Every few months the Bureau of Labor Statistics releases a job report. Every single time the numbers are used as a weapon. We’re told that by whomever is in power at the time tat things are great, after all unemployment is only at this rate. Then we’re told by whomever isn’t in power that the other people aren’t being honest with us, the numbers are really something else.

When one side claims that employment is at a certain level and another side claims that the real employment numbers are something else entirely, one gets the sense that nobody is being entirely honest.

So, let’s take a look at how the BLS comes up with these numbers and what they really mean.

The Sample

When it comes to statistics, one of the most important things to consider is the sample. Wikipedia has a portal dedicated to statistics if you want all the sordid details, but in general, a sample should be a microcosm of the entire population, and should be large enough to reduce margins of error to a reasonable level.

Information on the BLS sample is provided in their document, “How the Government Measures Unemployment.” Unlike the polls we often see which only poll a thousand people, the BLS sample is substantial: It is comprised of about 60,000 households, which translates into about 110,000 individuals. The households are chosen so that the sample is a good representation of the entire US population. This takes into account,for example, sex, ethnicity, and rural vs. urban residence.

With a sample size that large, the margin of error is going to be quite low – and that is what makes this data so valuable.

Another interesting thing is that the households in the sample are interviewed for 4 consecutive months, with 25% of the participants leaving the sample each month. Those households are then interviewed again after 8 months; after that, the household is removed from the sample forever. This means that about 75% of the sample is the same month to month, and from year to year, about 50% of the sample is the same. This is intended to more accurately reflect trends across the sample size.

Responses are then weighted and adjusted. Weighting of response is an attempt to fix issues when a certain segment of the population under- or over-represented in the sample. Numbers are also adjusted for seasonal work, so that the numbers look more consistent from month to month.

If it sounds like there is a bit of art to the science, you would be correct. Statisticians have to make decisions; sometimes, simply changing the wording of a question can change the responses, not everyone defines “Likely Voter” in the same way, and weighting a response poorly can have unexpected results.

That said, when it comes to the raw data supplied by the BLS, I’ve never heard of anyone claiming it to be inaccurate or biased.

What Does “Unemployed” Mean, Anyway?

But, when it comes to determining who is and who isn’t employed, the answer isn’t always so clear. The BLS supplies us with six – yes, six – different ways of looking at the employment statistics. These are the numbers that the politicians like to use to say that things are just peachy or a cataclysm like we’ve never seen is just on the horizon.

The U-3 value is the one most commonly used if you want to make the economy look good but retain some level of credibility. It is also the Official Unemployment Rate. No surprise there. The unemployment rate, in this case, is the number of people who are actively looking for work but cannot find a job. As of November, 2016, the seasonally adjusted unemployment rate is at 4.6%.

This sounds reasonable, right?

Well, yes and no. It does not include people who fall into the category of “discouraged workers.” These people are unemployed and have given up looking for a job for some reason. Maybe they feel there aren’t any jobs to be had; perhaps they’re receiving public assistance; perhaps living in mom’s basement is working out after all.

If you want to include those people then you want to look at the U-4 value. This sets unemployment at 5%.

But what about the people who are unemployed, have given up looking, but would be happy to work if they could? Oh, those people are in the “marginally attached” labor force. These people have looking for work at some point in the past year, but not in the last 4 weeks.

If you want to include those people then you want to look at the U-5 value. This sets unemployment at 5.8%.

Well, what about people who have a part time job, but would really like a full time job? Yup, there’s a special grouping for those people, too.

If you want to include those people, along with the marginally attached workers and the discouraged workers, then you want to look at the U-6 value. This sets unemployment at 9.3%.

Now, if you really want to make your economy look good, then you want to pay a visit to the U-1 value. This defines unemployed as not having had a job in the past 15 weeks. So, if you’re working for a temp agency, and you are lucky enough to have been assigned for a single day at some point in the last 15 weeks, you are not considered unemployed. Such a deal – the U-1 unemployment rate sits at 1.8%.

And then there is the Labor Participation Rate. This doesn’t have a U designation, but it includes all people at the age of 16 or above, except for those who do not want to work or who cannot work. Someone who doesn’t want to work may be a homemaker, for example, and someone who cannot work may be incarcerated. The unemployment rate using this metric is 37.3%.

Lies, Damn Lies, and Statistics

So we have all of these numbers and they all mean something different. Does that mean that someone using the U-3 rate in a discussion is wrong, and the person who is talking about the Labor Participation Rate is correct?

Not exactly. Each of these numbers does represent a different data point that has some value. It can be useful, for example, to know how many people haven’t had any work for 15 weeks. It can be useful to know how many people have given up looking for work, but would work if they could.

But, when one person says that the unemployment rate is 4.6%, and another says that it’s really 9.3%, and a third says that the other two are crazy, employment is only 1.8%, which one is correct?

Well, all of them. They’re just changing the definition of unemployed to suit their purpose at the time. They’re all counting on you not knowing what these numbers are and what they mean. As is usual, politicians and pundits and talking heads are all expecting you to be ignorant and to support whatever side simply out of trust.

After all, the Democrats are good guys, you can trust them, right? So obviously the Republican is lying. Or, the Democrats have their guy in office and they just want the numbers to sound good, so obviously the Democrat is lying. They want these numbers to be party politics instead of what it should be: an objective look at the information we can glean from the data.

So when you read an article or watch a video segment on employment rates in our country, pay special attention to which numbers are being used and the context in which they are being used. Don’t be a partisan voter; be an informed voter.

Personally, I think that the most important and useful number to look at is the Labor Participation Rate. It is simple. It doesn’t try to play games with how long you’ve been unemployed or the last time you’ve looked for work. Are you employed right now, and if not, do you want to be?

It is also the number that seems to reflect what we see across America: Towns full of people who would love to work, but their company has gone out of business, moved overseas, or has downsized in a desperate attempt to stay afloat.

It is the number that seems to match reality.