Cooking the unemployment numbers

Readability

Cooking the unemployment numbers

Last week, the New York Post’s John Crudele broke the story that the Cen­sus Bureau, which con­ducts the Cur­rent Pop­u­la­tion Sur­vey (CPS) that is the basis for the unem­ploy­ment rate reported by the Bureau of Labor Sta­tis­tics (BLS), has been fal­si­fy­ing the data since 2010. Cur­dele inter­viewed a per­son who was caught by the Cen­sus Bureau in 2010 sim­ply mak­ing up data, with the employee claim­ing his supe­ri­ors told him to do so because his region was not suc­cess­fully inter­view­ing enough peo­ple for the sur­vey. Accord­ing to an anony­mous source, that effort inten­si­fied in the months lead­ing up to the 2012 elec­tion, with Sep­tem­ber 2012’s data specif­i­cally fal­si­fied to Pres­i­dent Barack Obama’s favor, and con­tin­ues to the present time.

These alle­ga­tions are cur­rently being inves­ti­gated by both the House Over­sight Com­mit­tee and the Inspec­tor Gen­eral of the Cen­sus Bureau, with the BLS also quite inter­ested in them.

One place they can start is com­par­ing what came out of the CPS to a mea­sure of unem­ploy­ment con­ducted by Gallup, started in Jan­u­ary 2010. There are a cou­ple of key dif­fer­ences between the CPS and Gallup which make a com­par­i­son a bit harder:

- While the CPS uses a ref­er­ence week that includes the 12th of the month (5th of the month in Novem­ber), Gallup uses a 30-​day rolling aver­age.
– The CPS sur­veys (or claims to sur­vey) 60,000 peo­ple age 16 and over, while over the course of each 30-​day rolling aver­age, Gallup sur­veys 30,000 adults.

For­tu­nately, the BLS releases, as part of its dataset, the data from the por­tion of the CPS that cov­ers adults, or about 57,500 sur­veyed out of approx­i­mately 153,000,000 con­sid­ered part of the labor force. That allows an apples-​to-​apples comparison:

Gallup-CPS

For the most part, the CPS mea­sure of adult unem­ploy­ment is sig­nif­i­cantly lower than Gallup’s mea­sure. How sig­nif­i­cant? Let’s take a look at Gallup’s mea­sure on the day that puts the CPS ref­er­ence day right in the mid­dle of the rolling aver­age, the 27th day of the month (20th for Novem­ber and also Decem­ber to avoid an arti­fi­cial post-​Christmas spike). The CPS unem­ploy­ment was cal­cu­lated by divid­ing the num­ber of unem­ployed by the num­ber con­sid­ered to be in the work­force, so I could get much closer than to the near­est tenth of a per­cent­age point reported. The raw data was not avail­able for Gallup’s mea­sure of unem­ploy­ment, so I took the clos­est pos­si­ble num­ber to the CPS mea­sure that still rounded to Gallup’s tenth of a per­cent­age point reported.

Gallup-CPS divergence

Polls, which is what the CPS and Gallup mea­sures really are, come with a mar­gin of error, within which the true value can be expected either 90% or 95% of the time. For the CPS, the 90%-confidence mar­gin of error is +/-0.20 per­cent­age points and the 95%-confidence mar­gin of error is +/-0.24 per­cent­age points. For Gallup, the 90%-confidence mar­gin of error is +/-0.28 per­cent­age points and the 95%-confidence mar­gin of error is +/-0.34 per­cent­age points.

Two polls are con­sid­ered to be in good agree­ment when their val­ues are within each oth­ers’ mar­gin of error. Mean­while at least one poll has to be wildly incor­rect when the dif­fer­ence between the two is more than the sums of their mar­gin of error. Out of 46 months’ worth of data:

- 18 months saw Gallup’s and CPS’s mea­sures of unem­ploy­ment dis­agree by more than the com­bined 90%-confidence mar­gin of error of 0.49 per­cent­age points, with 17 months hav­ing Gallup’s mea­sure higher.
8 months saw the mea­sures of unem­ploy­ment dis­agree by between 0.28 per­cent­age points (Gallup’s 90%-confidence mar­gin of error) and 0.49 per­cent­age points, with another 3 months see­ing a dis­agree­ment between 0.20 per­cent­age points (CPS’s 90% con­fi­dence mar­gin of error) and 0.28 per­cent­age points.
17 months saw the mea­sures of unem­ploy­ment in “good agree­ment”, dis­agree­ing by less than 0.20 per­cent­age points.

When two polls wildly dis­agree more than they are in “good agree­ment”, one of them has to be wrong. Given the dis­agree­ment has been almost invari­ably in the administration’s favor, and there already was a proven round of fak­ery in the CPS, it sure looks like the offi­cial mea­sure of unem­ploy­ment has been cooked longer than a burnt turkey.

****************************************

[olimome­ter id=2]

It’s Sat­ur­day the last day of the month and the week.

Most impor­tant of all it is the end of the first month of my Mag­nif­i­cent Seven Blog­gers. Later today the final post of the month for my mag­nif­i­cent seven will go up and I’ll send out my final pay­ment of the month for that final arti­cle pay­ing them in full.

You have been treated to top­ics like Church his­tory from Pas­tor George Kelly Com­mon Core’s Kobayashi Maru from AP (Lady Lib­erty) Dil­lon , a bit of Venezuela from Fausta, Indoc­tri­na­tion vs edu­ca­tion from Baldilocks The san­ity of home­school­ing from Linda Szu­gyi , What Smok­ing Taxes actu­ally do to a cul­ture from Marathon Pun­dit and the Left’s Law­fare in Wis­con­sin from Steve Eggle­ston and today you will get one more piece from Steve round­ing out the con­tri­bu­tions from my Mag­nif­i­cent Seven.

Unlike the liti­gious blog­gers of the Huff­in­g­ton Post my seven have not just been promised expo­sure, each has been paid for their work since day one.

That’s why AOL will never come call­ing with $315 Mil­lion or even 2 Mil­lion because if they did they’d not only would they have to pay my writ­ers, but I’d likely just start a new even big­ger ven­ture along these same lines with the dough.

It would be very cool if even­tu­ally the story ended with a huge site, link­ing hun­dreds of con­ser­v­a­tive blog­gers and with our own staff of paid reporters, but for now we’re still the begin­ning and on this last day of the first month of my Mag­nif­i­cent Seven I’m $179 shy of end­ing that first month with all my writ­ers paid in full AND the mort­gage paid in full too.

One tip jar hit of $179 or two at $90 or Four at $45 will get it done.

I can’t promise that this will even­tu­ally grow into that ulti­mate con­ser­v­a­tive pow­er­house dri­ving sup­port­ing and elect­ing con­ser­v­a­tives as my ulti­mate plan, but suc­ceed­ing in the small things will make the large things possible.

Two final thoughts

We are offer­ing spon­sor­ships of both the Mag­nif­i­cent Seven & the Mag­nif­i­cent Panel now is an excel­lent time to jump on board, con­tact me here for more details

If I can get 63 more sub­scribers at $20 a month then the mort­gage and the mag­nif­i­cent seven will be paid in full EVERY MONTH.




Last week, the New York Post‘s John Crudele broke the story that the Census Bureau, which conducts the Current Population Survey (CPS) that is the basis for the unemployment rate reported by the Bureau of Labor Statistics (BLS), has been falsifying the data since 2010. Curdele interviewed a person who was caught by the Census Bureau in 2010 simply making up data, with the employee claiming his superiors told him to do so because his region was not successfully interviewing enough people for the survey. According to an anonymous source, that effort intensified in the months leading up to the 2012 election, with September 2012’s data specifically falsified to President Barack Obama’s favor, and continues to the present time.

These allegations are currently being investigated by both the House Oversight Committee and the Inspector General of the Census Bureau, with the BLS also quite interested in them.

One place they can start is comparing what came out of the CPS to a measure of unemployment conducted by Gallup, started in January 2010. There are a couple of key differences between the CPS and Gallup which make a comparison a bit harder:

– While the CPS uses a reference week that includes the 12th of the month (5th of the month in November), Gallup uses a 30-day rolling average.
– The CPS surveys (or claims to survey) 60,000 people age 16 and over, while over the course of each 30-day rolling average, Gallup surveys 30,000 adults.

Fortunately, the BLS releases, as part of its dataset, the data from the portion of the CPS that covers adults, or about 57,500 surveyed out of approximately 153,000,000 considered part of the labor force. That allows an apples-to-apples comparison:

Gallup-CPS

For the most part, the CPS measure of adult unemployment is significantly lower than Gallup’s measure. How significant? Let’s take a look at Gallup’s measure on the day that puts the CPS reference day right in the middle of the rolling average, the 27th day of the month (20th for November and also December to avoid an artificial post-Christmas spike). The CPS unemployment was calculated by dividing the number of unemployed by the number considered to be in the workforce, so I could get much closer than to the nearest tenth of a percentage point reported. The raw data was not available for Gallup’s measure of unemployment, so I took the closest possible number to the CPS measure that still rounded to Gallup’s tenth of a percentage point reported.

Gallup-CPS divergence

Polls, which is what the CPS and Gallup measures really are, come with a margin of error, within which the true value can be expected either 90% or 95% of the time. For the CPS, the 90%-confidence margin of error is +/-0.20 percentage points and the 95%-confidence margin of error is +/-0.24 percentage points. For Gallup, the 90%-confidence margin of error is +/-0.28 percentage points and the 95%-confidence margin of error is +/-0.34 percentage points.

Two polls are considered to be in good agreement when their values are within each others’ margin of error. Meanwhile at least one poll has to be wildly incorrect when the difference between the two is more than the sums of their margin of error. Out of 46 months’ worth of data:

– 18 months saw Gallup’s and CPS’s measures of unemployment disagree by more than the combined 90%-confidence margin of error of 0.49 percentage points, with 17 months having Gallup’s measure higher.
– 8 months saw the measures of unemployment disagree by between 0.28 percentage points (Gallup’s 90%-confidence margin of error) and 0.49 percentage points, with another 3 months seeing a disagreement between 0.20 percentage points (CPS’s 90% confidence margin of error) and 0.28 percentage points.
– 17 months saw the measures of unemployment in “good agreement”, disagreeing by less than 0.20 percentage points.

When two polls wildly disagree more than they are in “good agreement”, one of them has to be wrong. Given the disagreement has been almost invariably in the administration’s favor, and there already was a proven round of fakery in the CPS, it sure looks like the official measure of unemployment has been cooked longer than a burnt turkey.

****************************************

Olimometer 2.52

It’s Saturday the last day of the month and the week.

Most important of all it is the end of the first month of my Magnificent Seven Bloggers.  Later today the final post of the month for my magnificent seven will go up and I’ll send out my final payment of the month for that final article paying them in full.

You have been treated to topics like Church history from Pastor George Kelly Common Core’s Kobayashi Maru from AP (Lady Liberty) Dillon , a bit of Venezuela from Fausta, Indoctrination vs education from Baldilocks The sanity of homeschooling from Linda Szugyi , What Smoking Taxes actually do to a culture from Marathon Pundit and the Left’s Lawfare in Wisconsin from Steve Eggleston and today you will get one more piece from Steve rounding out the contributions from my Magnificent Seven.

Unlike the litigious bloggers of the Huffington Post my seven have not just been promised exposure, each has been paid for their work since day one.

That’s why AOL will never come calling with $315 Million or even 2 Million because if they did they’d not only would they have to pay my writers, but I’d likely just start a new even bigger venture along these same lines with the dough.

It would be very cool if eventually the story ended with a huge site, linking hundreds of conservative bloggers and with our own staff of paid reporters, but for now we’re still the beginning and on this last day of the first month of my Magnificent Seven I’m $179 shy of ending that first month with all my writers paid in full AND the mortgage paid in full too.

One tip jar hit of $179   or two at $90 or Four at $45 will get it done.

I can’t promise that this will eventually grow into that ultimate conservative powerhouse driving supporting and electing conservatives as my ultimate plan, but succeeding in the small things will make the large things possible.

Two final thoughts

We are offering sponsorships of both the Magnificent Seven & the Magnificent Panel now is an excellent time to jump on board, contact me here for more details

If I can get 63 more subscribers at $20 a month then the mortgage and the magnificent seven will be paid in full EVERY MONTH.