Fun with numbers
Tuesday, Jul 31, 2018 - Posted by Rich Miller
* Greg Hinz…
One study released each month is based on a survey of households. Known as the current population survey, it’s best known as being the basis for the unemployment rates that come out 12 times a year. The study also includes figures on local employment, and in the case of metropolitan Chicago the news is bad.
Specifically, the latest monthly survey indicated that compared to a year ago—click on Chicago-Naperville-Arlington Heights seasonally adjusted data to get an Excel file containing the numbers—the metro area has lost jobs, with employment dipping from an estimated 3,594,100 to 3,589,700. Given national growth of about 1.5 percent in that period, the drop here is awful.
However, BLS does another report each month that IDES uses to compile its metro employment report. In June, according to that report, employment in the Chicago-Naperville-Arlington Heights area grew 42,300, almost 1.2 percent, to 3.829 million.
So, Chicago lost 4,000-plus jobs. Or it gained more than 42,000. Both can’t be right, can they? […]
A BLS spokesman emphasizes that the household survey uses a significantly smaller sample than the employer survey, so its margin of error is larger. With a current margin of error in metro Chicago of plus or minus 63,000, the finding that the region lost more than 4,000 jobs is compatible with the region also being found to have gained 42,000 jobs.
Emphasis added.
- wordslinger - Tuesday, Jul 31, 18 @ 9:55 am:
Yes. The base is rather large. And monthly numbers are rather a tiny snapshot.
Monthly numbers are just fuel for the tronclodytes to set their hair on fire, when it suits their agenda.
- 47th Ward - Tuesday, Jul 31, 18 @ 9:56 am:
===the finding that the region lost more than 4,000 jobs is compatible with the region also being found to have gained 42,000 jobs.===
Oh, OK. So is this good news or terrible news?
- Texas Red - Tuesday, Jul 31, 18 @ 10:09 am:
” With a current margin of error in metro Chicago of plus or minus 63,000, the finding that the region lost more than 4,000 jobs is compatible with the region also being found to have gained 42,000 jobs”
Hence the saying “close enough for Government work” - which in an older age actually meant high attention to detail!
- Stuntman Bob's Brother - Tuesday, Jul 31, 18 @ 10:17 am:
Would an equally, or even more useful measurement be “total payroll dollars”? It seems like that would better capture economic activity better than the number of jobs alone. Not sure how that could be tracked accurately, unless the IRS could break out withholding taxes or something - as shown in the post, statistical sampling is a SWAG.
- wordslinger - Tuesday, Jul 31, 18 @ 10:24 am:
–Hence the saying “close enough for Government work” –
Dude, the base they’re working with is 3.6 million.
What do you want, the government to install a computer chip in everyone to track their every movement?
- Precinct Captain - Tuesday, Jul 31, 18 @ 10:32 am:
wordslinger - Tuesday, Jul 31, 18 @ 10:24 am
That answer is probably yes, if the targets are black or brown
- DuPage - Tuesday, Jul 31, 18 @ 10:38 am:
Some workers are keeping a low profile. They fear being hunted down by Trump/ICE, and being torn away from their families.
- Da Big Bad Wolf - Tuesday, Jul 31, 18 @ 10:39 am:
The base is more like nine and a half million. The MOE of 63,000 is pretty good considering. You can’t have statistics without MOE. It has nothing to do with attention to detail or government work.
- Annonin' - Tuesday, Jul 31, 18 @ 11:05 am:
Survey…a lot like sample size in pollin’ — not a hard count
- NoGifts - Tuesday, Jul 31, 18 @ 11:10 am:
And I guess it makes me wonder what is “a job.” If two part time positions are combined into a full time position, is it two jobs then one job? Or is the job a job no matter how many people fill it? Is it just the count of people who say they have a job * the number of jobs they go to? If the population survey is designed to count the unemployment rate, is it a good survey for also counting the number of jobs? Maybe we are just abusing the data.
- Archpundit - Tuesday, Jul 31, 18 @ 11:17 am:
BLS is one of the most accurate and reliable government agencies worldwide. They are the standard. As mentioned by other having an MOE is part of that accuracy and reliability.
Slagging on government agencies for being transparent in their processes is both wrong and dumb.
If you don’t understand, learn.
- Archpundit - Tuesday, Jul 31, 18 @ 11:19 am:
—–And I guess it makes me wonder what is “a job.”
They have definitions and are quite good at counting whether jobs are part or full time, but also looking at underemployment, etc. The above is just a small description of one aspect of the survey.
- wordslinger - Tuesday, Jul 31, 18 @ 11:26 am:
–The base is more like nine and a half million–
Employed base. “….with employment dipping from an estimated 3,594,100 to 3,589,700.”
- NoGifts - Tuesday, Jul 31, 18 @ 11:26 am:
I’m just saying looking at two different data sources and comparing their totals is not helpful because the surveys have different purposes and likely have different definitions. So the answer is probably “yes, they can both be right.”
- EconLady - Tuesday, Jul 31, 18 @ 4:10 pm:
Statistics is an actual science. The findings from a statistical analysis are not pronouncements. They are rigorous best guesses.