A noteworthy moment for an administration that famously refuses to apologize for its claims. Yesterday afternoon, Sarah Huckabee Sanders claimed that Donald Trump’s policies had tripled job creation for African-Americans in two years over what Barack Obama had done in eight years. The claim came in response to questions over whether Trump had said the “N-word” while working on The Apprentice:

By last night, however, Sanders was singing a different tune:

Correction from today’s briefing: Jobs numbers for Pres Trump and Pres Obama were correct, but the time frame for Pres Obama wasn’t. I’m sorry for the mistake, but no apologies for the 700,000 jobs for African Americans created under President Trump https://t.co/EXGvbliwlS

— Kayleigh McEnany 45 Archived (@PressSec45) August 15, 2018

Note that the Council of Economic Advisers is taking the hit for this publicly, attempting to take the blame off of Sanders. They released more pertinent figures after the briefing, making clear their context, even if Sanders had a bit of trouble framing them properly:

Per @PressSec briefing today: pic.twitter.com/uQzsAkOH2i

— CEA45 Archived (@WhiteHouseCEA45) August 14, 2018

Notice anything curious about those time frames? Even the updated numbers are spun a bit, although not quite as much as NBC News thinks:

The window Sanders used for Obama reflected the period from November 2008 through July 2010 — not his entire tenure, as she’d claimed — while her numbers for Trump accounted for a parallel stretch of time.

Aside from the apples-to-oranges comparison of black employment under each president, the press secretary’s mistake highlighted the CEA’s decision to measure the economic records of presidents starting on the day they were elected rather than the day they took office. That means Obama’s jobs numbers are weighed down by the recession at the end of George W. Bush’s presidency and Trump’s are buoyed by the final months of Obama’s presidency.

Comparing similar periods in two presidents’ terms is far from illegitimate, although one should start when the president actually takes office rather than when he wins the election. It also doesn’t take into account the sharp growth that should normally follow a deep recession, but that would tend to benefit Obama — theoretically. When one corrects the time frame, these are the actual numbers in black employment for the comparison of the first 18 months of each presidential term:

• Obama 1 (February 2009-July 2010): -474,000
• Obama 2 (February 2013-July 2014): 731,000
• Trump (February 2017-July 2018): 708,000

The CEA’s date shift isn’t very accurate, but it doesn’t produce much different data, at least in terms of comparisons. The US remained in recession the first several months of Obama’s first term, although job losses continued for several months after the technical recovery. In 2010, however, black employment increased overall by 377,000, although only by 234,000 at the end of July, the period for apples-to-apples comparison. Job growth in this demographic has been steady ever since:

Unfortunately for Sanders, she muffed this badly both ways. She used the wrong data (supplied by the CEA), and then represented it as the eight-year total under Obama. It was a nonsense claim all around, but especially in the context that Sanders initially provided. Even in its proper apples-to-apples time framing, though, it’s still not much of an argument.

That’s probably why Sanders felt compelled to offer a qualified apology on behalf of a president who famously detests such public statements. Will that get her in trouble with the boss? Today’s focus in the presidential Twitter account is far more on GOP wins in yesterday’s primaries rather than advancing claims on black unemployment. Don’t expect the White House to back away from the topic, though, perhaps in the future arguing that they’ve kept the pace up despite claims from economists that the job market has become saturated and further growth seemed unlikely. That’s an argument that the White House can make … hopefully with better data the next time around.