Kimberly Swygert of Number 2 Pencil has a masterful discussion of a recent study in the Harvard Educational Review claiming the SAT is biased against blacks. (Available only to subscribers)
It’s fascinating, even to us non-psychometricians. One highlight: the solution to the alleged test bias proposed by the author of the study is to limit the SAT to questions that are so difficult that virtually everyone is forced to guess the correct answer to virtually every question, which would (among other things, such as making the results useless) eliminate all group differences in scores.
Mr. Freedle’s hypothesis is clever, though certainly not remarkable, and amounts to little else than more surreptitious maneuvering to define aptitude away from itself.
I wonder how an analagous dissection of the 100 meter dash would be performed. Maybe compile a database of all varsity sprinters by race and time. Then parse each individual’s run into 25 meter quartile splits. What if, for equal 100M times, we found that blacks ran slightly faster in the first split, while whites (or Asian-Pacific-Islanders) outperformed in the final split?
Now, if you are willing to believe what your own eyes show you, you know that blacks vastly outperform all other races in short-distance sprinting at the world-class level (and all other levels). This must indicate some heretofore undiscovered bias in the 100M sprint itself. Otherwise, what accounts for the manifest performance discrepancy?
But back to the hypothesis–Mr. Freedle might then conclude that the “easier” part of the race (the initial 25M) is biased against white runners due to (insert cultural psychobabble). Therefore, to mitigate this bias, only the final split should be considered for purposes of determining a winner. Yet if even this results in continued racial inequities, Mr. Freedle may suggest an expansion of the range of “hard” splits to be timed. E.g., the split for say the 25th to 26th mile. Eventually, were this fully implemented, the 100M sprint would hardly be recognizable–or valid as a tool for measuring short-distance sprinting. But that’s irrelevant, cosmic equality would be achieved.
“Cosmic equality” – that’s priceless! The term definitely highlights the absurdity of parsing performances down to the micro-level and then reassigning points in an attempt to make everyone equal.