new | show | ask | jobs Github

topspin 4 days ago

[flagged]

▲

scarface_74 4 days ago | parent | next [-]

The President just fired the person in charge of reporting jobs numbers because he didn’t like the report.

Logically what do you think is about to happen?

▲

trealira 4 days ago | parent | prev | next [-]

Yeah, firing the labor statistics head because Trump say she's been faking the numbers to make him look bad actually makes it seem both obviously politically motivated and casts whatever comes after into doubt. Now their credibility is degraded.

That's different from just saying the numbers are obviously being faked under Biden or whatever with no real evidence because you just feel like the economy is bad and assume corruption. Now there actually does seem to be corruption!

▲

esseph 4 days ago | parent | prev | next [-]

It IS THE GOVERNMENT

▲

shmerl 4 days ago | parent | prev | next [-]

Yeah, tin party soldier never questions anything.

	▲	topspin 4 days ago \| parent [-]
		Also, whataboutery.

▲

hyperadvanced 4 days ago | parent | prev [-]

On this topic- last year it was somewhat common for R politicians to criticize the D regime for the “report high, revise low” strategy - if anything, I guess, this firing has been telegraphed. Anyway, those people were also called conspiracy theorists and politically motivated. There’s clearly a conflict of interest between the facts and what is politically expedient on all sides of the political system in the USA.

I personally would prefer that the jobs numbers apparatus was extremely conservative in the sense that it didn’t overstate the strength of the USA economy. I doubt Trump has that goal in mind necessarily, laudable as it might be.

▲

tbrownaw 3 days ago | parent [-]

> I personally would prefer that the jobs numbers apparatus was extremely conservative in the sense that it didn’t overstate the strength of the USA economy.

This sounds like a call for it to be biased low? I'd prefer zero net bias - overstate and understate equally often - and as little error as possible. (Also, I'd like a pet unicorn for the munchkin.)

The problem seems to be around the preliminary numbers and how widely they get reported. Maybe this is a case where excessive transparency and reporting partial data that's known to be inaccurate is negatively useful?

Or maybe there could be some way to get people to accept that it's known to be wrong, and only useful if you have the chops to account for that wrongness in whatever you're using it for? But humans in general seem to mostly be allergic to not knowing things, so...

Maybe the wrongness is predictable enough to model and account for, but publishing "expected correction" numbers along with the preliminary numbers would be extremely un-conservative in that doing that is speaking with your own voice rather than just collecting and reporting data.

	▲	hyperadvanced 3 days ago \| parent [-]
		Biased low if it must be - a more down-home way of phrasing that is “don’t count your chickens before they hatch,” which, in any case seems like better labor-statistical common sense than the other way around.