I agree on the studies that are currently ongoing -- the grant was awarded to the PI, not to the institution.
But this whole question of who 'owns' the data from research has been coming to ahead for a while. Common arguments are for one of:
* The PI
* The PI's institution
* The funder
The problem is that for years, the disposition of the data was never spelled out clearly in the RFPs. Most people had never heard of a DMP (Data Management Plan) until NSF started requiring them a few years ago.
So ... we get into the problem that because each grant can come up with a different DMP, we have to look to those to see who is the gatekeeper of the data. In some cases, the data is handed off to an IR (Institutional Repository; typically something managed by the library), and if that's spelled out in the DMP, then I'd say that the institution keeps control of the data. In some cases, it all needs to be sent back to the funder (NASA instrument contracts are like this, where the 'final data' must be deposited back to an ARC (Archive Resource Center)). But there might be other ones where the PI is personally responsible for access to the data.
Personally, I prefer the IR or funder, just because most scientists have no clue what they're doing when it comes to archiving data. See Data Sharing and Management Snafu in 3 Short Acts. You also run into problems when PIs retire / die / move / whatever. ... but I don't deal with medical data where you need to have an active gatekeeper (IRB, Institutional Review Board, or similar) where you might need someone with better understanding of the data.
So anyway ... without there being something specifically in the grants, the institution likely can lay claim to keeping a copy of the data ... but I don't know if they can necessarily stop the PI from taking a copy with him, or even the server holding it (if the hardware was paid for through his current grants). They *might* be able to get the IRB involved, and insist that they need to review what's being done with the data that's being moved.
In this particular case, though ... it's not only an NIH grant (which are clearly to the institution), but the PI has only been there for 8 years -- so he's taking data that was collected by previous PIs before him. I'd say that his trying to take all data from the department, and not just that which he was PI for is a rather sleazy move.
(disclaimer : I'm one of the moderators on StackExchange's Open Data site )