The latest jobs report was, well, kind of uneventful, close to economists’ expectations, no huge surprises. But it did also have some confusing, or seemingly contradictory, details down there in the data.

The unemployment rate rose slightly in February, to 4.1%, because the number of unemployed people went up while the number of people in the labor force went down. Meanwhile, the number of people employed in February fell by 588,000, while U.S. employers added 151,000 jobs to their payrolls.

Confused by that last one — employment went down while the number of jobs went up? Let’s break it down.

We call it the monthly jobs report, but it’s actually two reports mashed together by the Bureau of Labor Statistics, based on two surveys the agency does each month. And they can disagree, because they go to different sources for different information.

The household survey asks people whether they’re working or looking for work, while the payroll survey asks employers how many jobs they have.

That second survey’s a lot bigger, said Dori Allard with the Office of Industry Employment Statistics at the BLS: “121,000 businesses or government agencies, 631,000 individual worksites. And there is a lot more stability in those estimates than there is in the household survey, which is relying on interviews from approximately 60,000 households,” she said.

In the household survey, a change in employment is only considered “statistically significant” if it’s above 600,000.

Joe Brusuelas, chief economist at consulting firm RSM, said February’s employment decline was big, but not that big. 

“Given the large standard error inside that report, we really want to be cautious about overreacting,” he said.

Elise Gould, a senior economist at the Economic Policy Institute, agreed. For a single month, she said, the household survey is less reliable.

“When the surveys tell a different story, we got to take the payroll survey — give that more weight because of the sample size,” she said.

But, she said, over time, “the household survey can’t be ignored. Sometimes the household survey is better at predicting changes in the business cycle. It might find softening sooner,” she said.

The survey has been sending warning signals about long-term unemployment, said Brusuelas at RSM.

“There’s a large number of people who’ve been unemployed for six months or more,” he said.

But the survey is telling a different story about America’s least-educated workers, said Jane Oates, senior policy adviser at WorkingNation: “An increase in the labor market participation of people with less than high school, and a drop in their unemployment.”

As for the current divergence between the payroll and household surveys, Dori Allard at BLS said give it a few months.

“Over the long-term, the two series do tend to track well together,” she said. “Typically the trends are in the same direction.”