Although I think the redesigned survey is a big improvement, any time that significant changes are made in survey questions, there’s a risk that it will be difficult to make good comparisons of the data for the years before and after the changes were made.  However, the commentators who have expressed alarm that this will interfere with analysis of the ACA’s effects appear to be unaware of some key facts about when the changes take effect and the full range of Census Bureau products.  I appreciate their concerns, but they can rest assured that we will have plenty of good Census Bureau data to use as we analyze and debate the effects of the ACA.

Much of the news coverage last week regarding the CPS changes created an erroneous impression that is summed up in this headline: “Major Changes to U.S. Census Will Make It Nearly Impossible to Track How Obamacare Is Doing.”  Not all of the stories went quite that far in criticizing the CPS revisions, but most of the news stories missed the two key reasons why the revised survey questions are not a major problem:

• The Census Bureau isn’t changing its much larger survey of households, the American Community Survey (ACS), which yields far more reliable data about health insurance coverage, especially at the state level.  (The ACS is based on a sample of households that is 30 times larger than the one used for the CPS!)
• The data being released this fall, based on the new CPS questions, isn’t for calendar year 2014, it’s for 2013, which means that we will have comparable 2013 and 2014 CPS survey results – before and after the major ACA changes took effect.

In contrast to the ACS, the health questions that have long been used in the CPS ask people whether they have been uninsured for all of the last 12 months.  But research has shown that the answers people give are often about their current insurance status, not their status over the prior year.  The redesigned CPS addresses that problem by asking about coverage at the time of the survey and by looking back to January of the prior calendar year – capturing monthly insurance coverage information up through the month of the survey.  The new survey will provide information about marketplace participation, employer coverage offers and worker take-up, and the ability to track monthly transitions over a 15-month period – all welcome improvements on the prior survey.  (A summary of the advantages and drawbacks of the changes can be found here.)

I’m not suggesting that there isn’t a downside to changing survey questions.  The Wisconsin Department of Health Services has occasionally made changes to its annual Family Health Survey, and when those types of revisions disrupt the comparability of the data across years, it’s a significant source of heartburn for researchers and policy analysts like me.  However, the Census Bureau’s latest changes are made far less problematic by the fact that the much larger health survey remains the same.

Although I was disappointed that much of the press coverage last week missed a couple of key points that cast a much different light on this story, I was happy to hear commentators arguing for the importance of using good, comparable Census Bureau data.  Let’s hope that interest continues.  Health care policymaking will be improved if lawmakers from across the political spectrum can set aside their preconceived notions about the ACA and use objective analysis of data to guide their policy choices.