Why is there a focus on one type of error? Is there some reason why doctors would be biased toward wrongly denying disability benefits, rather than wrongly allowing?
What are the economic incentives to bias the decision one way or another?
The article does not go into detail, but I expect a doctor with a high false positive rate to not get much applications his way in the future as it costs the state a lot of money.
This is a key point, and points to so many thing being broken at the same time it hurts my head. The real smoking gun would be to find some sort of defacto evidence that doctors got the hint that denial gets them more business. That would sit in the next layer up, the assignment of cases. Alternatively, could it just be a big pool e.g. the race to get through them is a cash grab, so the faster you go you'll get to drink more of the milkshake? In that case denial is more of a way to avoid scrutiny.
Cutting government staff doctors in favor or more expensive contractor doctors is a red flag. You can't fire a staff doctor for approving too many applications, but you can direct more contract business towards doctors with a proven track record of denials.
It is also possible that there is some form of administrative punishment that makes false positives more burdensome than false negatives. If you are fined $1 for each false positive, and $0 for each false negative, you deny by default. If approvals are flagged for automatic recheck by the computer, and denials are only rechecked if an appeal is filed, you deny by default and let the real work fall to the appeals doctor. If the system branches in any way based on the APPROVE/DENY bit, that may affect how the bit is evaluated in the first place.
It's broken all the way down to the foundations. The state government in Tennessee is controlled by the state politics. And the state politics is very far from liberal democratic socialism. This is the state that had to have a literal gun battle in order to fight local government corruption [0], and is is equally important to note that shooting live rounds at the bastards didn't even get them to behave for an entire year. The veterans' coalition assembled to vote out the corruption was itself co-opted and re-corrupted almost immediately.
I can't say for sure, but I assume they begin with a presumption of each claim being invalid, then looks for proof that the claim is valid. Faster review pays more, so there's incentive to go fast, but is more likely to miss relevant positive proof of the claim.
What are the economic incentives to bias the decision one way or another?