One of the first exercises I was given in my first computer science class in 1962 was to program a correlation between the high school GPA, and the ACT and SAT scores of 20 or 30 college freshmen with their first year grades. Once we got the "best fit" we were then to apply our answers to another batch of 100 freshmen to see if the model was a good one. The idea of the exercise was to show how the brute force of computational power could solve a problem that would normally be the target for sophisticated mathematical and statistical formula called regression analysis. And the problem was explained as a timely and practical one.
I find it fascinating that 45 years later universities and businesses are still struggling with the "best fit" for predicting success in school or career choices. Ideally, we would like to know which high school graduate will find success at MIT so that we don't end up with failing students, impoverished parents, and frustrated teachers. Likewise, there are specialized paper and pencil tests to supposedly select people most fit for fire fighters, swimmers, wine tasters, baseball umpires, plumbers, ball room dance instructors, Bible scholars, and art collectors. It's as if we are afraid to use observation and experience as criteria when a nice "objective" test can shield us from charges of discrimination.
The problem is that some people test poorly even on subjects they have down cold. And some people test well on subjects in which they are walking disasters. Although virtually all colleges require either the SAT or the ACT, it has been shown again and again that neither predicts college success as well as high school GPA does and none of the three predicts ultimate career success.
In "Standardized Minds", after convincing us that there is a problem with using standardized tests, Peter Sacks explains the difficult task of turning around a battleship. There is a huge vested interest in the testing industry and traditions that aren't easily thrown out. "Standards" based on test scores don't equate well with standards based on observed performance. And there's always a lawyer or two in the wings waiting to yell "foul".
I found Standardized Minds an interesting read, maybe twice as long as it needs to be, but still interesting.
No comments:
Post a Comment