The puzzle in this month’s jobs report is the divergence between job growth — we added 244,000 jobs — and the unemployment rate, which rose by two-tenths of a percent. In theory, that could be explained by discouraged workers returning to the labor force. But the numbers say “nay”: Both the labor-force participation rate and the employment-to-population ration stood unchanged. Rather, as David Leonhardt explains, the change in the unemployment rate mostly seems to be statistical noise:

The rise in the unemployment rate is mostly a reflection of the fact that the rate fell by an artificially large amount over the previous several months. It doesn’t actually mean unemployment rose last month. Instead, it reflects a kind of statistical catch-up. The old picture of the job market, as presented by the household survey, had been too optimistic. (Did anyone really believe that the job market recently improved at its most rapid two-month pace since the 1950s, as the unemployment rate suggested?) Today’s report helps correct the picture. This is simply the nature of surveys: they have noise in them.

In general, my rule with jobs reports is to focus on the job numbers rather than the unemployment rate. When you add jobs, it’s pretty clear what’s going on. But though the unemployment rate at any given time is a pretty good measure of the labor market, the month-to-month change is frequently misleading. A jump in the unemployment rate could mean anything from “the economy is getting better and discouraged workers are returning to the labor market” to “the economy is tanking and workers are getting fired left and right.” If you don’t know why the unemployment rate moved, you don’t know what it means, whereas with jobs, you really only need to know if the number was positive or negative.