Memo from HHS on sample size

	DEPARTMENT OF HEALTH & HUMAN SERVICES	Office of the Secretary
		The Assistant Secretary for Planning and Evaluation Washington, D.C. 20201

TO: Brett O’Hara, Census Bureau

CC: Jennifer Day, Joanne Pascale, Amy Steinweg

FROM: Don Oellerich and Ben Sommers

Office of the Assistant Secretary for Planning and Evaluation

Department of Health & Human Services

SUBJECT: Operational Field Test for 2013 Current Population Survey Health Insurance Questions

DATE: April 5, 2012

Dear Brett,

As you know, we have been in conversation with you and other staff at the Census Bureau to offer our input into the formulation of new health insurance questions for the Current Population Survey Annual Social and Economic Supplement (CPS-ASEC). We greatly appreciate the opportunity to participate in these discussions, and thank you for seeking our input.

One issue that we would like to address further is the timing and nature of the field test for the 2013 CPS-ASEC. Our understanding is that Census plans to test the new health insurance questions in March 2013, and if all goes as planned, fully implement the new questions in March 2014. While the 2014 survey will still contain questions about insurance coverage during the 2013 calendar year, the entire survey instrument for these questions will have changed. This raises important issues in terms of a potential break-in-trend for the historical health insurance data from the CPS-ASEC, which would be particularly concerning given the onset of major insurance expansions under the Affordable Care Act in January 2014.

In light of this issue, we would like to propose that Census take the following two steps to minimize the impact of this change:

1. Increase the sample size of the 2013 field test so that the data can be used to test for changes in national estimates of overall insurance rates due to the new questions. This is particularly important since this will be the last CPS-ASEC completed before the Affordable Care Act expansions begin. Based on our own power calculations and our conversations with you and your staff, we think the planned test sample of 5,000 households (roughly 13,600 individuals) will be powered to detect a 1 percentage point change in overall uninsured rate. But we feel it is important to be able to detect an effect potentially half as small as that, which would require approximately another 10,000 interviewed households (24,000 individuals) in the test sample.

2. Make the microdata for the field test data available for public analysis once it is complete. This will be critical in enabling independent researchers to adequately address the potential change in trend when moving to the new CPS-ASEC data beginning in 2014.

We are happy to work with you and provide for any additional funding needs that these issues may raise. Overall, a key priority for our Department and the Administration as a whole is laying the groundwork for conducting informed and valid empirical analysis of the impacts of the Affordable Care Act. We appreciate all that the Census Bureau is doing towards this end, and we hope that attention to these concerns will further strengthen this important effort.