Part B. Supporting statement v4

SUPPORTING STATEMENT

U.S. Department of Commerce

U.S. Census Bureau

2013 Current Population Survey Annual Social and Economic Supplement Content Test

OMB Control Number 0607-<XXXX>

Part B – Collections of Information Employing Statistical Methods

Question 1. Universe and Respondent Selection

As discussed in Part A, Question 2 of the Supporting Statement, there are three objectives of this information collection:

• Operational test of increased content and modified instrument structure.

• Comparisons between the health insurance estimates from the Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC) redesign versus those from the CPS 2013 production instrument.

• Development of an alternative baseline for use in evaluating changes in health insurance reporting (of both coverage status and plan type) as a result of the implementation of the state-level health exchanges in 2014.

Each of these uses has different implications for the sample design. The sample for the production CPS will come from two sampling frames: 1) those in the regular rotation, and 2) retired sample introduced due to the Children’s Health Insurance Program (CHIP) related sample expansion. For the former sample, the earliest their CPS interviewing will have started is November 2011 and they will have been interviewed anywhere from zero to eight times (evenly distributed) at the point of their CPS ASEC 2013 production interview. For most sample members who were interviewed eight times, the 2013 interview will be their second ASEC. For the production retired sample, the earliest their CPS interviewing will have started is May 2011. Most sample members will have been interviewed eight times at the point of their CPS ASEC 2013 production interview, and for most the 2013 interview will be their first ASEC. For the redesign, retired CPS sample will be used. The earliest their CPS interviewing will have started is August 2010 and they will have been interviewed anywhere from eight to ten times at the point of their CPS ASEC 2013 content test interview. For most sample members, the 2013 interview will be their second ASEC. Below is a summary of these sample differences and sample sizes:

Table 1: Sampling frame for 2013 CPS ASEC production and redesigned instrument

	Earliest possible CPS interview	At the point of ASEC 2013 Interview		Approximate sample size
		# times interviewed	ASEC interview
CPS Production
Regular sample				78,000
Month-in-Sample 1-7	December, 2011	0-7	1st – 2nd	59,000
Month-in-Sample 8	November, 2011	8	2nd	7,000
Retired sample (CHIP expansion)	May, 2011	8 times	1st	12,000
CPS Redesign Retired Sample	August, 2010	8-10 times	2nd	15,000

In terms of comparing results from the production instrument and the content test, it will be possible to minimize sample design differences by focusing on particular subsets of the production sample. Specifically, among the regular CPS production cases, the sample of 7,000 cases who were interviewed eight times in CPS and once with ASEC will be a very close match to the CPS redesign sample in terms of exposure to and experience with CPS interviewing. The CPS production retired sample (n=12,000) will also have had a very similar experience as the CPS redesign sample, except that they will not have experienced an ASEC supplement before.

In terms of mode, the production CPS will be conducted using Computer Assisted Telephone Interviewing (CATI) as well as Computer Assisted Person Interviewing (CAPI), with about half the interviews being conducted over the telephone. The redesign will be conducted using only CATI. To address any potential mode effects, data from the CPS redesign will be compared to data from the subsets of the production CPS who were given a telephone-administered interview separately from the CPS CAPI cases. Analysis of the redesign will proceed taking into account any mode differences identified.

With respect to sample size, the Census Bureau determined that 5,000 households would be sufficient for the purposes of the feasibility test. However, the US Department of Health and Human Services Office of the Assistant Secretary for Planning and Evaluation (ASPE) and the Council of Economic Advisors proposed that the Census Bureau substantially increase the number of interviewed households from the planned 5,000 to 15,000 for the purpose of developing the alternative baseline and comparing estimates with the 2013 production ASEC.

Power analysis was performed by staff at ASPE and the Council of Economic Advisors and was confirmed by Census. The analysis showed that the increase in sample size would provide both a sufficient sample to detect differences between the test sample and the production 2013 CPS ASEC; this is critical for assessing any break in trend. See the power analysis prepared jointly by ASPE and the Council of Economic Advisors (Attachment K) and the memo on sample size and power from the Census Bureau (Attachment L).

The size of the sample will be 20,000 addresses with the goal of 15,000 completed households interviewed. The universe is the civilian noninstitutionalized population with no oversampling. We are estimating a 75 percent response rate,¹ which was determined based on the monthly response rate of the CPS. If the response rate is lower than 75 percent, the Census Bureau has additional CPS retired sample in reserve for the content test.

Question 2. Procedures for Collecting Information

All interviews will be conducted by telephone interviewers at the Census Bureau’s telephone facilities. Interviewers with at least two years’ experience working with CPS and at least three years’ experience working with standardized computer assisted telephone interviewing (CATI) questionnaires will be eligible to participate in the content test. Please see Attachment A for CPS sample design and methodology, and Attachment C for an example of the advance letter to be sent to respondents.

Question 3. Methods to Maximize Response

All interviewers working on the content test will have had two years of experience in working on the CPS and thus will have some familiarity with and skills at persuading respondents to participate in the survey. In addition to this general experience, interviewers will be trained in the content and purpose of the CPS ASEC content test, and they will be given a set of responses to address common objections from respondents. The telephone facilities also have specially-trained refusal conversion interviewers who will be assigned to the field test.

The initial parameters in the call scheduler are set at a maximum of 20 calls per case. However, this can be modified during data collection and adjusted accordingly. We have also drawn extra sample to hold in reserve in case the response rate for the content test is below 75 percent. If at about the halfway mark of the given field period it appears that we are at risk of missing our target number of completed interviews, we may release more sample, balancing the need for a target number of completes with upholding a reasonable response rate.

Question 4. Tests of Procedures or Methods

The CPS ASEC redesigned income and health insurance questions have been developed based on information collected from literature reviews, iterative cognitive testing, and, for health insurance questions, a series of split-ballot field tests.

The instrument will be evaluated upon whether it functions properly in the production environment. In particular, attention will be paid to any change in the speed of the instrument and storage issues.

Changes in the design of the questions will be compared to the production CPS ASEC health insurance instrument as well as other national surveys that have calendar year estimates of health insurance coverage (e.g., Survey of Income and Program Participation and Medical Expenditure Panel Survey). The focus of the comparisons will be on uninsurance, type specific coverage, policyholders, and dependents on private plans. As the sample size of the test was selected to examine differences in the key estimate of the uninsured rate for the new questions only at the national level, the examination these differences by subpopulations lack sufficient statistical power to determine differences in the impacts of the different questions by subpopulations.

The new questions on health insurance exchanges are the only new questions on the redesigned CPS health insurance section that have not undergone a field test or production. However, these questions have been tested in other ways, such as through focus groups and cognitive interviews of the CPS health insurance module. Interviews were conducted in Massachusetts, which passed legislation in 2006 similar to the ACA provisions concerning Health Insurance Exchanges. As health reform unfolds, the Census Bureau will work with its federal partners to refine the exchange questions.

Question 5. Contacts for Statistical Aspects and Data Collection

Several individuals will be involved in the statistical analysis, including:

1. Amy Steinweg, Brett O’Hara, Carla Medalia, Edward Welniak, Carmen DeNavas-Walt, and Jessica Semega, Social, Economic and Housing Statistics Division (301-763-3196)

2. Joanne Pascale, Center for Survey Measurement (301-763-4920)

3. Steve Ash, Demographic Statistical Methods Division (301-763-4294)

4. John Magruder, Telephone Center Coordinating Office (301-763-7789)

5. David Hornick, Demographic Statistical Methods Division (301-763-4183)

6. Lisa Clement and Tim Marshall, Demographic Surveys Division (301-763-5482)

Note: This is a limited list that is subject to change.

Attachments

1. CPS sample design and methodology

2. Confidentiality brochure

3. CPS ASEC Advance letterⁱ

4. Items booklet – health insurance

5. Survey instrument flow document

6. Survey instrument path diagrams

7. Items booklet – income and other questions

8. Report from Westat on CPS ASEC cognitive testing

9. References on CPS ASEC redesign

10. Memo from HHS on sample size

11. ASPE sample size calculations

12. Census calculations on power

13. Scripted scenarios for the health insurance module

1 Originally, DSMD estimated that amount a 5,000 household units would be able to detect a 1.0 percentage point change in the uninsured rate. On 4/05/2012, Brett O’Hara received a memo from Ben Sommers of ASPE stating that HHS would pay for increased sample to detect a 0.5 percentage point change in the uninsured rate between the 2013 CPS ASEC content test and the 2013 CPS ASEC production.

i The letter included in the package is the modified CPS ASEC letter. The letter will have the correct OMB number; currently, it does not. The letter may include a sentence that indicates that the email address to which the respondent may write is not secure.

5