Two of the bureaucracy's central planning agencies have come up with different answers as to the number of uninsured in America. Surprising nobody, the Administration chose to hype the report that shed the best light on it despite the fact that it is the less reliable sampling.
Here is a portion of an outstanding story from Joseph Antos at AEI:
This week’s double-barreled release of government statistics on health insurance coverage leaves us with only one question: How many Americans are insured because of Obamacare? Remarkably, the two highly regarded government surveys released this week do not even agree whether the number of uninsured increased or decreased. The survey that received a great deal of attention said there were 3.8 million fewer uninsured. The other, which was hardly noticed, found that there were 1.3 million more uninsured.
The Centers for Disease Control (CDC) reported preliminary results on the expansion of health insurance coverage. Its National Health Interview Survey(NHIS) interviewed 27,000 people in the first three months of this year. The survey estimates that the number of uninsured dropped by 3.8 million since 2013. That represents a 1.3 percentage point decline in the uninsured rate, from 14.4 percent last year to 13.1 percent early this year.
Estimates from an even larger survey of the uninsured from the nation’s premier statistical agency, the Census Bureau, were released a few hours later. The Census Bureau has been collecting information on health insurance for decades based on the Current Population Survey (CPS). Data were collected from a sample of 68,000 households in February, March, and April of 2014. That survey found that 42 million—13.4 percent of the population—were uninsured in 2013. Interesting, but last year’s uninsured rate tells us nothing about how much the Affordable Care Act (ACA) expanded health insurance coverage this year.
A day after the two main reports were issued, the CDC quietly placed another table on its website. The new table compares estimates from the NHIS and the CPS for the early months of 2014. It reports the NHIS result that 13.1 percent of the population lacked health insurance when they were interviewed in the January through March time period of 2014. But it also reports the CPS estimate that 13.8 percent were uninsured during the February through April interview period.
The difference in levels estimated by the two surveys is due to differences in data collection methods and time periods, sampling error, nonresponse rates, and date processing that the CDC says leads to minor differences in estimates. That suggests that the trend in the uninsured rate measured by the CPS should be similar to that measured by the NHIS. It isn’t.
The NHIS estimates a decrease in the uninsured of 1.3 percentage points, but the larger CPS shows an increase of 0.4 percentage points from last year to early this year. One survey says there were 3.8 million fewer uninsured Americans, while the other says there were 1.3 million more.
What’s going on?
It is no accident that the administration released CDC’s estimates early on September 16 followed shortly by the Census Bureau’s routine report on last year’s insurance coverage, delaying the comparable Census estimates to the next day. Good news about coverage gains drowned out the Census report. The high reliability of the NHIS was front and center in the press coverage. The contradictory evidence from Census was buried.
Although the NHIS is a highly reliable survey, the CPS is even more reliable. Its sample is 150 percent larger than that of the NHIS, which means that its estimates have significantly lower statistical variance. If the NHIS is the gold standard, then the CPS must be the platinum standard.
In fact, the CPS questionnaire has been substantially improved this year with the addition of new questions that do a better job of identifying both the type of insurance people have and the months they are covered. The previous version of the survey asked if the respondent had health insurance at any time during the last year. The current version asks if the respondent has coverage at the time of the interview as well as questions that directly ask about any specific months during the previous year of noncoverage. By asking about current coverage, the CPS can now be used to make the same comparisons over time as the NHIS. ...
