DUNS numbers are incorrect in `project_orgs` table

[jayhesselberth]

Sent the following to the NIH EXPORTER help folks on 2015 Mar 11, still waiting for fix.

For reference, I generated an R data package that pulls in the CSV formatted data from 2000-2014:

https://github.com/jayhesselberth/nihexporter

The problem I identified is that there are several organizations that have apparently been assigned to the same number in the PROJECTS tables. Given that this is supposed to be the authoritative number for cross-referencing with institution information, it would be nice if this were fixed.

For example, if I look at all of the organizations that have been assigned to DUNS number 001910777 across all fiscal years, I get the following result:

org.duns org.name count 1 001910777 JOHNS HOPKINS UNIVERSITY 3905 2 001910777 UNIVERSITY OF TEXAS MD ANDERSON CAN CTR 3912 3 001910777 UNIVERSITY OF VIRGINIA CHARLOTTESVILLE 5864 4 001910777 OSEL, INC. 33

In this case, Johns Hopkins is the correct one, but there are actually more assigned to UVA. In fact, UVA is only assigned to this DUNS number in the PROJECT tables, but it's actual DUNS number is 065391526.

Moreover, if I look up DUNS number 065391526, I get the following result:

org.name n() 1 UNIVERSITY OF COLORADO DENVER 7642 2 UNIVERSITY OF VIRGINIA 1328

with some of the DUNS numbers hitting UVA (under a different name though), but more with University of Colorado Denver.

I looked up a particular UVA grant (F31AT000058) in NIH REPOTER, and REPORTER has the correct UVA DUNS number on the DETAILS page. I would have though these are pulled from the same database, so it seems like the EXPORTER export of DUNS numbers is not working correctly.

[jayhesselberth]

Sent follow-up email on April 15. Still no word on whether this will be fixed.

[jayhesselberth]

This is now resolved by EXPORTER and the new tables will be released in v0.7