I'm a little late with this one. My psychologist readers might find two articles from the SIOP site interesting. The first is a discussion of the legal issues in Ricci The second is an opinion written after oral arguments were presented, but before the court ruled. The second piece provides some interesting information about the construction of the battery that was tossed by New Haven.
I'm not a lawyer, so I have no firm opinion on the legal matter, but the idea that a union can decide composition and weighting of components in an employee selection battery (p.3) is extremely problematic for validity of the selection protocol.
I offer the following extreme example to illustrate what was at issue in Ricci:
Imagine that a city government agrees to a union demand that an IQ test will be used to select hirees for ditchdigging positions. It's a safe bet that the IQ test is going to be much more valid as a race-sorting instrument than a predictor of future job performance for ditchdiggers. But let's say, for the sake of our example, that the city doesn't realize this beforehand. So you end up with the worst of all worlds with this method: high-adverse impact with a low validity coefficient for future job performance. Let's also imagine that an alternate test with less adverse impact and a higher validity coefficient for job performance was available--maybe a ditchdigging simulation.
After the fact, when the results from the IQ test come in, the city's hiring officials worry that they blew it. They wonder if the IQ test did a better job of sorting applicants by race than by future job performance. Fearing a lawsuit, they get a panel together to review the matter. They learn that an alternate battery was available--one that would have sorted less by race while actually doing a better job of predicting job performance. Realizing that this situation is ripe for a lawsuit, the hiring officials decide that it would be best to toss the results and start over using the alternate method that would better predict future job performance and withstand a legal challenge.
My example is extreme and makes the case very easy. Ricci was muddier than that. Whether the battery components and the weighting of components used by New Haven really was that bad and whether an alternate battery or a different weighting would have been better--don't know and can't say. That should be empirically determined.
And that is the testing issue in a nutshell.