UPDATE: Just because HHS botched their number doesn't mean you should botch it as well.

You can't read the full story without logging into a subscription account (maybe I should consider that here after 2/15/15?), but here's the opening paragraph from Inside Health Policy (my emphasis):

HHS Acknowledges Error In Enrollment Numbers, Says 6.7M Effectuated

HHS on Thursday (Nov. 20) acknowledged that it overstated enrollment by nearly half a million people because it had included both dental and medical health plans in recent numbers, and affirmed that there were 6.7 million people enrolled in health plans as of Oct. 15, or 400,000 less than previously asserted.

Two things:

  • The number is somewhere between 380,000 and 400,000 (it's my understanding that 380,000 have been confirmed, the maximum possible number could be 399,166.
  • More importantly, however: In what universe does 380,000 or even 400,000 get rounded up to "nearly half a million" when that'd be a 20% difference?

In other words, let's say you're rounding to a much larger number: No one's going to have a problem with 99.8 million being rounded up to 100 million in that instance; the 200K difference is only 0.2% of the total.

In this case, however, you're talking about a 20% difference (or a 24% difference if the 380K figure proves accurate).

OK, they do say "nearly" but still.

Considering that the entire point of the incident is that their numbers were misstated by 4.7% (7.3 / 6.97 = 4.7% too high), it seems like kind of a bad thing to be off by 20-24% in your own story about the incident.

That is all.

UPDATE: Turns out the actual number of dental plans ended up being 393,000.

So, that's 7.37 million minus 393K = 6,977,000.

Yes, that's right. The actual number of QHPs as of August 15th was...6.977 million. And yes, that you can round up to 7 million even without anyone saying boo.

UPDATE 12/03: I just heard from the woman who wrote the IHP story; she apologized for the error and is correcting it to "around 400K" which is a reasonable rounding.