Justification for Nonsubstantive Change Request

The Census Bureau requests approval of two non-substantive changes to the SIPP 2008 Panel questionnaire. The changes are: 1) Addition of questions to obtain data on the 2008 tax rebate; and, 2) Replacement of the 2008 health insurance questions with the questions used in the SIPP 2001 Panel.

1) Addition of 2008 Tax Rebate Questions

Justification: In 2001, the Consumer Expenditure Survey (CE) added several questions to the CE Interview survey to capture data on the 2001 Tax Rebate. The data collected allowed researchers to assess the effect of the 2001 rebate on consumer spending. In 2008, the CE is again adding to the CE Interview Survey a set of questions about the 2008 Tax Rebate recently approved by the Federal Government.

The SIPP is seeking approval to add the identical CE questions to the SIPP 2008 Panel waves 1 and 2. The CE questions include additional questions about the use of the rebate and about whether the rebate was received by check or direct deposit. The use of the rebate question is similar to a question asked in the Survey of Consumers conducted by the University of Michigan in 2001 regarding the 2001 Tax Rebate. At the request of researchers, the question on how the rebate was received is added to aid in their analysis.

Together, the questions will capture: the receipt of one or more tax rebates in the household, the month the rebate was received, the amount of the rebate(s), how the rebate(s) was received, and the primary use of the rebate(s). The rebate questions would be included in the SIPP 2008 Panel, Waves 1 and 2, from September 2008 through April 2009. This time period ensures we capture data for as many recipients of the 2008 tax rebate as possible, particularly respondents who wait until October 2008 to file a Federal tax return. The rebate questions are expected to take approximately one minute per adult to answer. This will increase the burden estimate by 1,575 hours.

To ensure these questions capture the needs of data users, the following researchers were given the opportunity to participate in the development of and review of these questions:

1. John Greenlees, Bureau of Labor Statistics, DPINR

2. David Johnson, U.S. Census Bureau

3. Robert McClelland, Bureau of Labor Statistics, DPINR

4. Jonathan Parker, Kellogg School of Management, Northwestern University

5. Ralph Rector, U.S. Department of Treasury

6. Nicholas Souleles, The Wharton School, University of Pennsylvania

OMB approved the 2008 Panel questionnaire in the Notice of Action dated March 9, 2007.

Burden Hours: The additional burden for the tax rebate questions is 1,575 hours for each of Waves 1 and 2.

2) Replacement of the 2008 Health Insurance Questions with the SIPP 2001 Health Insurance Questions

Justification: The SIPP 2004 Panel implemented a new questionnaire which included a revised health insurance section. Analysis of the 2004 data on health insurance has shown differences which can only be explained by the change in the questions. Throughout the panel, issues involving the quality of the private health insurance coverage arose. In Wave 1, the coverage rates for children came into question because the new design made it more difficult for adults to report coverage for their children. In subsequent waves, certain questions were not asked in order to reduce burden. As a result, coverage changes were not adequately captured. These dynamics being precisely the purpose of SIPP, the Census Bureau attempted to make instrument changes in Waves 5 and 6 in order to correct the situation. Those patches appear to have been improvements, but they were not complete corrections. Indeed, review of the wave 6 data still raise questions about whether the instrument and processing system correctly assigned private coverage status and characteristics to individuals. Given the complications inherent in the 2004 health insurance questionnaire and the implications for the processing system, improving the instrument for the 2008 panel has risks as well and in no way guarantees the success of health insurance coverage status in the 2008 panel.

Hence, for the 2008 Panel, we propose using the identical health insurance questions that were used for the SIPP 2001 Panel with two exceptions: 1) the Medicare claim number question (MCNUMB) will only collect the 2-digit suffix instead of the 11-digit Medicare number. The first nine digits of the Medicare number is the respondents Social Security number. Census Bureau policy prohibits obtaining social security numbers from respondents; and, 2) the word ‘private’ has been added to three screens (H4MNTH, H1KDCOV, H2KDCOV) to clarify the type of health insurance coverage. There are no known major instrument issues with the code used for the 2001 Panel questionnaire. Furthermore, the 2001 panel questionnaire yielded data which are comparable to benchmark estimates from other major surveys. The Census Bureau’s best assessment shows that the risk of serious problems is lower and the probability of high quality data is higher using the 2001 health insurance questions in the 2008 Panel.

For the 2008 Panel, we propose using the identical questions that were used for the SIPP 2001 Panel. The 2001 Panel questions yielded data which are comparable to benchmark estimates from other major surveys.

Burden Hours: The burden hours for the health insurance section will not change.