It’s official. Yesterday afternoon, a team of researchers from the Centers for Disease Control and Prevention (CDC), including the agency’s director, was forced to officially correct their deeply flawed study that wrongly blamed obesity for 400,000 U.S. deaths every year. Their statement read in part: “We are writing to inform the JAMA readership that through an error in our computations, we overestimated the number of deaths caused by poor diet and physical inactivity.

Although the CDC researchers owned up to one of their substantial mistakes, it is unfortunate that they did not reassess the larger problems with their highly controversial method. As we’ve told you before, the CDC relied on a flawed study that has come under fire from both inside and outside the agency for drastically overestimating the number of obesity-attributable deaths.

The CDC’s correction only came after the Center for Consumer Freedom pointed out that the agency had overestimated the number of deaths supposedly attributable to obesity, and The Wall Street Journal’s Betsy McKay broke the story. Despite yesterday’s mea culpa, the CDC team’s new estimate of 365,000 deaths a year fails to correct their study’s most glaring problems, which serve to greatly exaggerate the real figure.

“365,000 deaths” is quite a convenient number if you’re an obesity-hyper trying to scare the public. It probably won’t be long before you hear the farce that obesity causes 1,000 deaths a day. To reach this too-good-to-be-true conclusion, the CDC had to perform a little mathematical hocus-pocus. The original 400,000 estimate was actually rounded down from 402,553, even though the nearest 5,000 number was 405,000. This time around, after recalculating to correct the computational error, the CDC reports the actual figure as 362,304 — which is 360,000 if you round to the nearest 5,000. So now they’re rounded up rather than down, both times ignoring the rules of rounding by failing to go to the nearest 5,000. And both times they ended up with the more media-friendly figure.