People occasionally write about issues like race or sexual preference while declining to show the data behind their analysis. Without data, writers' attempts to show how the world works seem to reveal only their own prejudices.

From the Feministing blog (thanks Geoff):

What sealed the deal for me was when I heard the author got a $300,000 dollar book deal. That is fucking crazy. If he had been a person of color he would have never gotten so much attention or such a hefty book deal.
-- What is the big deal about stuff white people like?

"Never". No matter how fantastic or compelling, they would have never.... The author reveals their own biases here; it appears that they believe that society aligns itself to assign a lesser status to each and every single last person of color. Granted, this bias may exist in the media world, or it may not. We just don't know unless we have data.

One way to test whether publishers exhibit racist pricing runs as: compare advances to total book revenues between different race authors (we probably have to control for genre and for authors filched from other publishers). Non-racist publishers would have the same revenue to advance ratio across groups, racist firms would not. We'd then need to find a genre with a sizeable racial mix of authors and publishers, call the authors and ask them them about the advance (eeps! assume that we'll get the same normal distribution of embellishment/modesty across races), get the book sales from bookscan, and then run the numbers.

Collect the data and use it to speak truth to the world.