Archive for the ‘Statistical Bias’ Category

Biases: The favorite-long shot bias …

June 5, 2015

In gambling and economics, there’s an observed phenomenon favorite-long shot bias.

image

Here’s how it works …

 

(more…)

Biases: The favorite-long shot bias …

November 12, 2014

In gambling and economics, there’s an observed phenomenon favorite-long shot bias.

image

Here’s how it works …

 

(more…)

Biases: The favorite-long shot bias …

July 3, 2014

In gambling and economics, there’s an observed phenomenon favorite-long shot bias.

image

Here’s how it works …

 

(more…)

Biases: The favorite-long shot bias …

March 27, 2014

In gambling and economics, there’s an observed phenomenon favorite-long shot bias.

image

Here’s how it works …

 

(more…)

BLS: “Oops … we found some initial unemployment claims that we forget to report before the election”

November 15, 2012

Unbelievable !

Now that the election is done, the BLS has “caught up” on initial jobless claims reporting … their words, not mine.

Here’s a shocker …

They’ve figured out that unemployment is more of a problem than they’ve been reporting.

In the week ending November 10, the advance figure for seasonally adjusted initial claims was 439,000, an increase of 78,000 from the previous week’s revised figure of 361,000.

The 4-week moving average was 383,750, an increase of 11,750 from the previous week’s revised average of 372,000.

The consensus forecast for this week –- based on prior weeks’ reporting – was 375,000 … 64K lower than the BLS’ surprise number.

I say: Let’s raise taxes and get this economy moving again …

= = = = =
Technical note: While blame will be laid on Hurricane Sandy, keep in mind that (1) hurricanes temporarily boost employment of construction & trades workers, and (2) the affected areas were without electricty and many government offices (e.g. FEMA outposts) were closed … so, these initial unemployment claims are probably under-reported (as usual) … the fuller impact of the hurricane will show up in the next couple of weeks.

= = = = =
Ohio & PA

The highest numbers of new filings came from Pennsylvania and Ohio, where there were thousands of layoffs in the construction, manufacturing, and automobile industries.   During his campaign, President Obama highlighted his record of job creation in those states —  Ohio in particular.  Source

Oops.

* * * * *

Election predictions: The favorite-long shot bias …

November 9, 2012

Punch line: For some folks who predicted a Romney win over Obama, it  was simply heart over head.

For others, it may have been the favorite-long shot bias

= = = = =
In gambling and economics, there’s an observed phenomenon favorite-long shot bias.

image

On average, bettors tend to overvalue “long shots” and undervalue favorites.

That is, in a horse race where one horse is given odds of 2-to-1, and another 100-to-1, the true odds might for example be 1.5-to-1 and 300-to-1 respectively.

Betting on the “long shot” is therefore a much worse proposition than betting on the favorite.

Various theories exist to explain why people willingly bet on such losing propositions, such as risk-loving behavior, or simply inaccurate estimation.

Source

>> Latest Posts

Smoking gun: BLS streak comes to an end … coincidence?

November 8, 2012

As Gomer Pyle would say: Suprise, suprise, suprise.

This is absolutely unbelievable …

The BLS streak — understating initial unemployment claims – ended this week.

In all the prior 26 election season weeks, the BLS’s “headline number” under-reported initial unemployment claims … and cast the jobs situation as brighter than it really was.

The election was Tuesday, right?

Well, guess what.

Here’s what the BLS report this morning … read it carefully.

  • In the week ending November 3, the advance figure for seasonally adjusted initial claims was 355,000, a decrease of 8,000 from the previous week’s unrevised figure of 363,000.

English translation: Some how, the BLS was miraculously able to eliminate the reporting bias that had been consistently evident in the run up to the election.

Frankly, I’d expected them to wait a few weeks to create some distance from the election … then “modify” their reporting.

Nope.

Tell me again how the BLS is just a group of apolitical bureaucrats cranking out consistently reported facts.

The good news is that I can finally stop tracking and reporting the streak.

Quick, somebody call Jack Welch…

= = = = =

image

>> Latest Posts

Old song … BLS still under-reporting initial unemployment claims.

November 1, 2012

Yeah, yeah, yeah … I’m getting as tired writing about it as you’re getting reading about it.

But, the BLS streak — understating initial claims – continued this week.

Now we’re up to at least 26 election season weeks in a row that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on today’s BLS report, the number for the week ending October 20 was revised upward from 369,000 to 372,000 … making this week’s headline look 3,000 better.

These guys can’t be this sloppy or this stupid … can they?

image

= = = = =

To put today’s number in context. let’s flashback to the chart in yesterday’s post

The reported 4-week moving average is 374,000.

So, if the relationship of inital unemployment claims and the unemployment rate holds, tomorrow’s BLS report should be an unemployment rate of about 8.1%

That’s what all of my analyses say that the number is … but I’m still betting the under.

I think the BLS will fudge the numbers to keep the unemployment rate under 8%.

We’ll see tomorrow.

image

>> Latest Posts

Hard to believe … the BLS streak rolls right on.

October 25, 2012

I though the BLS might find some old time religion – or at least hire a new stats guy – since they got hammered on the incredible 7.8% unemployment number.

Not so.

And, the BLS streak — understating initial claims – continued.

Now we’re up to at least 25 election season weeks in a row that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on today’s BLS report, the number for the week ending October 13 was revised upward from 388,000 to 392,000 … making this week’s headline look 4,000 better.

Glad the election is only 12 days away.

Wanna bet that the BLS makes a post-election change to their methodology?

image

>> Latest Posts

Hoisted by their own pitards … BLS unemployment claims are in.

October 18, 2012

Last week, Team Obama was crowing about the huge drop in initial unemployment claims … proof poitive that the recovery was gaining steam.

They failed to mention the fact that the state of California sat on a pile of claims … making the numbers look better than they really were.

Well, as Rev. Wright would say, the chickens have come home to roost.

According to the BLS: “In the week ending October 13, the advance figure for seasonally adjusted initial claims was 388,000, an increase of 46,000 from the previous week’s revised figure of 342,000.”

Oops.

Headline: “Jobless claims increase 46,000”

Not exactly proof positive of an economy gaining steam.

My bet, Team Obama emphasizes that claims are overstated because of California.

Too bad.

* * * * *
And, yes … the BLS streak — understating initial claims – continued.

Now we’re up to at least 23 election season weeks in a row that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on today’s BLS report, the number for the week ending October 6  was revised upward from 339,000 to 342,000.

C’mon guys … get it right already.

image

>> Latest Posts

Unemployment claims are down (if you don’t count California) … and, yes, the BLS streak continues.

October 12, 2012

The BLS would morph into a punch line if the stakes weren’t so high.

Let’s do the easy part first.

Now we’re up to at least 22 election season weeks in a row that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending Sept. 29 was revised upward from 367,000 to 369,000.

I’ll complain to the BLS Commish when President Obama appoints one.

see the HFs post: BLS Commissioner’s post vacant since January

image

* * * * *

Bigger Issue this Week

This week, the BLS reported spectacularly good news …  claims down 30,000 (after revising last week’s claims up).

While the BLS report failed to mention the point, somewhere between 15,000 and 25,000 claims from California weren’t processed in time to be included.

Say what?

Business Insider did a nice job decoding the situation:

Some of the jobless claims in one large state–California–were not included in the claims reported to the Department of Labor this week. 

When a state’s jobless claims bureau is short-staffed, sometimes the state does not process all of the claims that came in during the week in time to get them to the DOL.

Our source [at the BLS]  believes that this is what happened this week.

The California claims that were not processed in time to get into this week’s jobless report will appear in future reports, most likely next week’s or the following week’s.

In other words, those reports might be modestly higher than expected.

Our source believes that the number of California claims that were not processed totaled about 15,000-25,000.

Thus, if one were to “normalize” the overall not-seasonally-adjusted jobless claims number, it would increase by about 15,000-25,000.

This week’s “normalized” jobless claims number, therefore, would be about 355,000-365,000, not the 339,000 that was reported.

Are you kidding me?

And, Business Insider missed a key line in the BLS report:

“The largest increases in initial claims for the week ending September 29 were in New York (+2,764) and  California (+2,069)”

So, the missing California claims may be even higher … if the missing regions kept pace with the rest of the state

This is getting silly.

>> Latest Posts

Why would anybody distrust the BLS numbers?

October 8, 2012

The BLS is  an independent organization that just reports the facts, right?

Former GOP administration insiders are coming to the BLS’ defense, testifying that the number crunchers are innocent as babies.

So, why should anybody be suspicious just because  the “household survey” is giving answers that conflict with the “establishment survey” and  is reporting job gains greater than in any other month for the past 30 or 40 years?

Here are four documented reasons to be skeptical:

1. The administration has used bullying tactics with outside pollsters – specifically Gallup

Recently, Team Obama didn’t like Gallup’s polling numbers.  So campaign chief Axlerod called them to provide some statistical counseling, and and Attorney General Holder launched a DOJ investigation of Gallup’s human resource practices. Suddenly, Gallup’s  poll numbers turned more favorable to the President, including a 1-day 12 point improvement in Obama’s approval rating.  Coincidence?   Source

2. The administration has been bullying defense contractors to violate Federal employment laws.

ABC is reporting that the White House has told defense contractors to not issue layoff notices until after the election. They even went as far as to offer to pay for any legal fees associated with their violating the law by not giving employees proper notice. Specifically, “defense contractor Lockheed Martin heeded a request from the White House – one with political overtones – and announced it will not issue layoff notices to thousands of employees just days before the November presidential election.” Source

3. Obama has used bullying tactics with other “independent government agencies – specifically the CBO.

The Congressional Budget Office is supposed to be strictly objective, and completely detached from the Administration. But, during the ObamaCare debate, when the CBO numbers weren’t looking favorable, the President ordered CBO director Doug Elmendorf to the White House for counseling.  The next week, the CBO revised its numbers.  The new estimates were  more favorable towards ObamaCare.  Another coincidence? click for news report

4. The BLS has a recent track record of questionable numbers.

Most important, for at least the last 22 election season weeks, the same BLS that reports the unemployment statistics has systematically underreported weekly initial unemployment claims by an average of roughly 1% – about 3,000 claims per week – and then revised the estimates up the next week.

Why is the preliminary under-reporting a problem?

Because each week’s “headline” number of changes in unemployment claims is derived by taking the current week’s preliminary number and comparing it to the prior week’s revised number.

For example, in the week ending September 22, the preliminary number (367,000) was compared to September 15th’s revised number (363,000) and and 4,000 drop in unemployment claims was reported. The September 15 preliminary number – the basis for the September 15 report — was 359,000. So, comparing preliminary estimates for the two weeks, unemployment claims increased by 8,000 not 4,000. And, based on the past 22 weeks of initial underreporting, that number is likely to swell when the September 22 number is revised – most certainly upwards.

image

* * * * *

My point: the Administration has demonstrated a willingness to bully supposedly independent groups, both in and out of government.

And, the BLS has exhibited some curious statistical reporting.

Still believe the latest unemployment report?

>> Latest Posts

It’s Thursday, so … guess what?

October 4, 2012

Yep, the BLS announced this weeks initial unemployment claims, and you know what?

They revised last week’s headline number up.

Now we’re up to 81 out of 82 weeks — and, at least 22 election season weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending Sept. 15 was revised upward from 359,000 to 363,000.

In itself, the 4,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

I’m now starting to conclude the latter.

The BLS has plenty of statisticians on payroll … and this is an elementary stats problem

* * * * * *

Let’s try a new way of reporting … here’s a picture.

image

Note that the preliminary estimate (the blue line) is ALWAYS low … by a couple of thousand.

Hint to BLS: just add 2k or 3k … or .8% to your prelim forecast !

>> Latest Posts

The last time that I talk about BLS reporting bias …

September 28, 2012

… certainly won’t be today

Unbelievable, they did it again this week.

I promise that I’ll stop writing about BLS reporting bias when the streak ends, but …

Now we’re up to 80 out of 81 weeks — and, at least 21 election season weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending Sept. 22 was revised upward from 382,000 to 385,000.

In itself, the 2,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

I’m now starting to conclude the latter.

The BLS has plenty of statisticians on payroll … and this is an elementary stats problem

* * * * * *

Let’s try a new way of reporting … here’s a picture.

Note that the preliminary estimate (the blue line) is ALWAYS low … by a couple of thousand.

Hint to BLS: just add 2k or 3k … or .8% to your prelim forecast !

image

* * * * *

Here are the nums … but the picture says it all.

image 

>> Latest Posts

Unemployment claims drop by 3,000 … well, not really.

September 21, 2012

Yesterday’s headline’s trumpeted a 3,000  drop in initial unemployment claims.

Hooray. Right?

Of course not, the BLS revised last week’s number up by 3.000 so that it could report this week as being down by 3,000.

Huh?

Now we’re up to 79 out of 80 weeks — and, at least 20 election season weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending Sept. 8 was revised upward from 382,000 to 385,000.

In itself, the 3,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or 3k … or .8% to your prelim forecast !

image

I promise that I’ll stop writing about BLS reporting bias when the streak ends.

>> Latest Posts

Sorry, but the BLS streak continues …

September 14, 2012

I promise that I’ll stop writing about BLS reporting bias when the streak ends.

Now we’re up to 78 out of 79 weeks — and, at least 19 election season weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending Sept. 1 was revised upward from 365,000 to 367,000.

In itself, the 2,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or 3k … or .8% to your prelim forecast !

image

* * * * *

And, oh yeah, the initial jobless claims increased by 15,000 … above the consensus estimates … and consistent with an unemployment rate higher than 8.1%.

image

>> Latest Posts

Sorry to harp … but, the streak continues

September 7, 2012

Some loyal readers have suggested that I get off this case … That I’ve made my point.

I promise that I’ll stop writing about BLS reporting bias when the streak ends.

Now we’re up to 77 out of 78 weeks — and, at least 18 weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on yesterday’s BLS report, the number for the week ending August 25 was revised upward from 374,000 to 377,000.

In itself, the 3,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or 3k …  or .8% to your prelim forecast !

image

>> Latest Posts

Unbelievably, the streak rolls on: BLS under-reports initial unemployment claims … again!

August 31, 2012

Am I the only person in the world to to think this is nuts?

Media sure isn’t reporting it …

Now we’re up to 76 out of 77 weeks — and, at least 17 weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on yesterday’s BLS report, the number for the week ending August 17 was revised upward from 372,000 to 374,000.

In itself, the 2,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or .8% to your prelim forecast !

image

* * * * *

Almost forgot … the preliminary unemployment claims for the week of Aug. 25 are reported even vs. the Aug. 11 preliminary number and up 2K vs the revised Aug. 18 number.

In other words, no indication that a corner has been turned.

image

>> Latest Posts

The streak rolls on: BLS under-reports initial unemployment claims … again!

August 27, 2012

Still again …

Now we’re up to 75 out of 76 weeks — and, at least 16 weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on last Thursday’s BLS report, the number for the week ending August 17 was revised upward from 366,000 to 368,000.

In itself, the 2,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or .8% to your prelim forecast !

image

* * * * *

Almost forgot … the preliminary unemployment claims for the week of Aug. 18 are up 6K vs. the Aug. 11 preliminary number and up 4K vs the revised Aug. 11 number.

In other words, no indication that a corner has been turned.

image

>> Latest Posts

The streak rolls on: BLS under-reports initial unemployment claims … again!

August 10, 2012

I can post this post on auto-generate, I guess …

Now we’re up to 73 out of 74 weeks — and, at least 14 weeks in a row — that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending July 28 was revised upward from 365,000 to 367,000.

In itself, the 2,000 isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or .8% to your prelim forecast !

image

* * * * *

Almost forgot …

The 4-week moving average of initial unemployment claims bumped up 2,250 to 368,250 … suggesting that the corner hasn’t been turned yet.

Amazing: BLS under-reporting streak continues …

July 27, 2012

Now we’re up to 71 out of 72 weeks — and, at least 12 weeks in a row —  that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending July 14 was revised upward from 386,000 to 388,000.

In itself, the 2k isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or .8% to your prelim forecast !

image

>> Latest Posts

BLS streak shamelessly continues …

July 20, 2012

BLS bias continues

Now we’re up to 70 out of 71 weeks that the BLS’s “headline number” has under-reported the number of initial unemployment claims … and cast the jobs situation as brighter than it really is.

Based on Thursday’s BLS report, the number for the week ending July 7 was revised upward from 350,000 to 352,000.

In itself, the 2k isn’t a big deal.

But, in context it is

Again, I ask: statistical bias or political bias?

If the former: fix it already, BLS.

Hint to BLS: just add 2k or  .8% to your prelim forecast !

image

* * * * * *

Increase in Unemployment Claims

And, don’t miss the big point: initial unemployment claims increased by about 10% to 386,000

I expect Team Obama to whine:

“Therefore, it is important not to read too much into any one monthly report and it is informative to consider each report in the context of other data that are becoming available.”

Safe bet since, as we reported before, that’s exactly what they’ve said each month for over 2 years.

image

>> Latest Posts