I would agree it is hard but not impossible to test for. The test being very long, at some point people just figure no one can finish it... and in some ways that is how I feel about work in life.

I imagine if you told people "we are not testing for correctness, just number of problems solved" then you would skew this test.

However when the tests were administered in the longitudinal study, the researchers didn't know which metrics would be important so they just put a lot of problems on it.