Unemployment rate to 8.8%

[pinqy]

Do you know enough probability theory to figure out the odds on getting a hand in a game you have never played before? Where do you think the sample errors come from? It comes from probability theory.

Yyes, I am well aware of the sample and non-sampling errors associated with the CPS. You do not appear to be.

The size of the sample vs. accuracy is all based on either census data or extrapolations there from vs. sample size plus data on response rates or it is totally useless.

And the population controls used are updated every January. This causes a readjustment and some changes. You haven't answered the question about why you think it's outdated. And the sample selection is NOT based on the Census.

For example landline households with or without caller ID, being on or not being on the do not call list have different response rates and if not adjusted for those rates the results are not worth compiling because of sample self selection error. That is covered in every Stat 101 course that is worth taking.

I have corrected you on this multiple times...the CPS is NOT a random telephone survey, so none of what you're saying here is relevant.

Then there is respondent error, does the respondent have accurate and up to date information on all household members? How well are the survey questions tested for reliability of response? In each and every post of yours on this subject you have ignored error generators. You are either an undergrad or else someone who never earned a paycheck based on statistical analysis.

Please point out where the subject has come up and I ignored it? I've cited the margins of error for the UE rate in this very thread.

Respondent error is considered, and it's well known that, particularly because second parties can answer, that the income results from the March supplement are way off (more people report earning less than the minimum wage than the minimum wage). BLS has a team of psychologists who work with the sample survey design. All surveys are well tested before fielding.

I'm post grad and have earned a paycheck for many years based on statistical analysis.

When will you realize that the CPS is NOT a random telephone survey and quit jabbering about people not answering the phone?

 

[bigrebnc1775]

Answer the fucking questions or quit wasting my time.

I answered you those who have recieved or are recieving or who have be denied unemployment are on a record that is on file. They are kept track of. Which brings us back to the original post of 8.9 percent being incorrect. Your argument has failed.

Try again: A person was denied benefits in November last year. There is a record of that, sure, but what's their employment status NOW??????

Where are the records for people WHO NEVER APPLIED FOR BENEFITS?

How will you get any DEMOGRAPHICS?

Yes, there are records of how many people are currently recieving UI benefits, and the offices will have records of denying claims and ending claims, but there's no tracking before or after that.

And did you not notice that the number of unemployed under the survey is much higher than the number of people receiving benefits?

You keep ignoring all this

You keep ignoring the fact sports, You said there is no way of knowing about the 200,000 who no longer are recieving unemployment checks. The government knows what happen to them Why don't they report what happen to those people? Could it be they were either disqualified or ran out of benefits?

 

[pinqy]

I answered you those who have recieved or are recieving or who have be denied unemployment are on a record that is on file. They are kept track of. Which brings us back to the original post of 8.9 percent being incorrect. Your argument has failed.

Try again: A person was denied benefits in November last year. There is a record of that, sure, but what's their employment status NOW??????

Where are the records for people WHO NEVER APPLIED FOR BENEFITS?

How will you get any DEMOGRAPHICS?

Yes, there are records of how many people are currently recieving UI benefits, and the offices will have records of denying claims and ending claims, but there's no tracking before or after that.

And did you not notice that the number of unemployed under the survey is much higher than the number of people receiving benefits?

You keep ignoring all this

You keep ignoring the fact sports, You said there is no way of knowing about the 200,000 who no longer are recieving unemployment checks. The government knows what happen to them Why don't they report what happen to those people? Could it be they were either disqualified or ran out of benefits?

Two seperate things here. First, there is no way of knowing the flow of UI receipiants because it's not published. The reason it's not published is because it would take way too long to compile and publish. UI benefits numbers come out every week. DOL has 3 days to do the count of initial claims (which it gets from the states which gets it from the local offices) and perform seasonal adjustment before release. The state data is often incomplete or not reported in time and statistical estimates are made. The numbers are revised twice as the state reports are more completed.

Continuing claims have a week lag: 8 days for the count and seasonal adjustment, and that still gets revised due to incomplete/missing data.

Federal claims have a 2 week lag: 13 days to get the count (no seasonal adjustment).

Those are just the levels and don't require looking at individual records. What you want is for every single record to be looked at and verified for reason of termination of benefits, and even if you did that, there would be a lot of vagueness. If someone is terminated because of faulty address or some other technical issue, we don't know if they found a job or not. And it would just take too long to count all those records and double check.

An example: The official Employment level comes from a survey of non-farm businesses who contribute to UI. The sample is about 410,000 worksites and they're just asked number of employees, avg hours, avg wages. Now it's possible to get the full count from UI records, and that is done, but it takes months to do.

So your scheme to take the full admin count, while theoretically possible, would take so long as to be useless.

And again, you'd still be missing alll the people who never applied, and you'd have no tracking of anyone no longer recieving benefits past the time they stopped recieving.

Person A is on UI. Gets a job. Terminates UI. 3 months later does he still have a job? No way to know unless he's applied again, which he might not.

Do you really think nobody's thought of this before? Are you that egotistical? It's been thought of and rejected as impractical and not giving enough details.

 

[bigrebnc1775]

Try again: A person was denied benefits in November last year. There is a record of that, sure, but what's their employment status NOW??????

Where are the records for people WHO NEVER APPLIED FOR BENEFITS?

How will you get any DEMOGRAPHICS?

Yes, there are records of how many people are currently recieving UI benefits, and the offices will have records of denying claims and ending claims, but there's no tracking before or after that.

And did you not notice that the number of unemployed under the survey is much higher than the number of people receiving benefits?

You keep ignoring all this

You keep ignoring the fact sports, You said there is no way of knowing about the 200,000 who no longer are recieving unemployment checks. The government knows what happen to them Why don't they report what happen to those people? Could it be they were either disqualified or ran out of benefits?

Two seperate things here. First, there is no way of knowing the flow of UI receipiants because it's not published. The reason it's not published is because it would take way too long to compile and publish. UI benefits numbers come out every week. DOL has 3 days to do the count of initial claims (which it gets from the states which gets it from the local offices) and perform seasonal adjustment before release. The state data is often incomplete or not reported in time and statistical estimates are made. The numbers are revised twice as the state reports are more completed.

Continuing claims have a week lag: 8 days for the count and seasonal adjustment, and that still gets revised due to incomplete/missing data.

Federal claims have a 2 week lag: 13 days to get the count (no seasonal adjustment).

Those are just the levels and don't require looking at individual records. What you want is for every single record to be looked at and verified for reason of termination of benefits, and even if you did that, there would be a lot of vagueness. If someone is terminated because of faulty address or some other technical issue, we don't know if they found a job or not. And it would just take too long to count all those records and double check.

An example: The official Employment level comes from a survey of non-farm businesses who contribute to UI. The sample is about 410,000 worksites and they're just asked number of employees, avg hours, avg wages. Now it's possible to get the full count from UI records, and that is done, but it takes months to do.

So your scheme to take the full admin count, while theoretically possible, would take so long as to be useless.

And again, you'd still be missing alll the people who never applied, and you'd have no tracking of anyone no longer recieving benefits past the time they stopped recieving.

Person A is on UI. Gets a job. Terminates UI. 3 months later does he still have a job? No way to know unless he's applied again, which he might not.

Do you really think nobody's thought of this before? Are you that egotistical? It's been thought of and rejected as impractical and not giving enough details.

NO it's not two seperate things. We are talking about the 200,000 drop from the 600,000 who were on the unemployment wagon. Those people are on record. But if you insist on saying

And again, you'd still be missing alll the people who never applied, and you'd have no tracking of anyone no longer recieving benefits past the time they stopped recieving.

The unemployment numbers are higher than the false numbers of 8.9% Good day sir. You lost

 

[pinqy]

NO it's not two seperate things. We are talking about the 200,000 drop from the 600,000 who were on the unemployment wagon. Those people are on record.

I'm sorry, you're now claiming only 600,000 were receiving unemployment? And then you're going to go on and say that's more than the 13.5 million official unemployed? And while there are records, it would take too long to compile and publish, and you'd still have no idea about all the people never receiving or applying for benefits or what has subsequently happened to those no longer receiving benefits.

But if you insist on saying

And again, you'd still be missing alll the people who never applied, and you'd have no tracking of anyone no longer recieving benefits past the time they stopped recieving.

The unemployment numbers are higher than the false numbers of 8.9% Good day sir. You lost

I do keep saying it and you keep refusing to address it. The Current Population Survey accounts for all the people who never applied or were never eligible for UI benefits. So why the hell do you think a count of UI benefits alone will give a higher number? Again, 8,770,443 receiving benefits, 14,060,000 total Unemployed (not seasonally adjusted). At 90% confidence the number is between 13,729,845 and 14,390,155 (+-2.3%)

Give me your numbers and where you're getting them from. You keep saying it's higher, but you haven't given any evidence. Hell, until a few days ago you thought the numbers came from benefits.

 

[bigrebnc1775]

NO it's not two seperate things. We are talking about the 200,000 drop from the 600,000 who were on the unemployment wagon. Those people are on record.

I'm sorry, you're now claiming only 600,000 were receiving unemployment? And then you're going to go on and say that's more than the 13.5 million official unemployed? And while there are records, it would take too long to compile and publish, and you'd still have no idea about all the people never receiving or applying for benefits or what has subsequently happened to those no longer receiving benefits.

But if you insist on saying

And again, you'd still be missing alll the people who never applied, and you'd have no tracking of anyone no longer recieving benefits past the time they stopped recieving.

The unemployment numbers are higher than the false numbers of 8.9% Good day sir. You lost

I do keep saying it and you keep refusing to address it. The Current Population Survey accounts for all the people who never applied or were never eligible for UI benefits. So why the hell do you think a count of UI benefits alone will give a higher number? Again, 8,770,443 receiving benefits, 14,060,000 total Unemployed (not seasonally adjusted). At 90% confidence the number is between 13,729,845 and 14,390,155 (+-2.3%)

Give me your numbers and where you're getting them from. You keep saying it's higher, but you haven't given any evidence. Hell, until a few days ago you thought the numbers came from benefits.

I'm sorry, you're now claiming only 600,000 were receiving unemployment?

Where have you been in this discussion? The 600,000 was from last months new numbers count. the drop came from people losing their benefits due to running out of time, or lying. You're dancing so much you are getting drunk.

It's new numbers added onto the already exisiting old numbers.

Hell, until a few days ago you thought the numbers came from benefits

You keep saying they don't keep track of the numbers. They have to keep track

 

[pinqy]

Ok, let's try this: Lay out EXACTLY what you would define as Unemployed, how you would define the Unemployment rate (unemployed as a percent of what), what details about the unemployed you would want (age, race, sex, veteran's status, disability, duration, reason, industry, job), how often you would put out these figures, how you would collect and continue to track people. Hopefully, you'll see by yourself the conclusion that every economist/statistician in the last 50 years has come to...that a survey is the only method to put out such data on a monthly basis (excluding extremely socialized countries where employment/unemployed is under major control of the government).

If you don't see it, I'll explain it, but you must promise to actually read thoroughly and try to understand instead of hand-waving the difficulties I've pointed out.

And for the umpteenth time, yes, levels of people receiving UI checks is kept and aggregated. Refusals and denials are NOT published and remain at the state level...AND it would take a long time regularly sort out all that data, AND it only tells us what happened to that person that week and not any other week, AND there is NO tracking, except by survey, of people who don't apply for benefits.

 

[editec]

Pinqy you have my sincerest sympthies.

Every time the issue of unemployment statsitics comes up somebody or the other informs people how BLS compiles them.

And each and every time they do, people rush in to tell them they must be wrong.

I've been watching this exact same debate on various boards for over a decade.

They're not called KNOW-NOTHINGS for nothing, ya know.

Ignorance, ignorance that is completely impervious to facts or logic, ignorance that revels in glofiying their ignorance is their super-power.

What's the old expression?

You can't teach pigs to sing. It wastes your time and it annoys the pigs

 

[bigrebnc1775]

Pinqy you have my sincerest sympthies.

Every time the issue of unemployment statsitics comes up somebody or the other informs people how BLS compiles them.

And each and every time they do, people rush in to tell them they must be wrong.

I've been watching this exact same debate on various boards for over a decade.

They're not called KNOW-NOTHINGS for nothing, ya know.

Ignorance, ignorance that is completely impervious to facts or logic, ignorance that revels in glofiying their ignorance is their super-power.

What's the old expression?

You can't teach pigs to sing. It wastes your time and it annoys the pigs

If you're agreeing with him then both of you haven't a clue. If both of you are agreeing on the samething. "A POLL" Then I wouldn't make any bets on those numbers.

 

[bigrebnc1775]

Ok, let's try this: Lay out EXACTLY what you would define as Unemployed, how you would define the Unemployment rate (unemployed as a percent of what), what details about the unemployed you would want (age, race, sex, veteran's status, disability, duration, reason, industry, job), how often you would put out these figures, how you would collect and continue to track people. Hopefully, you'll see by yourself the conclusion that every economist/statistician in the last 50 years has come to...that a survey is the only method to put out such data on a monthly basis (excluding extremely socialized countries where employment/unemployed is under major control of the government).

If you don't see it, I'll explain it, but you must promise to actually read thoroughly and try to understand instead of hand-waving the difficulties I've pointed out.

And for the umpteenth time, yes, levels of people receiving UI checks is kept and aggregated. Refusals and denials are NOT published and remain at the state level...AND it would take a long time regularly sort out all that data, AND it only tells us what happened to that person that week and not any other week, AND there is NO tracking, except by survey, of people who don't apply for benefits.

Now for you shrimp fry

None of this

(age, race, sex, veteran's status, disability, duration, reason, industry, job),

Has to do with the topic of those who have lost unemployment benefits. You are basing your whole argument of unemployment being 8.9 on a servey, that does not account for honesty. I liad out in another post what are the things to look for as to how the true unemployment numbers look. Keep believing in the tooth fairy it may just come and bring you a dollar for your first lost tooth.

 

[pinqy]

None of this

(age, race, sex, veteran's status, disability, duration, reason, industry, job),

Has to do with the topic of those who have lost unemployment benefits.

Never said it did. But it has everything to do with measuring the labor force and potential labor force, and you CANNOT get any of that by looking at UI data unless you go through every single application and you still wouldn't get all the info.

You are basing your whole argument of unemployment being 8.9 on a servey, that does not account for honesty.

Yes, it does. We know it's an ESTIMATE, and nobody claims it's exact, precise, or 100%. But it's better than anything else.

I liad out in another post what are the things to look for as to how the true unemployment numbers look. .

Why do your "true unemployment numbers" completely ignore people who don't apply or receive government assistance? How does looking at food stamp receipiants tell you how many people are unemployed?

 

[pinqy]

Pinqy you have my sincerest sympthies.

Every time the issue of unemployment statsitics comes up somebody or the other informs people how BLS compiles them.

And each and every time they do, people rush in to tell them they must be wrong.

I've been watching this exact same debate on various boards for over a decade.

They're not called KNOW-NOTHINGS for nothing, ya know.

Ignorance, ignorance that is completely impervious to facts or logic, ignorance that revels in glofiying their ignorance is their super-power.

What's the old expression?

You can't teach pigs to sing. It wastes your time and it annoys the pigs

I try, but he clearly has no interest in a real discussion. I like how he first criticized the numbers for coming from the UI claims, and then when he had to admit they weren't, he did a 180 and claimed that they should come from UI claims. Mind boggling.

 

[bigrebnc1775]

None of this

(age, race, sex, veteran's status, disability, duration, reason, industry, job),

Has to do with the topic of those who have lost unemployment benefits.

Never said it did. But it has everything to do with measuring the labor force and potential labor force, and you CANNOT get any of that by looking at UI data unless you go through every single application and you still wouldn't get all the info.

You are basing your whole argument of unemployment being 8.9 on a servey, that does not account for honesty.

Yes, it does. We know it's an ESTIMATE, and nobody claims it's exact, precise, or 100%. But it's better than anything else.

I liad out in another post what are the things to look for as to how the true unemployment numbers look. .

Why do your "true unemployment numbers" completely ignore people who don't apply or receive government assistance? How does looking at food stamp receipiants tell you how many people are unemployed?

]

Never said it did.

You keep posting it as if it's a factor

But it has everything to do with measuring the labor force and potential labor force, and you CANNOT get any of that by looking at UI data unless you go through every single application and you still wouldn't get all the info.

No it does not have anything to do with the 200,000 drop from last month.

Why do your "true unemployment numbers" completely ignore people who don't apply or receive government assistance? How does looking at food stamp receipiants tell you how many people are unemployed

That would mean the unemployment numbers would be higher than the government says.

 

[william the wie]

Ok, let's try this: Lay out EXACTLY what you would define as Unemployed, how you would define the Unemployment rate (unemployed as a percent of what), what details about the unemployed you would want (age, race, sex, veteran's status, disability, duration, reason, industry, job), how often you would put out these figures, how you would collect and continue to track people. Hopefully, you'll see by yourself the conclusion that every economist/statistician in the last 50 years has come to...that a survey is the only method to put out such data on a monthly basis (excluding extremely socialized countries where employment/unemployed is under major control of the government).

If you don't see it, I'll explain it, but you must promise to actually read thoroughly and try to understand instead of hand-waving the difficulties I've pointed out.

And for the umpteenth time, yes, levels of people receiving UI checks is kept and aggregated. Refusals and denials are NOT published and remain at the state level...AND it would take a long time regularly sort out all that data, AND it only tells us what happened to that person that week and not any other week, AND there is NO tracking, except by survey, of people who don't apply for benefits.

Now for you shrimp fry

None of this

(age, race, sex, veteran's status, disability, duration, reason, industry, job),

Has to do with the topic of those who have lost unemployment benefits. You are basing your whole argument of unemployment being 8.9 on a servey, that does not account for honesty. I liad out in another post what are the things to look for as to how the true unemployment numbers look. Keep believing in the tooth fairy it may just come and bring you a dollar for your first lost tooth.

UE and UI are unrelated. The problems with UE are:

Compounding sample error. (The size of the population is known maybe +/- 0.5% error. Potential labor force is fairly well known with actual number of illegal aliens being the biggest source of error say +/- 1-2%. Actual labor force the main source of error is the estimate of the size of the black and gray markets being subtracted from the labor force before it gets to BLS. Since police budgets are dependent on crime stats and the FBI compiles the stats that are used by the census bureau to feed data to BLS this is a big source of error. ["Freakonomics" by Levitt & Dubner has an anthropological study of Chicago drug gangs showing the difference between criminal income/employment reported to the census bureau by the FBI vs. actual, the gap is gigantic] this normally unreported error takes people out of the labor force in job lots. What pinqy is talking about is what percentage of error can be expected from the fraction of census data the BLS works with.)

Sample self selection. (Only people who choose to reply are counted. Without longitudinal studies of employment among those who reply vs. those who do not the error rate cannot be guessed at.)

A simple test from number theory says that the UE numbers are probably in error. I don't remember the name of the proof but the problem was figuring out why numbers games were always based on the last 3-4 digits instead of the first 3-4 digits of a daily number the gangsters did not control. It turns out that 1-4 is the first significant digit 60% of the time and more than 50% of the time for the second digit instead of the expected 44% and 40%. You can verify this proof by using a ruler and a logarithmic scale: 1-4 is more than 60% of the distance of 1 to higher 1 on a log scale and then repeat the process for second digits. UE numbers do not conform to this rule sufficiently in either first or second digit in raw numbers or percentages to reflect accuracy and never have.

 

[bigrebnc1775]

Ok, let's try this: Lay out EXACTLY what you would define as Unemployed, how you would define the Unemployment rate (unemployed as a percent of what), what details about the unemployed you would want (age, race, sex, veteran's status, disability, duration, reason, industry, job), how often you would put out these figures, how you would collect and continue to track people. Hopefully, you'll see by yourself the conclusion that every economist/statistician in the last 50 years has come to...that a survey is the only method to put out such data on a monthly basis (excluding extremely socialized countries where employment/unemployed is under major control of the government).

If you don't see it, I'll explain it, but you must promise to actually read thoroughly and try to understand instead of hand-waving the difficulties I've pointed out.

And for the umpteenth time, yes, levels of people receiving UI checks is kept and aggregated. Refusals and denials are NOT published and remain at the state level...AND it would take a long time regularly sort out all that data, AND it only tells us what happened to that person that week and not any other week, AND there is NO tracking, except by survey, of people who don't apply for benefits.

Now for you shrimp fry

None of this

(age, race, sex, veteran's status, disability, duration, reason, industry, job),

Has to do with the topic of those who have lost unemployment benefits. You are basing your whole argument of unemployment being 8.9 on a servey, that does not account for honesty. I liad out in another post what are the things to look for as to how the true unemployment numbers look. Keep believing in the tooth fairy it may just come and bring you a dollar for your first lost tooth.

UE and UI are unrelated. The problems with UE are:

Compounding sample error. (The size of the population is known maybe +/- 0.5% error. Potential labor force is fairly well known with actual number of illegal aliens being the biggest source of error say +/- 1-2%. Actual labor force the main source of error is the estimate of the size of the black and gray markets being subtracted from the labor force before it gets to BLS. Since police budgets are dependent on crime stats and the FBI compiles the stats that are used by the census bureau to feed data to BLS this is a big source of error. ["Freakonomics" by Levitt & Dubner has an anthropological study of Chicago drug gangs showing the difference between criminal income/employment reported to the census bureau by the FBI vs. actual, the gap is gigantic] this normally unreported error takes people out of the labor force in job lots. What pinqy is talking about is what percentage of error can be expected from the fraction of census data the BLS works with.)

Sample self selection. (Only people who choose to reply are counted. Without longitudinal studies of employment among those who reply vs. those who do not the error rate cannot be guessed at.)

A simple test from number theory says that the UE numbers are probably in error. I don't remember the name of the proof but the problem was figuring out why numbers games were always based on the last 3-4 digits instead of the first 3-4 digits of a daily number the gangsters did not control. It turns out that 1-4 is the first significant digit 60% of the time and more than 50% of the time for the second digit instead of the expected 44% and 40%. You can verify this proof by using a ruler and a logarithmic scale: 1-4 is more than 60% of the distance of 1 to higher 1 on a log scale and then repeat the process for second digits. UE numbers do not conform to this rule sufficiently in either first or second digit in raw numbers or percentages to reflect accuracy and never have.

It's all done by a poll and depending on what your view is so goes the poll.

 

[pinqy]

Why do your "true unemployment numbers" completely ignore people who don't apply or receive government assistance? How does looking at food stamp receipiants tell you how many people are unemployed

That would mean the unemployment numbers would be higher than the government says.

How do you figure? The current data doesn't ask any questions about government assistance. The survey is of the Adult Civilian Population. People are asked:
Did you work last week (or were absent due to leave or temporary illness)? If yes, they are employed.
Did you actively look for work in the last 4 weeks (or were on temporary layoff)? If yes, they are unemployed.

Then further questions about specific job search activity, demographics, reason for not looking, part time or full time, etc.

There are NO questions about UI benefits or government assistance and they play no role in determining the figures, so I don't know why you keep saying including people never eligible would make the government figures higher, when such people are already included.

 

[pinqy]

[It's all done by a poll and depending on what your view is so goes the poll.

How does one have a "view" about whether or not one has worked or looked for work? You don't think this is an opinion poll where people are asked what they think the unemployment numbers are, do you? People are asked what they did in the previous week. There's no "view" to that.

 

[pinqy]

Compounding sample error. (The size of the population is known maybe +/- 0.5% error. Potential labor force is fairly well known with actual number of illegal aliens being the biggest source of error say +/- 1-2%. Actual labor force the main source of error is the estimate of the size of the black and gray markets being subtracted from the labor force before it gets to BLS.

Untrue. The Labor Force, being the sum of Employed and Unemployed, is derived from the CPS itself. The monthly population numbers used are the Adult Civilian Non-Institutional Population estimates from Census. No estimates of black and gray markets are subtracted from the labor flrce before it gets to BLS because the Labor Force is a derived number, not a set thing. The CPS does not question nor does it care about legal versus illegal.

Sample self selection. (Only people who choose to reply are counted. Without longitudinal studies of employment among those who reply vs. those who do not the error rate cannot be guessed at.)

The response rate is consistanly over 90% (of eligible households), so that's not a big factor.

But what I'm waiting for you is to recant your false claim that not answering the phone, or having a cell phone turned off, or not having a phone is a factor. If a household does not answer the phone, there's a personal visit to follow up.

 

[william the wie]

A simple test from number theory says that the UE numbers are probably in error. I don't remember the name of the proof but the problem was figuring out why numbers games were always based on the last 3-4 digits instead of the first 3-4 digits of a daily number the gangsters did not control. It turns out that 1-4 is the first significant digit 60% of the time and more than 50% of the time for the second digit instead of the expected 44% and 40%. You can verify this proof by using a ruler and a logarithmic scale: 1-4 is more than 60% of the distance of 1 to higher 1 on a log scale and then repeat the process for second digits. UE numbers do not conform to this rule sufficiently in either first or second digit in raw numbers or percentages to reflect accuracy and never have.

It's all done by a poll and depending on what your view is so goes the poll.[/QUOTE] Given the huge second digit variance from expected 1946-80 on annual and monthly figures (My own research for my own purposes feel free to replicate.) of around 50% of UE numbers being X.5% or X.9% I ignore the US UE numbers as cooked books. I also got a copy of the old Facts on File "European Historical Statistics: 1750-1975" and doublechecked against those numbers and found much less evidence of manipulation (the Belgian and UK data seemed hinky but given sample size that could easily be random fluctuation.) I haven't seen such obvious cooking in more recent US data but I just figure that others did published work on the cooked books and the cooking became more subtle.

UE data like everything like all other data is published to advance an agenda since I don't know the explicit agenda I don't use it.

 

[pinqy]

Given the huge second digit variance from expected 1946-80 on annual and monthly figures (My own research for my own purposes feel free to replicate.) of around 50% of UE numbers being X.5% or X.9%

Can you link to your source data for that?