Census Changes Health Insurance Coverage Survey To Improve Accuracy

Census Bureau To Revise Survey Questions Measuring Health Coverage. On April 15, the Census Bureau announced that it will change the questions regarding health insurance coverage in the census survey in order to produce more accurate results. The bureau's director, John H. Thompson, explained that the change “will provide a more accurate baseline for assessments of changes in insurance coverage”:

The recent changes to the Current Population Survey's questions related to health insurance coverage is the culmination of 14 years of research and two national tests in 2010 and 2013 clearly showing the revised questions provide more precise measures of health insurance through improved respondent recall. 

This change was announced in September 2013 and implemented because the evidence showed that reengineering the questions provides demonstrably more accurate results. The Census Bureau only implements changes in survey methodology based on research, testing, and evidence presented for peer review.

The revised questions were implemented in the Current Population Survey beginning in February 2014 in time for data collection covering calendar year 2013, and thus will provide a more accurate baseline for assessments of changes in insurance coverage, including that of the Affordable Care Act. [Census Bureau, 4/15/14 via PR Newswire]

Right-Wing Media Claim Census Changing Survey To Hide Effects Of ACA

Wall Street Journal: “Cooking The Obamacare Stats.” In an April 15 editorial, The Wall Street Journal suggested that the change in survey questions was made for political purposes in order to provide more favorable numbers for the Affordable Care Act:

The Census says the new CPS is more accurate, and demographers and statisticians of all persuasions have argued for years that the old version overreports the number of uninsured relative to other surveys. But the inflated figures served the politicians who were plugging national health care, and Democrats in 2009 and 2010 used them to exaggerate the problem amid the push to put more of American health care under government direction.

As with the IRS targeting of conservative political groups, this sudden change will undermine public trust in the supposedly nonpartisan institutions of government. Muddying a useful source of information about ObamaCare's results is definitely unfortunate, but our guess is that it wasn't coincidental. [The Wall Street Journal, 4/15/14]

Fox's Varney: “We Know That Obamacare Is In Trouble But Now We Can't See How Badly. How Convenient.” On the April 16 edition of Fox News' America's Newsroom, Fox Business host Stuart Varney called the change “very convenient” and claimed that the Obama administration “is struggling with Obamacare and now cannot produce the relevant numbers on the impact of Obamacare. Very convenient.” [Fox News, America's Newsroom, 4/16/14]

Fox Nation: Survey Changes “To Hide Effects Of Obamacare.” Fox Nation posted an excerpt of a Mediaite piece on the Census change under the headline “Obama Census Bureau Changes Health Care Survey Questions To Hide Effects Of ObamaCare”:

[Fox Nation, 4/15/14]

Hot Air: Census Change Guarantees Number Of Uninsured Will Drop. A post on Hot Air claimed the change was being made to artificially lower the number of uninsured after the law's first year:

As it turns out, the change the Bureau is making all but guarantees that the number of uninsured next year, post-ObamaCare, will be lower than than the number last year, pre-ObamaCare. This drug will work, even if the feds have to tweak how success is measured to ensure it. [Hot Air, 4/15/14]

New Survey Methodology Has Been “In The Works For Years”

Census Director: New Questions Are “The Culmination Of 14 Years Of Research.” In the press release announcing the change, Thompson explained that the “recent changes to the Current Population Survey's questions related to health insurance coverage is the culmination of 14 years of research and two national tests in 2010 and 2013 clearly showing the revised questions provide more precise measures of health insurance through improved respondent recall”:

Based on more than a decade of comprehensive research to improve the measure of the Nation's uninsured population and give the country better data, the U.S. Census Bureau redesigned the health insurance coverage questions for the Current Population Survey. The new method captures current health insurance coverage, in addition to coverage throughout the entire year.  Results from two separate national test of the new questions in 2010 and 2013 showed that the new method captured health insurance coverage better than the old method. Specifically, it improved the respondent recall of when they were covered by health insurance and then works backwards through time about specific months of coverage. Based on the test results and 14 years of research, the new questions were included in this the February, March and April CPS (reflecting the collection of data for 2013). The CPS will act to provide an accurate baseline for 2013 to facilitate the assessment of the impact of the Affordable Care Act implemented in 2014. [Census Bureau, 4/15/14 via PR Newswire]

Census Expert: Change Is “Scientifically Driven.” The Huffington Post quoted Terri Ann Lowenthal, a Census expert, who pointed out that changes such as the one the Census just made “are scientifically driven, but sometimes poorly timed.” The article went on to note that “this particular change has been in the works for years”:

Census watchers are adamant that the change is not being made for political reasons, but to improve the data. “These changes are scientifically driven, but sometimes poorly timed,” said Terri Ann Lowenthal, a former overseer of the Census for Congress and co-director of the Census Project.

Indeed, this particular change has been in the works for years, and was the subject of detailed, peer-reviewed studies. The most recent study from 2013 can be found here. Another study was conducted in 2010, and other studies and discussions go back as far as the early years of the George W. Bush administration.

In order for data to stay relevant, Lowenthal explained, surveys need to be modified and updated. “There has to be changes in questions, and frankly there may never be a good time,” she said. [The Huffington Post, 4/15/14]

White House Spokeswoman: Change Was Announced In September For Comment, In The Works For 10 Years. In an interview on Fox News' America's News HQ, White Housecommunications director Jennifer Palmieri pointed out that the Census change was announced in September 2013 and subjected to public comment. Palmieri also noted that the decision is the result of an ongoing process that started under President George W. Bush's administration. [Fox News, America's News HQ, 4/16/14]

Change Will Still Measure ACA's First-Year Impact

Sarah Kliff: Survey Will Measure Before And After Major ACA Changes. In a post on Vox, Sarah Kliff pointed out that the survey will apply to 2013 data, meaning the impact of Obamacare's exchanges will be reflected in the new data:

The Census Bureau reports the health insurance rate with a one-year delay; in September 2013, for example, the agency reported the percent of Americans without coverage in 2012. It will most likely report the uninsured rate for 2013 sometime this coming fall. 

In other words: The survey will make it difficult to compare the uninsured rate for 2012, the last year for the old questions, and 2013, the first year for the new questions. But making the change now means that 2013 and 2014 -- the year before and after Obamacare's big programs started -- are using the same question set. 

Census officials told the Times that the changes will make the survey a more accurate measure of who actually has health insurance coverage. The new survey questions are expected to show a higher uninsured rate. In a test last year, they found that 10.6 percent of Americans said they did not have health insurance when using the new questions -- compared to 12.5 percent when people were surveyed with the old ones. [Vox, 4/15/14]

Kaiser's Larry Levitt: New Census Questions Will Measure 2013-2014 Change. In a series of tweets, Kaiser Family Foundation senior vice president Larry Levitt noted that the new data “will be able to measure the change from 2013 to 2014”:

[Twitter, 4/15/14]

TPM: New Survey Will Provide Apples-To-Apples Comparison Of “The Most Critical Period” Of Obamacare. Talking Points Memo noted that the revisions will keep measurement consistent between 2013 and 2014, “the most critical period during which Obamacare's main coverage expansion took effect”:

The fear was that the metric would change between 2013 and 2014, making it very difficult to discern the health care law's impact on insurance coverage. That would've been problematic. But the Census revisions, set for this fall, will measure data from 2013 onward. That means researchers will have a direct comparison from 2013 to 2014, the most critical period during which Obamacare's main coverage expansion took effect.

“We will still have an apples to apples comparison of 2013 vs. 2014 in the current [census],” Larry Levitt, vice president of the nonpartisan Kaiser Family Foundation, said in an email. He said the new Census questions are a more accurate indicator of insurance coverage. [Talking Points Memo, 4/15/14]