BoardEx–DIME Crosswalk

In my paper “The Political Transformation of Corporate America, 2001–2022” (available here), I use supervised machine learning to merge two widely used datasets: BoardEx, which contains data on corporate directors an executives, and the Database on Ideology, Money in Politics, and Elections (DIME), which contains data on campaign contributions and ideology scores for campaign donors in the United States. I have made the crosswalk linking these two databases freely available for anyone to use. The latest version links to v4 of DIME and runs through the end of 2024.

You can download the crosswalk here.

If you use the crosswalk, please cite: Steel, Reilly S. 2025. "The Political Transformation of Corporate America, 2001–2022." American Political Science Review.

I’d love to hear about how you’re using the data, so please feel free to shoot me a note and let me know!

Frequently Asked Questions:

Q: The crosswalk doesn’t produce any links! What’s going on?

A: There are multiple versions of the crosswalk, corresponding to different versions of DIME. Make sure you’re using the right one.

Q: I’ve noticed that a BoardEx directorid number is sometimes linked to multiple DIME bonica.cid numbers. What’s going on?

A: There are multiple possible reasons why a single directorid may link to multiple bonica.cid numbers. First, Bonica’s entity-resolution algorithm sometimes assigns multiple bonica.cid numbers to the same individual (e.g., when a person moves and changes jobs; for example, I am personally in DIME with one bonica.cid for when I was living in Vermont and another bonica.cid when I was living in New York), whereas directorid numbers are (as far as I can tell) truly unique because human analysts ensure that each individual receives only one directorid. This virtually guarantees that there will some directorid numbers will have multiple links (as they should). What to do with that fact is up to the researcher. In the APSR paper, when an individual was assigned multiple bonica.cid numbers, I took a weighted average of the CFScores for each bonica.cid. Second, there could be an error in which I’ve erroneously linked a directorid to a bonica.cid. The validation exercises (described in the Supplemental Materials) suggest that such errors are rare, but I don’t doubt that it has happened.