http://www.nber.org/papers/w9873
Good try with this one. In a cursory reading I could see the glaring flaws in that experiment. They even cover them in paragraph 3.5 of their own paper.
"First, our outcome measure is crude, even relative to the previous audit studies."
They go so far as to suggest that their lack of controls would
"As a result, our findings may under-estimate the extent of discrimination."
If that is not a clear statement of expecting certain results before the experiment then I don't know what does. That experiment was designed to achieve a specific result.
I would not consider that study to be legitimate evidence of anything other than a poorly run experiment with a clear agenda.
The line you're quoting, "our findings may under-estimate the extent of discrimination," contextually refers to the possibility that an employer may not see the name or associate the given name with a specific race.
The weaknesses of the overall experiment are certainly listed, and you're right to point out the "crudeness" of the initial measure.
But...here's another from 2010 that more thoroughly examines the issue. .
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2915472/
Take a look if you want to. As I noted before, this thread is far past its expiration date and should die peacefully somewhere in The Round Table.
But, on the topic of institutional racism, it seems much more difficult to utterly disprove its existence than to acknowledge that, at least to some extent, there are elements at play.