@Data10 I am certain sampling bias plays a role, but there are a lot more variables to consider. We know for example standardized tests are less accurate at the tails than the centre. We also know that tests administered to children are less accurate than those administered to adults. In more recent time, we also learned that heredity plays a greater and greater role as we mature. Personality and opportunity (luck) must also be accounted for etc.
To reduce the “noise” inherent in the data, we need to focus more on meta-analysis and the test data need to be current. I think a personality inventory and structured interview are needed as well. In short, we need more tests, not less. Here is an article I was looking at lately. It is an excellent summary of what I always suspect and knew. Curious what your criticism of it would be.
http://www.annualreviews.org/doi/full/10.1146/annurev-orgpsych-031413-091255
In brief, my opinion is that standardized testing reduces personal bias. It seems like I have much less faith in subjective evaluations than you do, but I am sure this world is big enough for more than one opinion though.
Got to go.