Barack Obama has rescued America from Economic Disaster

[DGuller]

Ugg, yes, the dreaded random sample. At least Clinton didn't get his way with sampling and we still have an exact census for population totals.

We do that once every 10 years and it costs a fair chunk of change. And honestly, what does it matter if the unemployment rate is 9.7 as opposed to 9.3?

If I recall correctly, the margin of error for unemployment is actually 0.1%. 

[Strix]

People who don't understand statistics fear it.

Btw, how does the survey take place? If they go door to door or call people at homes, then I assume the unemployed rates would be higher than in reality, since people who are employed are more likely not to sit at home.

No idea as to the mechanics of it, but I'm pretty sure that government statisticians are not anywhere near that incompetent.

The Department of Labor has the mechanics on their website. They basically survey 60,000 people each month. They survey those same 60,000 for X number of consecutive months (I think 4-6). They actually never ask if they are employed or not but infer it based on the answers to other questions. I guess after the 4-6 months they switch to a new 60,000 sample.

[Strix]

Ugg, yes, the dreaded random sample. At least Clinton didn't get his way with sampling and we still have an exact census for population totals.

We do that once every 10 years and it costs a fair chunk of change. And honestly, what does it matter if the unemployment rate is 9.7 as opposed to 9.3?

On the bright side, unemployment will go down once the census starts next year. They pay pretty good.

[Strix]

What's so dreaded about a sample?

It isn't exact by its very nature and can be easily manipulated. It's a useful tool but shouldn't be relied on for important things.

[alfred russel]

Ugg, yes, the dreaded random sample. At least Clinton didn't get his way with sampling and we still have an exact census for population totals.

We do that once every 10 years and it costs a fair chunk of change. And honestly, what does it matter if the unemployment rate is 9.7 as opposed to 9.3?

It matters due to the trend. At the end of Q1 the OECD was projecting unemployment to continue increase through Q4 2010 when it would reach 10.5%. While the earlier unemployment figures were worse than their projections, the fact we had a decline (or even were just stable) could mean the worst unemployment is over--a year and a half early.

[crazy canuck]

Dont forget that a drop in the unemployment rate can also mean that a significant number of people have simply removed themselves from the job market out of despair that they cant find a job.

We need a lot more data before anyone can suggest that a drop of .1% means things might be improving.

[alfred russel]

What's so dreaded about a sample?

It isn't exact by its very nature and can be easily manipulated. It's a useful tool but shouldn't be relied on for important things.

So every month, every single person in the country should have to fill out a form describing their employment status?

The only practical way to get information is through a survey, which does involve some uncertainty. If you aren't comfortable with that uncertainty, then you should just ignore all economic reporting and commentary.

[alfred russel]

Dont forget that a drop in the unemployment rate can also mean that a significant number of people have simply removed themselves from the job market out of despair that they cant find a job.

We need a lot more data before anyone can suggest that a drop of .1% means things might be improving.

I bet you that US GDP growth for Q3 is positive. You up for it?

[Strix]

So every month, every single person in the country should have to fill out a form describing their employment status?

The only practical way to get information is through a survey, which does involve some uncertainty. If you aren't comfortable with that uncertainty, then you should just ignore all economic reporting and commentary.

I am comfortable with the uncertainty. That isn't the issue. I am uncomfortable with the weight that is given to a sample that has such uncertainty.

I doubt it would cost that much more to survey those people who have come to the end of the unemployment benefits to find out if they found a job, stopped looking, or ran out of eligibility.

Using sampling to determine if the economy is getting better or worse is speculation at best. And basing how the government approaches the recession on speculation doesn't seem to be the best course of action to take.

The last set of unemployment rates might be a total false reflection of the true unemployment rate because the "sample" of people might have changed (as they do every so often). Any increase or decrease might just be based on the new set of variables and have nothing to do with the economy.

[crazy canuck]

Dont forget that a drop in the unemployment rate can also mean that a significant number of people have simply removed themselves from the job market out of despair that they cant find a job.

We need a lot more data before anyone can suggest that a drop of .1% means things might be improving.

I bet you that US GDP growth for Q3 is positive. You up for it?

Thats just the point isnt it.  We dont know yet.  We really dont know what the drop in the unemployment rate means yet.

[Hansmeister]

Dont forget that a drop in the unemployment rate can also mean that a significant number of people have simply removed themselves from the job market out of despair that they cant find a job.

We need a lot more data before anyone can suggest that a drop of .1% means things might be improving.

We do know from the data that this is exactly what happened.  Employment declined by 270,000.  The labor force declined by 800,000.  Ta-daa, unemployment went down, but we're still worse off than the month before.  Actually, the first sign of the labor market improving will probably be a spike in unemployment as more optimistic people rush to join the labor market, but i don't expect that to happen until next summer at the earliest.  The big question will be when Obama's insane spending spree sends the economy back into a tailspin?

[Faeelin]

Dont forget that a drop in the unemployment rate can also mean that a significant number of people have simply removed themselves from the job market out of despair that they cant find a job.

We need a lot more data before anyone can suggest that a drop of .1% means things might be improving.

We do know from the data that this is exactly what happened.  Employment declined by 270,000.  The labor force declined by 800,000.  Ta-daa, unemployment went down, but we're still worse off than the month before.  Actually, the first sign of the labor market improving will probably be a spike in unemployment as more optimistic people rush to join the labor market, but i don't expect that to happen until next summer at the earliest.  The big question will be when Obama's insane spending spree sends the economy back into a tailspin?

Never, because the God Emperor's prescience will avert that before it happens.

[DGuller]

What's so dreaded about a sample?

It isn't exact by its very nature and can be easily manipulated. It's a useful tool but shouldn't be relied on for important things.

Its inexactness is very calculable, and there is no reason why it can't be riled on for important things.  What's your alternative, a monthly census?

[Faeelin]

I guess my question as to why we should care what Hans thinks about the topic is because he was among those saying, "Don't worry about the deficits, the rapid economic growth of 2009-2012 will leave us with a surplus."

Shame the old board's gone, to digt up some priceless quotes.

[DGuller]

I am comfortable with the uncertainty. That isn't the issue. I am uncomfortable with the weight that is given to a sample that has such uncertainty.

Then you're not really comfortable with uncertainty.

I doubt it would cost that much more to survey those people who have come to the end of the unemployment benefits to find out if they found a job, stopped looking, or ran out of eligibility.

Maybe it wouldn't cost much more, but then it would be very stupid from a statistics point of view.  You would be calculating unemployment figures from a biased sample, which is a big no-no in statistics.

Using sampling to determine if the economy is getting better or worse is speculation at best.

No, comments like that are speculation at best.

The last set of unemployment rates might be a total false reflection of the true unemployment rate because the "sample" of people might have changed (as they do every so often). Any increase or decrease might just be based on the new set of variables and have nothing to do with the economy.

Again, the uncertainty is 0.1%, so it's not like we can go from boom to depression figures over a change of a sample.