Unemployment numbers, like all good statistics, can be deceiving.

Sometime next week, the Bureau of Labor Statistics will release the unemployment rate for February. No doubt, all the major news outlets will do as they did last month and prominently feature that number in headlines and articles.

Before that happens, it’d help to understand what that number actually means.

Here’s the crash course: The Bureau of Labor Statistics has six different ways of measuring un- and under-employment.

When journalists and policymakers cite the “official” unemployment number, they’re looking at the third measurement, which is sometimes called U3. That number includes those jobless people who are available to work (i.e. not students) and have actively looked for a job in the last four weeks.

That official U3 measure is the one used to calculate everything from unemployment insurance expenditures to the “stress test” that banks will face in the next couple of weeks. And it’s the measure used by everyone in Washington including, say, Federal Reserve Chairman Ben Bernanke, who estimated earlier this week that unemployment—which rose to 7.6 percent in January—is likely to rise a bit more this year, topping out between 8.5 or 8.8 percent.

So it’s the most important measure.

But a lot of people—including Paul Krugman and a handful of less-famous academics (PDF)—say the U3 does a bad job representing the state of the labor force in America, since it doesn’t include those workers who haven’t looked for a job in the last month but say they want one (“discouraged” or “marginally-attached”(PDF) workers), or those who pick up part-time hours, but can’t find a full-time job due to the economy.

These folks argue that the broadest measure of unemployment (U6) is better, since that measure gives us a better idea of how many people would be working—or would be working full-time—if the economy were better.

A front page New York Times story mentioned the problem earlier, saying that while unemployment for January was 7.6 percent, “that number understated the weakness in hiring because hundreds of thousands of people dropped out of the labor force over the last year. On top of that, 3.1 million additional people were working part time because they could not find full-time jobs.”

A Financial Week story also noted a Merrill Lynch study conducted by economist David Rosenberg, indicating that the “actual unemployment rate” (as opposed to the “official” rate) this month has hit 1994 levels:

As Mr. Rosenberg explained, what the official unemployment rate misses is the vast degree of ‘underemployment’ as companies cut back on the hours that people who are still employed are working. Those hours have declined 1.2% in the past twelve months.

The BLS still counts people as employed if they are working part-time, but the number who have been forced into that status because of slack economic conditions has ballooned nearly 70% in the past year, according to the study…

That’s pretty significant. While official unemployment is at 7.6 percent, that Merrill Lynch study would put it at 13.9 percent.

The Wall Street Journal blog Real Time Economics noted the disparity in unemployment statistics too, warning that the gap between the numbers is probably going to get worse:

Because it’s a relatively young data series, the U-6 doesn’t get much attention beyond researchers. But it may deserve more focus over the coming year as the labor market continues its purge. Many employers are still focused on cutting jobs quickly to get through this downturn, pushing job-seekers aside for an extended period. After long searches — we’re in the fifteenth month of this recession — some job hunters are likely to give up and wait out the recession. Without the broader unemployment rate, many of them wouldn’t be counted.

We didn’t always measure unemployment like we do now. According to a Reuters , the government rejiggered the official unemployment rate forty years ago, excluding from the labor force those jobless people who had not actively looked for work in the past year. That, of course, had the effect of making unemployment appear lower almost overnight.

In 1994, the BLS redefined the official unemployment number again, excluding people who had not actively looked for work in the previous month. That change in definition is particularly important because we’ve seen lots of comparisons between unemployment today and unemployment in the early ’90s. But, since the definition of “unemployment” has changed in the meantime, it’s an apples and oranges situation.

Haley Sweetland Edwards is a writer in New York.