supporting_statement_b_FINAL

SUPPORTING STATEMENT

U.S. Department of Commerce

U.S. Census Bureau

Comparing Health Insurance Measurement Error (CHIME)

OMB Control Number 0607-<XXXX>

Part B – Collections of Information Employing Statistical Methods

Question 1. Universe and Respondent Selection

Medica Research Institute, an affiliate of Medica Health Plans based in Minnesota, is providing sample. Medica Research Institute received approval through the Internal Review Board (IRB) to conduct this study, and to provide Census with a file of phone numbers of enrollees. Medica will mail an advance letter to their enrollees, explaining that they are collaborating with the Census Bureau on a study. Medica’s letter will invite enrollees who do not wish to participate to opt-out by calling in to the Medica call center. Medica will then draw a sample from among enrollees who did NOT opt out for transfer to Census.

The sample will be comprised of enrollees in a range of different coverage types offered by Medica. Specifically, the study sample will be a stratified random sample of Medica subscribers under age 65 who reside in Minnesota in one of six sampling strata:  (1) Medicaid, (2) direct purchase through the insurance marketplace (MNsure), (3) direct purchase not through marketplace (4) employer sponsored insurance, (5) MinnesotaCare, and (6) subscribers who have moved between public and private health insurance markets over the 15 month period prior to the sample draw.  The sample file will be de-duplicated by address so that only one subscriber per address/household can be sampled.

The sample size was limited to 5,000 completed household interviews due to budget. To maximize the utility of the study, a Technical Advisory Group (TAG) was convened to provide expertise and input on a number of design issues, including selection of questionnaire treatments, strata and sample size per strata. Table 1 displays the assumptions and power calculations used to determine the final target sample size per strata.

Table 1: Power Analysis Assumptions and Associated Case Count Estimates

Parameters	Assumptions
Number of questionnaire treatments	2
Number of strata (insurance markets)	5
Alpha level	0.10
Power	0.80
Minimum detectable difference between treatments	4 ppt
Case counts needed by strata under assumptions*
ESI (5%)	1,190
Medicaid (17%)	3,538
Non-group within the marketplace (11%)	2,454
MinnesotaCare (17%)	3,538
Non-group outside marketplace (11%)	2,454
Total person-records required	13,174
Number of households interviewed	5,312
*Based on administrative records, 100% of cases are enrolled in this coverage. The percentage in parenthesis for each strata represents the estimated under-reporting for each coverage type based on what can be gleaned from the literature. ESI assumptions are based on Davern et al, 2008; Medicaid are based on an average from experimental findings in Call et al, 2012; no data exists for marketplace under-reporting so we assume it will be lower than Medicaid, as most marketplace enrollees pay a premium, which could help them identify the coverage as from the marketplace. We assume MinnesotaCare under-reporting to be similar to Medicaid and that non-group outside the exchange will be similar to non-group coverage within the marketplace.

More detail on the TAG and other technical details of the study design can be found in the White Paper (Attachment A).

Sample from all six strata will be comingled when Medica provides it to Census. There will be a variable on Medica’s sample file that indicates strata/coverage type, and this indicator will be embedded in Census’s control number/case ID. However, for operational purposes, interviewers will not know coverage type, and sample from all five coverage types will be treated equally in terms of contact procedures, refusal conversion and general level of effort.

Upon contact with a resident, the interviewer will attempt to identify a household member 18 years old or older to conduct the survey. A single household respondent will be asked questions about all household members. Based on response rates from a related research study conducted in 2010 (the Survey of Health Insurance and Program Participation), which were 48 percent in the random digit dial sample and 61 percent in the Medicare enrollee sample, we anticipate a response rate in the range of 50 percent. For operational purposes, however, to ensure we meet our target goal of 5,000 households, we will draw sample assuming a 30 percent response rate (resulting in a total sample of 16,667 sample units). We will also have some data on all non-respondents from the Medica records, which will enable a sophisticated non-response analysis.

Question 2. Procedures for Collecting Information

All interviews will be conducted by telephone interviewers at the Census Bureau’s telephone facility in Hagerstown, Md. In order to minimize interviewer effects, interviewers will be randomly assigned to one of two groups – one for each questionnaire treatment. Each interviewer group will work on both questionnaire versions over the course of the entire study. In order to allow each questionnaire version “equal access” to fresh sample and fresh interviewers, data collection will be divided into two field periods of equal length, and within each field period both questionnaire versions will be worked evenly. In order to control for interviewer effects, interviewer groups will first be randomly assigned to one questionnaire version, and they will work on only that version during the first field period. At the end of the first field period, each interviewer group will be rotated off that first questionnaire version and on to the other questionnaire version. They will receive a brief training on the new questionnaire version, focusing just on the differences between their previous questionnaire and the new one. They will then work on just that second questionnaire version throughout the second field period. Thus, over the course of the survey, the interviewers will work in both field periods and both questionnaire treatments. In each field period, an interviewer will be working on a single questionnaire treatment.

Question 3. Methods to Maximize Response

All interviewers working on the CHIME survey will have had several years of experience in working on Random Digit Dial (RDD) surveys, and thus will have some familiarity with and skills at cold-calling respondents and persuading them to participate in a survey. In addition to this general experience, interviewers will be trained in the content and purpose of the CHIME survey, and they will be given a set of responses to address common questions or objections from respondents.

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. Daily progress reports will be produced and distributed to a team and decisions about modifying the call parameters will be made on a weekly basis. Furthermore, at the end of the first field period a more complete analysis can be conducted in order to adjust the parameters for the second field period, if appropriate.

Question 4. Tests of Procedures or Methods

The CPS ASEC has recently undergone a complete redesign based on testing that began in 1999. The ACS has also undergone numerous tests. For example, in 2005, cognitive testing was conducted on the ACS basic health insurance module, and in 2011-2012, ACA-related adaptations to the basic series were tested with Massachusetts state residents. Furthermore, as of this writing, cognitive testing is being conducted on further adaptations to the ACS for health reform. The CHIME team is in close contact with the Census representatives on the ACS testing team in order to gain from the findings as much as possible to inform the CHIME question design.

Question 5. Contacts for Statistical Aspects and Data Collection

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

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

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

• Olga Koly, Census Telephone Center Coordinating Office (301-763-1948)

• Kathleen Call, State Health Access Data Assistance Center (612-624-3922)

• Angie Fertig, Medica Research Institute (952-992-3299)

• Don Oellerich, HHS/ASPE (202-690-7409)

Attachments:

CHIME questionnaire

Solicitation letter sent to enrollees by Medica

3