Monday, May 29, 2006

Testing Is The Messenger

Via Phi Beta Cons, an excellent discussion at Opinion Journal about why SAT scores have fallen. The author makes a point with which I heartily agree, that the reason test scores have declined is not because the test is unfair, but instead because American students are poorly educated in math and reading:

That the new SAT tests more reading comprehension than the old test did is a good thing. Colleges complain that their incoming students don't have sufficient skills to read and analyze the kind of material that their professors will assign them. I hope that the new SAT's emphasis will make students realize that you can't get much of an education if you can't read.

Maybe the decline in SAT scores will force people to notice that their children are not getting good educations. If your children don't read or do math, why would you think that they would do well on the SAT? I would love to get into a time machine and go back to 1960 and give this new SAT to high-school students back then. I suspect that they would do much better than today's students. If we want people to get good scores on the SAT, I have a suggestion. Stop complaining about how unfair the test is and do your homework.

There is one point on which I can offer some additional background. The author responds to people who say that scores have declined because the newest version of the test (introduced in 2004) is about an hour longer than previous versions by saying that his own students show no decline in performance in later sections than in earlier ones. I take this to mean that a kid who misses 8 questions in the first verbal section misses the same proportion of questions in the last verbal section.

I can attest that his experience is not unusual. I have analyzed a sufficient number of student tests (in the thousands) and there is, with the exception of a few outliers, no difference in an individual's performance on the last section of a given type than on the first section of that type. The student's performance is entirely related to the content of the question and how well they have learned that content. If a test contains 10 geometry questions, and the studenty knows plenty of geometry, it is probable that he or she will 8 of the 10 correctly, regardless of when they appear in the test.

What students tend to report is that they get tired later in the test. This report is probably what creates the myth of the over-long test, once parents start to gossip about their kids' SATs. What the kids don't realize, perhaps, and what the parents don't want to know, is that this is usually preemptive excuse making. Their is no decline in actual performance in the latter parts of the exam.

Technorati tags: | |

No comments: