
A.1

Data Elements and Use ..................................................................................... 6

A.2

Burden Estimates for COBRA Subsidy Study Respondents ................................ 18

A.3

Cost to the Federal Government ...................................................................... 19

A.4

Timeline for Data Collection and Publication ................................................... 20

Appendix A: 60-Day Federal Register Notice

Appendix B: Public Comments and Responses

Appendix C: Memorandum on the TAA Experiment

Appendix D: Draft Advance Letter and Study Information Sheet

Appendix E: Draft IVR Screening Questionnaire

Appendix F: Draft COBRA Subsidy Study Survey

iv

06859 OMB Part A

Mathematica Policy Research

PART A: SUPPORTING STATEMENT FOR PAPERWORK ACT SUBMISSION
The U.S. Department of Labor (DOL) contracted with Mathematica Policy Research to
conduct an evaluation of the impact of a subsidy for health benefits under the Consolidated
Omnibus Budget Reconciliation Act (COBRA) that was provided by the American Recovery and
Reinvestment Act (ARRA) of 2009. The subsidy was available to workers who experienced
involuntary termination of a job from September 2008 to May 2010, were eligible for COBRA at the
time of job loss, and were not eligible for certain other health insurance options. The overall aim of
the Mathematica evaluation is to determine whether and how people who had employer-sponsored
health insurance maintained health care coverage after employment termination and whether the
COBRA subsidy provided by ARRA led to increased health care coverage. DOL is requesting
Office of Management and Budget (OMB) clearance for approval to conduct a one-time survey of
randomly selected unemployment insurance (UI) recipients (COBRA Subsidy Study Survey) as part
of this evaluation.
1. Circumstances Necessitating the Data Collection
COBRA offers workers and their dependents with employer-sponsored health benefits from
firms with 20 or more employees the opportunity to extend their coverage for limited periods of
time if they lose their benefits due to such circumstances as voluntary or involuntary job loss,
reduction in the hours worked, death, divorce, and other life events. COBRA continuation coverage
laws are administered by several agencies, including DOL, Health and Human Services, and the
Treasury Department. DOL’s interpretive and regulatory responsibility is limited to the disclosure
and notification requirements of COBRA. DOL is involved in the enforcement of COBRA but does
not collect program data regarding its utilization or the characteristics of those enrolling in it, nor
does any other government agency.
ARRA, as amended most recently by the Continuing Extension Act of 2010, provided for
temporary premium reductions. Eligible individuals paid only 35 percent of their COBRA
premiums, with the other 65 percent reimbursed to the coverage provider through a tax credit. To
qualify, individuals must have experienced a particular COBRA qualifying event—involuntary
termination of a covered employee’s employment—and not otherwise have been eligible for group
health insurance or Medicare. The involuntary termination must have occurred between September
1, 2008, and May 31, 2010. The premium reduction applied to periods of health coverage that began
on or after February 17, 2009, and could be claimed for up to 15 months. In particular, eligible
individuals who lost a job from September 1, 2008, to February 17, 2009, could claim the subsidy for
premium reductions only for periods of insurance that began on or after February 17, 2009.
Official statistics on the number of individuals that have participated in the subsidy program are
not available. Because the Internal Revenue Service (IRS) processes payroll tax data for premium
reduction withholdings, it can provide the number of policyholders that took the subsidy. However,
data on the number of dependents enrolled, the share of the population that is eligible for and
utilized the subsidy, the number of months enrollees received the subsidy, and the length of time
they were enrolled in COBRA all remain lacking. The IRS data can be used to calculate the direct
cost of the subsidy, but detailed data on individuals are required to study its impact, including who
benefits from the subsidy and how people make insurance decisions after job loss. No data currently
exist that can be used to study these individual characteristics.

1

06859 OMB Part A

Mathematica Policy Research

To fulfill this data limitation, DOL is sponsoring a study focusing on whether people who lost
their jobs maintained health care coverage and whether they used COBRA. The analysis will
describe the factors associated with being eligible for the subsidy, providing insights into the types of
people who had the potential to benefit from it. The study will document COBRA and other health
insurance enrollment decisions among individuals experiencing job loss and how these decisions
relate to characteristics of the individuals and their families. This will provide a picture of how
people make these health care decisions and the impediments they might face in maintaining health
care coverage for themselves and their dependents when employer-sponsored coverage is no longer
available.
The study will also provide an estimate of the impact of the availability of the subsidy on
COBRA enrollment, health, and employment outcomes. Measuring the impact of the subsidy and
developing a better understanding of health insurance decisions after job loss will enable DOL to
evaluate the efficacy of the subsidy and inform future policies aimed at increasing health insurance
coverage.
The study requires collecting information on people who lost a job and were eligible for
COBRA at the time, including both people who did and people who did not qualify for the subsidy.
Key domains in which data will be collected include employment, health insurance coverage, health
and health care utilization, financial well-being, and knowledge of and participation in COBRA.
This information will be collected in a one-time survey, the COBRA Subsidy Study Survey,
administered to people who lost a job and were eligible for COBRA. As described in detail in the
next section, the survey will focus on three groups of COBRA-eligible individuals, defined by
subsidy-eligibility status. The study’s sample frame, described in detail in section 1 of Part B, requires
that people be screened to determine whether they were COBRA-eligible at the time of job loss and
will therefore be included in the survey. It will take about 6 months to obtain the 5,800 completed
surveys needed to address the study’s research questions, with fielding initiated upon receipt of
OMB approval.
2. How, by Whom, and for What Purpose the Information Is to Be Used
a.

Overview of the Evaluation

DOL sponsored this study to learn about the effects of the ARRA subsidy on COBRA
coverage and duration, and to paint a broad picture of potential COBRA recipients and their health
coverage status. The study will answer several questions:
1. What is the impact of the subsidy on COBRA take-up and other outcomes such as health,
duration of health insurance coverage, and unemployment duration?
The impact analysis is a key measure of the efficacy of the subsidy. This information could be used,
for example, to calculate the number of people who enrolled in COBRA but would not have done
so without the subsidy. Phrased somewhat differently, the impact analysis will be able to capture the
incremental change in take-up rate in COBRA resulting from the offer of the subsidy. The analysis
will estimate the direct effect of the subsidy’s availability on COBRA enrollment, as well as its
indirect effect on employment, health, and other outcomes that may be affected through enrollment
in COBRA. The study will also examine how the impact varies by worker characteristics, such as
income and health status.

2

06859 OMB Part A

Mathematica Policy Research

2. What are the characteristics of COBRA-eligible and subsidy eligible individuals?
Documenting the number and characteristics of people eligible for COBRA and people eligible for
the subsidy will provide evidence on which workers are most likely to benefit from each program.
Since the subsidy can be claimed only by COBRA-eligible individuals, this analysis will provide a
context for understanding the potential of the subsidy to reach particular groups.
3. What are the characteristics of COBRA enrollees who experienced job loss?
Comparing the characteristics of COBRA enrollees to those of eligible nonenrollees provides direct
information about who benefits from COBRA. Describing enrollees and identifying predictors of
enrollment will help provide a context for understanding individual and family health insurance
decisions. Survey respondents’ stated reasons for enrolling or not enrolling in COBRA and their
choices of alternative health plans provide additional insight into the enrollment decision from the
perspective of the worker.
b. Data Needs
A unique set of data is needed to detect the impacts of the offer of the subsidy on COBRA
coverage, and to describe the behaviors and decision making of COBRA-eligible individuals. Data
are needed on (1) people’s socioeconomic, demographic, and household characteristics; (2) people’s
knowledge and awareness of COBRA and the subsidy and how they obtained this information; (3)
key outcomes that potentially could be affected by the subsidy, including COBRA take-up and
enrollment duration, as well as other health and employment outcomes; and (4) predictors of these
outcomes, including cost of enrolling in COBRA, ability to pay (e.g., income, debt), preferences
(e.g., health at time of job loss, demographics), quality of the COBRA coverage (i.e., plan
characteristics), availability of alternative coverage options, and environmental factors such as
characteristics of the firm from which the worker separated and the local area in which the worker
lives.
Because information on the groups of interest is not readily available in any existing survey or
administrative data, the COBRA Subsidy Study Survey will be the primary source of data for the
study. A random sample of UI recipients will serve as the basis for identifying, in the target
population of workers, individuals who lost a job and became COBRA-eligible. The sample will be
further divided into three groups targeted for data collection: (1) a subsidy-eligible sample comprising
people who were eligible for the ARRA COBRA subsidy; (2) a subsidy-comparison sample consisting of
people who would have been eligible for the subsidy except for the timing of the job loss; and (3) a
subsidy-ineligible sample made up of job losers who were eligible for COBRA but did not meet the
subsidy criteria for reasons other than timing of job loss. Data collection will focus on individuals
who lost their job (1) during the period in which the subsidy was in place (February 17, 2009, to May
31, 2010), to collect information from the subsidy-eligible population; and (2) on the period after the
subsidy ended (June 1, 2010, to March 31, 2011), to collect information from COBRA-eligible
individuals who would have been eligible for the subsidy except for the timing. Data collection for
the subsidy-ineligible sample will focus on individuals from both time periods.
Data collection for the study will take place over a six-month period beginning as soon as OMB
clearance is received. Since not all UI recipients belong to the target population, the study will use a
two-stage data collection process. Stage 1 will screen sample members for eligibility for the survey
and determine to which of the study groups an individual belongs, and Stage 2 will collect the data
required for the evaluation. In Stage 1, the random sample of UI recipients will be asked whether
they had employer-sponsored health insurance at the time they lost their job; this is to determine
3

06859 OMB Part A

Mathematica Policy Research

whether the person is eligible for COBRA, a requirement for participation in the full survey. Before
being screened out, people who did not have coverage will be asked if their employer offered
coverage and, if so, whether they were eligible for it. This information will provide insights into why
job losers are not eligible for COBRA. People who did have coverage (i.e., those eligible for
COBRA) will be asked questions about Medicare and availability of other coverage to determine
which of the three study groups they belong to.
Sample members who pass the screener will be administered the full survey until the target
sample sizes of each study group are reached. The full COBRA Subsidy Study Survey includes a set
of modules that encompass demographic characteristics; employment and job search; income,
program participation, and financial well-being; and, most important, health care use and type,
knowledge of COBRA health insurance and of the ARRA COBRA subsidy, health status of the
worker and his or her family, and unmet health needs. Below are the types of data elements that will
be collected in the survey.
• Demographic, household, and area characteristics include items such as age, gender,
race/ethnicity, education, marital status, household composition and size, and
characteristics of the local area. 1 This information will be used to describe the samples,
and as factors that could influence the decision to enroll in COBRA coverage.
• The employment and work search modules will be used to collect information on the
characteristics of the job prior to separation and subsequent jobs, including such
information as hours worked and earnings, industry and occupation, benefits offered on
the job, reasons for job separation, whether the person looked for work and reasons for
not looking, time spent looking for employment, and reasons for rejecting offers
received. Information on employment prior to job loss will be used to describe the
samples and as predictors of outcomes, while information on employment after job loss
will be used as outcomes. Information will also be collected about expectations of
finding employment at the time of job loss.
• Income and its sources includes items such as spousal earnings; income from interest,
dividends, and rent; income received from participation in various programs, including
UI, food stamps (or Supplemental Nutrition Assistance Program), welfare or general
assistance, Social Security at the time of job loss; and monthly income at the time of job
loss. In addition to its influence on COBRA enrollment, this information will be used to
determine subgroup impacts by income and poverty status at the time of job separation.
Furthermore, information on participation in programs after job loss will be collected
and used as an outcome measure.
• Financial well-being includes items such as the types of loans, trouble making bill or
loan payments, food insufficiency, and amount of savings. When these items are
measured at the time of job loss they represent factors that could influence COBRA
take-up. When they are captured twelve months after job loss, they reflect outcomes that
could be affected by loss of health insurance (and hence might be influenced indirectly
by the offer of the subsidy).
1 Characteristics about the local area will be obtained by linking the zip code information in the UI administrative
data to secondary data sources.

4

06859 OMB Part A

Mathematica Policy Research

• Health and health care utilization include items such as self-reported health status,
use of medical care before and after job loss, and the individual’s own and family
members’ health insurance coverage before and after job loss. Some measures, such as
health status and use of medical care prior to job loss, might influence COBRA take-up.
Others, such as health status and use of medical care after the time of job loss, reflect key
outcomes that could be influenced by the availability of the subsidy.
• COBRA and ARRA subsidy awareness and eligibility include items such as
knowledge of the program, whether the worker was aware of various details of each
program and whether they were eligible, how he or she was informed, whether the
information was easy to follow, and reasons for participating or not participating. This
information will provide a context for understanding how people learn about the
programs, and other factors that might be related to why they did or did not use
COBRA coverage or use the subsidy.
c

Purpose of the Data

Table A.1 links the data elements to their source—survey, screener, or UI data—and primary
use in the study. The table provides a justification for why survey questions capturing items in each
data element were developed for the survey and how they will be used in the study’s analysis. Data
will be collected for one of five uses: (1) descriptive/contextual; (2) covariate/background; (3)
outcomes; (4) key impact outcome; and (5) stratification. Information collected for
descriptive/contextual purposes will provide a context for interpreting the impact of the subsidy and
the factors correlated with COBRA take-up, while information collected for covariate/background
purposes will be used as correlates of the health insurance decisions. Information collected as
outcomes will be used to highlight the myriad of ways in which the subsidy’s availability might affect
the lives of individuals who lost their job. Information collected as key impact outcomes will be
formally modeled in the impact analysis as influenced directly or indirectly by the ARRA subsidy.
Some of the covariate/background information will also be used as categories of subgroup analysis.
Depending on the nature of the data elements, variables will be measured at one or more of
three points in time—at the time just prior to job loss, at six or twelve months after job loss, and the
time between job loss and time of survey. The data elements measured just prior to job loss are
intended to capture the factors that are likely to have influenced the recipient’s health insurance
decisions, and that provide background information on the sample and that can be used as
covariates in our analyses. Data elements measured at six or twelve months after job loss will
capture the more proximal changes in individuals’ circumstances that result from job loss and
potential loss of employer health insurance – including health care utilization and financial wellbeing. Data elements that capture experiences from the time of job loss to time of survey include
key outcomes such as employment and health insurance coverage that are likely to be influenced by
the offer of the subsidy and will also help inform the cost-benefit of the offer and take-up of
subsidies. As shown in Table A.1, data elements measured after job loss will generally be used as
outcomes while those measured at job loss will generally be used as covariates in multivariate
analysis to predict outcomes or as descriptive, background measures to provide characteristics of
individuals who were eligible for the subsidy (for example).

5

6

Descriptive/
Contextual

Covariate/
Background

Survey

X

Survey
UI data

X
X

Survey
UI data

X

Outcomes

X
X

Survey

X

Survey
Survey

X
X

Survey

X

Survey

X

Survey
Survey

X

X
X

X
X
X
X
X

Survey
Survey
Survey
Survey
Survey
Survey
Survey
Survey

Stratification
Variable

X

Survey

Survey
Survey
Survey

Key Impact
Outcomes

X
X
X
X
X
X
X
X

X

Mathematica Policy Research

Demographic, household, and area characteristics
Demographic characteristics: Age, gender, race and ethnicity, education, marital
status
Household characteristics: Household composition and size, children outside the
house, employment status of adult household members
Area characteristics: Zip code most recent address(used to link to secondary data)
Employment and work search
Employment before job loss
Earnings, hours worked, occupation, industry, benefits offered, separation reason,
firm size < 20, seasonal/temporary job, represented by union, company move or
close
Date of job separation
Employment after job loss
Duration of initial unemployment spell
Other job characteristics (after UI claim): Earnings, hours worked, occupation,
industry, benefits offered, start and stop date, separation reason, firm size < 20,
seasonal/temporary job, union member, represented by union
Current employment: Earnings, hours worked, benefits offered, seasonal/temporary
job, Occupation, industry, firm size < 20, union member, represented by union?
Work search activity: Looked for work, reasons not working and rejecting job offers
Income and its sources
Income at job loss
Monthly income prior to job loss
Spousal earnings, interest, dividends, rent, program participation, welfare, Social
Security, etc. (including sources)
Income after job loss
Number of months and average income from each source
Current monthly income
Financial well being
Time of job loss
Living arrangement (own home, rent, etc.)
Type and amount of debt
Savings: for 3 months living expenses, 6 months, total
12 months following job loss
Trouble making bill/loan payments (which ones)
Moved, sold valuables, withdraw from retirement savings to pay bills
Bankruptcy, home foreclosure
Food insufficiency
Health and health care utilization
Health at time of job loss
Self reported for self and family
Conditions that limited work
Chronic health conditions (self and family)
Pregnancy

Information
Source

06859 OMB Part A

Table A.1 Data Elements and Use

Descriptive/
Contextual

Covariate/
Background

Outcomes
X
X

Key Impact
Outcomes

Stratification
Variable

X

X
X
Screener
Survey
Survey
Survey

X

X
X
X

Survey

X
X

Survey

X

Survey

X

Survey
Survey
Survey
Survey

X
X

X

X

X

X

Screener
Survey
Survey
Survey
Survey

X
X
X

Survey
Survey

X
X

Survey

X

X

06859 OMB Part A

7

Health at the time of the survey
Self rating of health
Condition that limits the amount of work
Health care use 6 months after job loss
Physician visits increase, decrease, same (as 6 months prior to job loss)
Prescription medicine use increase, decrease, same (as 6 months prior to job loss)
Health care coverage at job loss
Covered by employer’s plan (reason why not)
Adequacy of plan, whether covers doctors want to see
Monthly premium
Covered by another plan (besides employer’s plan)
Health care coverage since job loss
Enroll in any health insurance coverage immediately after job loss
Current coverage, start date, monthly premium, eligible for Medicare
Other coverage since job loss: source, reason end, gaps (self and family), months,
medical needs and financing
Change in health status (6 months before/after job loss): self or family, doctor visits,
use prescriptions, emergency room visits
COBRA and subsidy awareness
COBRA
Enrollment in COBRA after job loss
Family members enrolled in COBRA
Reasons participating (or not)
Duration of COBRA enrollment
Reasons not eligible for COBRA (ineligible for employer-sponsored insurance,
eligible but not enrolled)
Reason ended
Coverage without COBRA
Monthly premium
Would enroll in COBRA if: 90%, 80%, 65%, 35% subsidy
COBRA awareness: knowledge of, familiarity with, how informed, ease of
information
ARRA subsidy: use assistance, start and stop date, reason not take up subsidy
ARRA subsidy awareness: knowledge of, familiarity with, how informed, ease of
information

Information
Source

X

X

Mathematica Policy Research

Notes
Data elements marked in the “Descriptive/Contextual” column will be used to provide a context for interpreting the impact of the subsidy and
the factors correlated with COBRA take-up. These data include stated reasons for health insurance decisions, reasons for COBRA or subsidy ineligibility,
and knowledge of the programs. The “Covariate/Background” column indicates factors determined at or before the time of job loss that may be
correlated with health insurance decisions. These will be used both in descriptive analyses comparing groups based on COBRA or subsidy eligibility
status, and as covariates in the impact analysis. Data elements marked in the “Outcomes” column are variables measured after the time of job loss. The
distributions of these variables will be compared among the study groups. A subset of these, marked as “Key Impact Outcomes” will be used as
outcomes in the formal impact analysis since they are likely to be impacted directly or indirectly by the availability of the subsidy. The impact analysis
will be performed on subgroups defined based on data elements marked in the “Stratification Variable” column.

06859 OMB Part A

Mathematica Policy Research

d. Who Will Use the Information
DOL will use the information from the data collection to assess the effect of the ARRA subsidy
on COBRA enrollment. The information will answer DOL’s questions about impacts of the benefit
of the subsidy on the health and employment of those that involuntarily lose their job. DOL will
also use the new information about the broader COBRA-eligible population to gain a better
understanding of the decision to enroll in COBRA coverage and of the outcomes associated with
not enrolling.
Data collected will be released as a public use file without personally identifiable information, so
that public policy and social science researchers will have access to the rich source of data collected.
Additional research based on this data set may expand on this study’s initial goals, and further
analyses tailored to a particular situation may inform specific policy decisions related to health
insurance coverage. As a result, research used to develop policy initiatives to promote health care
coverage and positive health outcomes will be grounded in a stronger knowledge base. The public
use data file will be formatted to the specifications of data.gov for uploading.
Ultimately, these data will benefit researchers, policy analysts, and policymakers in a wide range
of program areas. The data collection will enable the study to create a better understanding of the
health insurance decisions of people who lose a job. Measuring the effect of the ARRA subsidy on
COBRA enrollment may inform the efficacy of other public programs aimed at increasing health
care coverage. Additional analyses will provide valuable information on potential barriers to health
care enrollment and on the types of individuals and families at greatest risk of becoming uninsured.
3. Use of Improved Technology to Reduce Burden
A dual-mode approach will be used to facilitate the anticipated high volume of screening
required to identify the survey sample. Sample members will be screened either (1) by calling in to a
state-of-the-art interactive voice response (IVR) system; 2 or (2) by speaking with a Mathematica
interviewer. The IVR was selected because it will help screen out large numbers of people who
would not have been eligible for COBRA coverage in the relevant time periods and thus decrease
the costs of contacting sample members. The IVR system may also increase response rates by
appealing to a subset of sample members who prefer this option and might not respond to
interviewer-initiated contact attempts. It is also helpful in connecting with sample members for
whom the telephone contact information obtained from the UI administrative records is invalid,
since some of these sample members will call in on their own.
The IVR pre-screening interview is designed to place minimal burden on the respondent
because it is easy to understand and can be completed in about two minutes. At the start of the IVR
call, a brief introduction that has been pre-recorded will be played for the sample member. In
addition to project specific information, the introduction will contain important IVR information,
such as at any point during the call, please press the “*” button to be transferred to an interviewer.
After answering a few questions to confirm their identity, sample members will answer one prescreening question to determine their eligibility for the study. Sample members who do not pass the
screen will be asked two additional questions to gather some information about their ineligibility for
IVR technology involves the use of a touch-tone telephone to interact with a database to enter into or acquire
information from the database.
2

8

06859 OMB Part A

Mathematica Policy Research

COBRA. These questions all require a yes or no response, and the IVR will be programmed to use
appropriate skip logic. Ineligible sample members will be thanked for calling in and told that they
have not been selected for the study.
Sample members who are “screened in” to participate will be transferred to an interviewer at
Mathematica’s Survey Operations Center (SOC) to complete the screening process and full
interview, if appropriate. Prior to any transfer to an interviewer, respondents will be asked to enter
their 10-digit telephone number in case they are disconnected. A dedicated line at the SOC will
accept these IVR transfer calls so that interviewers are alerted to the source of the call and will enter
a code that ensures that the proper question fills and paths are followed. If the transfer is not during
hours when interviewers for the project are staffed, the sample member will receive a message that
an interviewer will call back when the SOC re-opens, and will request that the sample member enter
the best time when they can be contacted. The SOC will also receive daily data transfers from the
IVR system that will allow for identification of sample members who completed the screener but did
not leave a voicemail message or sample members who did not complete the screening questions.
The IVR will be administered by Interviewing Services of America (ISA), a company
experienced in developing and administering IVR systems. All respondents who indicate that they
were covered by employer-sponsored health insurance at the time of their job loss will be
transferred to the SOC to complete the screening, verification, and interviewing process. Sample
members who report no such coverage will be screened out after reporting whether their employers
offered health insurance and their eligibility for it. Sample members who opt to call in to speak with
a live interviewer (or who are called by an interviewer) will complete the same pre-screening
questions as those who call into the IVR. These questions, and the full interview, will be conducted
with computer-assisted telephone interviewing (CATI). CATI will work efficiently in conjunction
with the IVR, as sample members who are identified as potentially eligible by the IVR will be able to
continue the interview in CATI immediately upon transfer to Mathematica’s SOC. The IVR was not
considered for the full survey for several reasons, including the length of the interview, inclusion of
open-ended questions, and the need for probing.
CATI was selected as the data collection mode for the full survey because telephone interviews
are more cost-effective and less burdensome on respondents than in-person interviews, given the
flexibility they allow for scheduling. CATI also increases efficiency and decreases burden on
respondents compared to paper-and-pencil interviewing because it accepts only valid responses and
can be programmed to check for logical consistency across answers. Interviewers are thus able to
correct errors during the interview, eliminating the need to call back respondents to clarify or obtain
missing data. To aid respondents’ recall of events, the CATI program will be preloaded with useful
information from the UI administrative data file, such as the job separation date, UI claim date, and
the name of the employer associated with job separation shortly before the initial UI claim. CATI is
also beneficial because it virtually eliminates dialing errors by making calls through an auto-dialer
linked to the system. The automated call scheduler will simplify scheduling and rescheduling of calls
to respondents at their convenience and can assign cases to specific interviewers, such as those
fluent in Spanish.
4. Efforts to Identify Duplication
An estimate of the impact of the ARRA subsidy on enrolling in COBRA and on health and
other outcomes can be computed using data collected. Information needed to estimate such impacts
goes far beyond what is available. No other data collection effort has been conducted or planned
that would provide the information needed to estimate this impact. Further, to avoid duplication in
9

06859 OMB Part A

Mathematica Policy Research

with other ongoing studies of the UI population this study has coordinated efforts with
Mathematica’s DOL-funded evaluation of the Unemployment Compensation Provisions of ARRA.
5. Methods to Minimize Burden on Small Businesses or Entities
Small businesses and other small entities are not required for this data collection effort.
6. Consequences of Not Collecting the Data
Data will be collected only once. The COBRA Subsidy Study Survey will provide the only
source of data for UI recipients on eligibility for and familiarity with COBRA and familiarity with
the ARRA subsidy in addition to data on health and health insurance before and after job loss. If the
COBRA Subsidy Study Survey were not conducted, the impact of the ARRA subsidy for COBRA
enrollment would remain unknown as would a description of who used the subsidy.
7. Special Data Collection Circumstances
All federal guidelines will be strictly adhered to in the collection of all data required for the
COBRA Subsidy Study. There are no special circumstances that would require straying from such
strict adherence.
8. Federal Register Notice
a.

Federal Register Notice and Comments

As required by 5 CFR 1320.8 (d), a Federal Register Notice, published on December 12, 2011
(FR, Vol. 76, No. 238, pp. 77263-77264), announced the study of the Impact of the ARRA Subsidy
on COBRA Take-Up and provided the public an opportunity to review and comment on the
planned data collection and evaluation. Comments from this notice were due within 60 days of the
publication, in accordance with the Paperwork Reduction Act of 1995. A copy of this 60-day notice
is attached as Appendix A. A second Federal Register Notice will be published for a 30-day period,
coinciding with submission of this OMB clearance request, and will provide the public a second
opportunity to respond. No comments were received from the public during the initial 60-day
posting.
b. Consultations Outside the Agency
To ensure that the best decisions were made regarding the research, sample, and questionnaire
designs for the study as well as for the data collection plans, experts from outside the agency were
consulted, and their input has helped to shape the evaluation. These experts included project staff
from Mathematica and members of the project’s Technical Working Group. The experts consulted
are listed below, along with telephone contact information.
Mathematica Staff
Dr. Anu Rangarajan, Project Director
Ms. Julita Milliner-Waddell, Survey Director
Dr. Nathan Wozny, Researcher
Dr. Nan Maxwell, Senior Researcher
Dr. Frank Potter, Senior Fellow
Dr. Eric Grau, Senior Statistician

(609) 936-2765
(609) 275-2206
(609) 936-2795
(510) 830-3726
(239) 558-5956
(609) 945-3330
10

06859 OMB Part A

Mathematica Policy Research

Dr. Hanley Chiang, Researcher
Ms. Grace Roemer, Associate Director of
Princeton Survey Research
Ms. Yingying Xu, Program Analyst
Ms. Karen Cybulski, Senior Survey Researcher

11

(617) 674-8374
(609) 936-2782
(609) 750-4043
(609) 936-2797

06859 OMB Part A

Mathematica Policy Research

Members of the Technical Working Group
Dr. Randall Bovbjerg, Senior Fellow Health Policy Center,
Urban Institute
Dr. Jonathan Gruber, Professor of Economics,
Massachusetts Institute of Technology
Ms. Wendy Hamlett, Director, Massachusetts
Medical Security Program
Dr. Howard (Rocky) King, Senior Policy Advisor on
Health Care Reform, Oregon Dept. of Business
and Consumer Affairs
Dr. Brigitte Madrian, Professor of Public Policy,
Harvard University

(202) 261-5685
(617) 253-8892
(617) 626-6099
(503) 947-2345
(617) 495-8917

9. Respondent Payments
We will employ several proven methods to ensure a high response to the COBRA Subsidy
Study Survey. One such method is the offer of an incentive to survey completers, which can help
increase cooperation among sample members and thus help increase response rates. High response
rates, in turn, help achieve sample representativeness, which is critical to achieving high data
quality—that is, data that are complete, valid, reliable, and unbiased. Offering incentives can help
achieve these goals. The offer of incentives will be coupled with other methods known to increase
response, such as good techniques for locating mobile sample members; refusal avoidance;
persistent and professional contact protocols, including the use of agency letterhead for mail
contact; and high refusal-conversion rates. Declining response rates in telephone surveys (Curtin et
al. 2005) and the concomitant rise in effort and costs associated with achieving high response rates
have made the use of incentives a more common practice for survey studies. Evidence on the
benefits of offering incentives has become available in the last several years. Incentives can help
achieve high response rates by increasing the sample members’ propensity to respond (Singer et al.
2000). Studies offering incentives show decreased refusal rates and increased contact and
cooperation rates. Among sample members who do initially refuse to participate, incentives increase
refusal-conversion rates. By increasing sample members’ propensity to respond, incentive payments
have been found to significantly reduce both the number of calls required to resolve a case and the
number of interim refusals. Thus, incentive payments help contain costs, and pass some of the costs
of conducting the survey as a gain to the participant rather than into additional survey operations.
Last, while incentives help gain cooperation to increase the overall response rate, they also
increase the likelihood of participation from subgroups with a lower propensity to cooperate with
the survey request. This is an important consideration for ensuring the representativeness of the
survey respondents and the quality of the data being collected. For example, Jäckle and Lynn (2007)
found that incentives increased the participation of sample members more likely to be unemployed,
a key characteristic of the sample for the COBRA Subsidy Study. There is also evidence that
incentives bolster participation among those with lower interest in the survey topic (Schwartz et al.
2006; Jäckle and Lynn 2007; Kay 2001) resulting in data that is more nearly complete. In the case of
the COBRA Subsidy Study, sample members who are working and now insured, may be less
interested in participating in this study than those who are unemployed or uninsured; an incentive
offer will likely bolster participation among this important group. Further, paying incentives does
not impair the quality of the data obtained (such as item nonresponse or the distribution of
responses) from groups that would otherwise be underrepresented in the survey (Singer et al. 2000).
12

06859 OMB Part A

Mathematica Policy Research

An incentive payment will be offered to all respondents who are determined eligible and who
complete the full survey. No incentive will be offered or paid to respondents who complete only the
screening interview. A payment of $40 will be offered to respondents who use the IVR for initial
screening and complete the survey within four weeks of receiving the invitation, and $30 will be
offered to non-IVR completers or to IVR completers who complete the survey after the four-week
window. Because of the high volume of screening required (an estimated 22,000 to 26,000
interviews), and the fact that many sample members will not pass the eligibility screens, the higher
incentive offer is intended to encourage early use of the IVR. Increased use of the IVR will help to
reduce interviewing costs by eliminating time and labor spent attempting to reach households and
conducting the screening questions. In addition, it will reduce locating costs by encouraging call-ins,
as some sample members will have outdated contact information. The potentially significant cost
savings of screening sample members through the IVR makes it worthwhile to encourage as many
sample members as possible to use this system and justifies offering them a slightly higher incentive.
The importance of achieving a high response rate and preventing differential attrition across
research groups makes offering incentives a critical addition to our intensive efforts to successfully
establish contact with prospective respondents and gain their cooperation with the planned data
collection. To leverage fully the benefits of offering incentives in the COBRA Subsidy Study, the
incentive offer will be mentioned in our advance letter to sample members, making it clear that it
will be provided to “eligible survey completers.” To help gain cooperation, interviewers will also
mention the proposed incentive when they determine that a sample member is eligible for the study.
Mathematica has conducted several studies of the general UI claimant population as well as
studies for which the UI claimant population served as the sample frame. The planned incentive
amount for the COBRA Subsidy Study is on par with those studies which used similar methodology
and used the UI claimant population as a sample frame, as in the Impact Evaluation of the Trade
Adjustment Assistance Program (TAA study) for the DOL. For example, Mathematica recently
conducted an incentive experiment as part of the TAA study, in which non-respondent sample
members were offered one of three incentive amounts—$25 (the status quo), $50, and $75—in a
split-ballot experiment to evaluate the impacts of the incentive protocol. Responses were also
tracked by whether the sample member was in the TAA group that had received services or not, and
their comparison group counterparts. Finally, a number of procedural changes were also
implemented and their effectiveness assessed. Response rates increased from 41 percent to 55
percent overall after the higher incentives were offered, and increases ranged from 8.7 percentage
points to 15.8 percentage points for each of the four groups examined. 3 In particular, response rates
increased from 53.7 percent with an incentive offer of $25 to 62.4 percent after an increased
incentive offer of $50 to $75 for group A. Similar differences were observed for the other three
groups, with response rates increasing from 37.0 percent to 52.8 percent for Group B, from 42.7
percent to 57.2 percent for Group C, and from 34.4 percent to 50 percent for Group D. (See
Appendix C for a detailed memo describing the TAA experiment.)
Based on the results of the experiment, the incentive offer for the TAA study was increased to
$50 for both existing, non-responding sample members and for newly released sample members.
The differences in response between the $50 and $75 incentive groups were not significant. The
3 The four groups, referred to as A, B, C, and D, represent TAA participants, TAA-eligible nonparticipants, and
their respective comparison groups.

13

06859 OMB Part A

Mathematica Policy Research

proposed $30 and $40 incentive amounts for the COBRA Subsidy Study are, therefore within the
range of amounts that have been provided in other federal surveys of similar populations. In the
TAA study, much of the improved response came through an increase in the number of call-ins by
sample members.
10. Privacy
Participants will receive assurances of privacy in an advance letter and a study information sheet
describing the survey (see Appendix D) and again at the outset of the interview as part of the
introductory script. Sample members who call the IVR will also be told, in the prerecorded message,
that all answers will be kept private. In addition, if an IVR caller is connected to the SOC to
complete screening and interviewing, they will be transferred to a project-specific phone line where
their information will be verified and privacy assurances repeated.
Interviewers will be trained in privacy procedures and will be prepared to describe them in full
detail, if needed, and to answer any related questions raised by participants. A set of frequently asked
questions (FAQs) and responses has been developed to assist interviewers with inquiries. These
FAQs are incorporated into the CATI program and can be accessed by interviewers at any point in
the interview. They include responses to generic questions typically encountered in surveys as well as
questions and answers specific to the COBRA Subsidy Study. For example, if asked about how their
answers will be used, the interviewer will explain that individual answers will be combined with
those of other respondents and presented in summary form only and used for research purposes
only—for example, “Blank percent of respondents reported that they are now employed.” Further,
respondents will be reassured that their names will never be used in any reports or other documents
provided to DOL or any other agencies.
All data items that identify respondents will be kept only by the evaluator contractor
(Mathematica) and by ISA, for use in assembling records data and in administering the screening and
full interviews. Any data DOL receives will contain no personal identifiers, thus precluding
individual identification.
It is the policy of Mathematica to efficiently protect private information and data in whatever
medium it exists, in accordance with applicable federal and state laws and contractual requirements.
In conjunction with this policy, all Mathematica staff shall:
1. Comply with the company’s Confidentiality Pledge, which is signed by all full-time, parttime, and hourly Mathematica staff, and with Mathematica’s Security Manual procedures
to prevent the improper disclosure, use, or alteration of private information. Staff may
be subjected to disciplinary and/or civil or criminal actions for knowingly and willfully
allowing the improper disclosure or unauthorized use of private information.
2. Access private and proprietary information only in performance of assigned duties.
3. Notify their supervisor, the project director, and the Mathematica Security Officer if
private information has been disclosed to an unauthorized individual, or used or altered
in an improper manner. All attempts to contact Mathematica staff about any study or
evaluation by individuals who are not authorized access to private information will be
reported immediately to both the cognizant Mathematica Project Director and the
company’s Security Officer.

14

06859 OMB Part A

Mathematica Policy Research

In addition, many Mathematica staff, including some COBRA project staff, have received
security clearance from the Social Security Administration and are experienced with the stringent
security requirements of collecting sensitive and personally identifying information.
To allow external verification and replication of the study findings, as well as additional
research, a public use data file containing key analysis variables created for the COBRA Subsidy
Study will be produced at the end of the study. The public use file that is produced will follow the
current OMB checklist on privacy to ensure that it can be distributed to the general public for
analysis without restrictions. Steps will be taken to ensure that individual study participants cannot
be identified in indirect ways. For example, categories of a variable will be combined to remove the
possibility of identification due to a respondent being one of a small group of people with a specific
attribute. Variables will also be combined in order to provide summary measures to mask what
otherwise would be identifiable information. Variables we will carefully scrutinize include age, race/
ethnicity, household composition, location, household income and assets, and rare health
conditions. We will scrutinize these and other such variables for small cell sizes and outliers, as well
as combinations of variables that could potentially lead to any identification. Depending on the
nature of the responses we find, we will either top-code or categorize outliers, and/or combine data
in neighboring cells as needed. Statistical methods will be used to add random variation within
variables that would be otherwise impossible to mask, if necessary. Finally, variables that could be
linked to identifiers by secondary users will be removed. Additional research based on the public use
data set may expand on this study’s initial goals, and further analyses tailored to a particular situation
may inform specific policy decisions related to health insurance coverage.
a.

General System Security Description

Mathematica’s computer facilities include state-of-the-art hardware and software. The hardware
and software configurations have been designed to facilitate the secure processing and management
of both small- and large-scale data sets.
1.

Facility

The doors to Mathematica’s office space and SOC are always locked, and all SOC staff are
required to display current photo identification while on the premises. Visitors are required to sign
in and out and must wear temporary ID badges while on the premises. Any network server
containing private data must be in a controlled, limited-access area. All authorized external access is
through a server under strict password control.
2.

Network

Data stored on network drives is protected with the security mechanisms available through the
network operating system used on our primary network servers: Novell Netware 5 – 6.5. These
versions of Novell Netware are compliant with the C2/E2 Red Book security specifications.
Netware is certified at the National Computer Security Center’s Trusted Network Interpretation
Class C2 level of security at the network level. The network is protected from unauthorized external
access through the PIX Firewall from CISCO. This firewall resides between our network and the
communications line over which our Internet traffic flows.
Access to all network features, such as software, files, printers, Internet, email, and peripherals,
is controlled by userid and password. Network passwords must contain a minimum of eight
characters and must be a combination of numbers and letters. All userids, passwords, and network
15

06859 OMB Part A

Mathematica Policy Research

access privileges are revoked within one working day for departing staff and immediately for
terminated staff. All staff are required to log off the network before leaving for the day.
3.

Printers

Printer access is granted to all staff with a valid userid and password. The physical hard disks on
which the printer queues reside are subject to the same security/crash procedures that apply to the
file servers. Printer stations are appropriately monitored according to the sensitivity of the printed
output produced. No private or proprietary data or information can be directed to a printer outside
Mathematica’s offices.
4.

Electronic Communication

Ethernet is used for internal email communications over the network. As Ethernet
communications use Novell Netware with built-in userid and password protections and Windows
NT Challenge Handshake Authentication Protocols, sensitive information in both email text and
attachments can be safely transmitted. Email transfer is also encrypted when sent to or from the
Mathematica gateway facility, which allows staff to check and send emails from home. A dedicated
private line supports cross-office communications between Mathematica’s offices.
b. Treatment of Data with Personal Identifying Information
All data containing personal identifying information (PII)—including SSN, name, home
address, date of birth, and telephone number—are considered to be sensitive, or private, data. The
COBRA Subsidy Study is in compliance with the aforementioned company security policies. Listed
below are specific details regarding the handling and processing of private information in this
evaluation.
1.

Access

Private electronic files are stored in restricted-access network directories. Access to restricted
directories is limited on a need-to-know basis to staff who have been assigned to and are currently
working on the project. When temporarily away from their work area, project staff are instructed to
close files and applications. Access to their workstations locks within a set number of minutes, and
they must use a password to regain access through the protected screen saver.
2.

Electronic Communication

Staff are forbidden to transmit sensitive study information as a regular file attachment to an
internal email; they are instructed instead to use the “insert shortcut” feature in Outlook to include a
shortcut to the file. This allows the receiver to go to the file directly but will not allow access to
unauthorized individuals. In addition, staff are instructed to avoid including sample member names
or other PII in internal emails, so that there is no potential for these to be viewed by others.
Emails sent outside Mathematica are not automatically encrypted, and therefore neither the text
nor attachments are secure. Before sending an email containing sensitive information, the sender is
obligated to ensure that the recipient is approved to receive such data. When files must be sent as
attachments internally and outside Mathematica, staff are instructed to use WinZip 9.0 (256-bit AES
encryption) to password-protect the file. When sample member name and contact information are

16

06859 OMB Part A

Mathematica Policy Research

sent outside Mathematica, the information is included in a secure attachment rather than in the text
of the email.
3.

COBRA Subsidy Study Databases

Project databases containing private information are password protected and accessible only to
staff currently working on the project. To access the project’s database, users must first log onto
their workstations and then upon starting the database, log in again using a separate prompt. Project
databases will be removed and securely archived at the end of the data-processing period.
4.

Telephone Interviewing

Telephone interviewers for the COBRA Subsidy Study Survey will be seated in a common
supervised area. As part of the verification process, interviewers will have access to respondents’
names and birthdates, as well as the last four digits of their SSN. Birthdate and last four SSN digits
will be displayed on the computer screen only during the sample member verification. Interviewing
staff for this project receive training that includes general SOC security and privacy procedures, as
well as project-specific training that includes explanation of the highly private nature of this
information, instructions to not share it or any PII with anyone not on the project team, and caution
about the consequences of any violations.
5.

Locating

Staff who work on updating sample member contact information when the original contact
information is not successful must have access to key identifying information for short periods.
Locating staff receive training that includes general SOC security and privacy procedures, as well as
project-specific privacy training with clear instructions on what data and databases can be accessed
and what data are required and can be recorded.
Locators may talk to sample member’s family, relatives, or other references to obtain updated
contact information. To protect the sample member, locators are given scripts on what they can and
cannot say when using these sources to obtain information. For example, interviewers will be
instructed not to tell anyone that the sample member has been selected to participate in a study of
COBRA. Rather, they will indicate that Mathematica is trying to reach the sample member for an
important study sponsored by DOL.
6.

Locating and Calling Contact Sheets

Project team members keep only the minimum amount of printed private information needed
to perform assigned duties. Hard-copy materials (such as locating or calling contact sheets)
containing data with any individual-level identifiers (e.g., name, street address) are stored in a locked
cabinet/desk when not being used. When in use, such materials are carefully monitored by a project
supervisor and are never left unattended. At the conclusion of the project, a complete disposition of
all remaining sample will be conducted, and the contact sheets and other associated materials will be
either archived or destroyed per agreement with DOL’s Project Officer.
7.

Hard-Copy Printouts

Sensitive temporary work files, used to create hard-copy printouts and stored in temporary
work files on local hard drives, are deleted on a periodic basis. Private hard-copy output that is no
17

06859 OMB Part A

Mathematica Policy Research

longer needed is shredded or stored securely. Test printouts of data records carrying personal
identifiers that are generated during file construction are shredded.
8.

Data Files

When possible, electronic files without personal identifiers are created for everyday use. Data
and sample files that must contain sensitive data are stored in a restricted-access location on the
network. Access to data and sample files is granted only at the request of the project director (Anu
Rangarajan) or the survey director (Julita Milliner-Waddell). This folder is restricted to staff who are
currently working on the project and is available only to the staff who must have access to all the
sample information to select and process the sample or to process the data files. Sensitive data that
are no longer needed in the performance of the project will be magnetically erased or overwritten
using Hard Disk Scrubber or equivalent software, or otherwise destroyed.
To carry out data collection using the IVR, our IVR vender, ISA, will need data about sample
members. All data transmission between ISA and Mathematica will take place via a secure site, and
all files will be encrypted and password protected. ISA will sign a project-specific services
agreement with Mathematica and will comply with all privacy protocols set forth in this clearance
request. ISA agrees to maintain the privacy of all information it receives as part of this study.
11. Questions of a Sensitive Nature
Sections C and D of the survey of UI recipients being conducted for the COBRA Subsidy
Study contain some questions—about employment and health insurance—that may be considered
sensitive by sample members who have been unemployed or without health insurance for a long
time. Some questions about health status, chronic health conditions, and medical visits and expenses
(Section F), income and participation in transfer programs (Section G), and financial well-being
(Section H) may also be considered sensitive by some sample members. Obtaining information
about these potentially delicate topics is integral to addressing the research questions posed by the
study, both in order to describe the characteristics of COBRA- and subsidy-eligible individuals, to
describe the characteristics of COBRA enrollees, to describe the outcomes that these groups
experience, and to assess the impact of the subsidy on COBRA enrollment. The health of the
worker and dependents at the time of job loss and the income and financial ability to purchase
COBRA are critical influences on COBRA enrollment. Information on employment and health
insurance will also be critical for conducting a cost-benefit assessment to the government of offering
the subsidy.
The survey questions have been worded to show the highest level of objectivity and sensitivity.
Interviewers will also be trained to show sensitivity to respondents while remaining impartial. In
addition, many questions in the current survey have been included without modification from other
surveys of similar populations, such as the TAA Evaluation, the Accelerated Benefits
Demonstration, and the National Survey of American Families, and have been used extensively with
no evidence of harm. All questions in the current survey, including those deemed potentially
sensitive, have been thoroughly pretested.
Further, as described in section 10, all participants will be assured of privacy at the outset of the
interview. All survey responses will remain private and will be reported in aggregate in any reports or
publically available documents, summary format, eliminating the possibility of individual
identification.
18

06859 OMB Part A

Mathematica Policy Research

12. Hour Burden of the Collection of Information
The screening survey for the COBRA Subsidy Study can be completed with either an IVR or an
interviewer. The questions are designed to determine whether the sample member was eligible for
COBRA benefits at the time of job loss, as those who do not meet this broad criterion will be
ineligible for the study. After their identity is verified and a callback number collected, COBRAeligible sample members will answer and confirm one screening question before being transferred to
an interviewer at Mathematica. Sample members who do not meet the eligibility criterion for the
study (participation in an employer-sponsored health plan at job loss) will answer a maximum of
three screening questions. The screening interview is estimated at two minutes for both groups of
respondents. A copy of the IVR screening questionnaire is included as Appendix E. Completion of
the full survey for eligible sample members is estimated at 45 minutes. A copy of the full COBRA
Subsidy Study Survey instrument is in Appendix F.
The time burden for administering the study eligibility screening interview is estimated at
between 733 and 867 hours for the anticipated 22,000 to 26,000 screener respondents. For the full
interview, the total estimated time burden is 4,350 hours for the projected 5,800 respondents. These
estimates, presented in Table A.2, are based on timing tests conducted with Mathematica staff and
on pretests conducted with individuals who met the study criteria. Pretest respondents were referred
by Mathematica staff. Hard copy instruments were used to administer the pretest interviews which
ranged from 35 to 59 minutes, for an average of 51 minutes. However, we expect to realize a
reduction in administration time of between five and eight minutes, because of the efficiencies that
will be gained by using CATI.
Table A.2 Burden Estimates for COBRA Subsidy Study Respondents
COBRA Subsidy Study Survey
Screeners 4

Number of Respondents
Responses per Respondent
Minutes per Response
Total Respondent Burden (Hours)
Total Burden Cost

22,000 – 26,000
1
2
733 – 867
$10,555 - $12,485

Full Interviews
5,800
1
45
4,350
$62,640

The total burden cost of conducting this survey is estimated to range between $73,195 and
$75,125. This cost represents the time to complete both screening and full interviews multiplied by
the number of projected respondents at an estimated average hourly wage of $14.40 per hour. 5
13. Estimated Total Cost Burden to Respondents and Record Keepers
There will be no financial costs borne by respondents.

4 Since the best estimate for the number of screeners ranges from 22,000 to 26,000 respondents, the total burden
for screeners is presented as a range, not to exceed 867 hours or $12,485.

This hourly wage estimate is the midpoint of wages reported by participants in another DOL study: the initial
Individual Training Account Evaluation. In that study, hourly wages for the ITA study participants ranged between
$13.60 and $15.20 (McConnell et al. 2006).
5

19

06859 OMB Part A

Mathematica Policy Research

14. Estimated Annualized Cost to the Federal Government
The total cost of the study to the federal government is $3,499,443. Over the two-year study
period, this translates to an annualized cost to the federal government of $1,749,721. These costs
include the following major expense categories required to conduct this study of the Impact of the
ARRA Subsidy on COBRA Take-Up:
Table A.3 Cost to the Federal Government
Activity

Kickoff Meeting
Develop Workplan, Evaluation, Design & Analysis Plans
Sample Frame & Survey Design

Cost

13,508
89,968

360,000

Questionnaire Development

51,953

OMB Clearance

42,852

Obtain UI Administrative Data
Survey Management
Survey Operations – Interviewing 6
Survey Operations – Locating
Survey Operations – QC
Survey IS Programming
Technical Work Group Meetings

130,000
225,000
1,606,373
155,000
55,000
140,000
102,070

Data Analysis
Report Writing

180,879
186,635

Client Briefings
Public Data File

33,217
71,734

Project Management

55,254

Total Cost to the Government

$3,499,443

15. Changes in Burden
This is a new data collection effort. Therefore, there are no changes in burden.
16. Publication Plans and Project Schedule
a.

Tabulations
The COBRA Subsidy Study Survey will be used to examine:
1. Impact on COBRA enrollment of a 65 percent subsidy, including duration of
enrollment and other outcomes (e.g., health, health care coverage, employment).

6 Represents the costs of screening, administering the 45-minute survey, including costs for all mailings and contact
attempts, and incentives.

20

06859 OMB Part A

Mathematica Policy Research

2. Characteristics of job losers who are eligible for COBRA (and the ARRA
subsidy), including the reasons for people are ineligibles for the program.
3. Characteristics of job losers who enroll in COBRA, including their reasons for
enrolling and a comparison to the characteristics of job losers who do not enroll in
COBRA and their reasons for not enrolling.
Additional details on the estimation of overall impacts, impacts by subgroup, and descriptive
analysis are provided in section 2 of Part B.
b. Project Schedule and Publication Plans
The project schedule for data completion and publication is in Table A.3.
Table A.4 Timeline for Data Collection and Publication
Tasks

Schedule

Administer COBRA Subsidy Study Survey

May 2012 to November 2012
(pending OMB approval)

Mathematica briefs DOL staff on study findings

September 2013

Create public use data file

August 2013 to September 2013

17. Reasons for Not Displaying Expiration Date of OMB Approval
The expiration date issued by OMB for this data collection will be included with all materials
sent to sample members.
18. Exception to the Certification Statement
Exception to the certification statement in item 19 of Form 83-I is not requested for this data
collection.

21

06859 OMB Part A

Mathematica Policy Research

REFERENCES
Curtin, Richard, Stanley Presser, and Eleanor Singer. “Changes in Telephone Survey Nonresponse
over the Past Quarter Century.” Public Opinion Quarterly, vol. 69, no.1, spring 2005, pp. 87-98.
Singer, Eleanor, John Van Hoewyk, and Mary P. Maher. “Experiments with Incentives in Telephone
Surveys.” Public Opinion Quarterly, vol. 64, no. 2, summer 2000, pp. 171-188.
Jäckle, Annette, and Peter Lynn. “Respondent Incentives in a Multi-Mode Panel Survey: Cumulative
Effects on Nonresponse and Bias.” Working paper presented to the Institute for Social and
Economic Research, University of Essex, Colchester, United Kingdom, 2007.
McConnell, Sheena, Elizabeth Stuart, Kenneth Fortson, Paul Decker, Irma Perez-Johnson, Barbara
Harris, and Jeffrey Salzman. “Managing Customers’ Training Choices: Findings from the
Individual Training Account Experiment.” Washington, DC: Mathematica Policy Research,
Inc., December 2006.
Schwartz, Lisa K., Lisbeth Goble, and Edward M. English. “Counterbalancing Topic Interest with
Cell Quotas and Incentives: Examining Leverage-Salience Theory in the Context of the Poetry
in America Survey.” Proceedings of the American Association for Public Research. Montreal,
Canada: American Association for Public Opinion Research, 2006.
Schochet, Peter, Jillian Berk, and Pat Nemeth. “Short-Term Results of the New Survey Procedures
for the TAA Evaluation.” Memo to Sande Schifferes. Princeton, NJ: Mathematica Policy
Research, Inc., November 2008.

22