Time and time again
The WSJ has an article today on the issue of timing in standardized testing. This is the first article I've seen that brings together the issues of timing in K-12 testing and at the post-K12 level.
Beginning in 2005, the new federal education law will require each state to test every child every year in reading and math and, starting in 2007, in science. The law carries sanctions for the schools -- children in low-performing schools can transfer to better ones, taking their funding with them -- and those sanctions will get steeper every year...Timing standardizes conditions so that student results can be compared among schools and across states. That especially matters on so-called norm-referenced tests like the Iowa Test of Basic Skills, which is published by the University of Iowa, and TerraNova, which is published by McGraw-Hill Cos....To make sure every child has the same chance to do well, most norm-referenced exams come with elaborate instructions about lighting conditions, breaks, classroom temperature and time. "If we want unstandardized measures, we have grades," says Wayne Camara, research director at the College Board, whose SATs are norm-referenced exams.
But there are equally strong operational reasons to time tests, he adds. Schools want to know when their classrooms will be free for other uses, parents want to know when to collect their kids, and the test companies need to budget for proctors. The SAT also worries about a wily test taker walking away with a test booklet if students are allowed to leave whenever they choose. And grade schools worry about giving up teaching time. "There's too much going on in school, too much to learn to give unlimited time. It's wasteful," says H.D. Hoover, professor of education at Iowa and the senior author of the Iowa test.
The 13-part Iowa test is given in 5½ hours and designed so that 95% of test takers finish it, he says. The College Board says 75% of youngsters finish 100% of the SAT, but that it falls a bit short of its other goal -- that 100% of youngsters finish 75% of the test. Test companies insist their exams aren't "speeded" -- that is, they don't make children rush through questions to see how many they can complete in a short amount of time.
Good point. Much of the journalistic discussion I've seen on test time limits doesn't note the distinction between a "timed" test and a "speeded" test. A completely untimed test is, for the reasons given above, unworkable, impractical, and vulnerable to abuse. Good tests are designed so that, although test takers can't lollygag, almost everyone should be able to finish within the time limits.
The article also points out, "...timing complicates testing for disabled children, who account for about 12% of all youngsters nationally", and while the author seems fairly objective, the point of the article seems to be supportive of the policy of removing time limits from tests for all children.
This is a very big issue in testing right now, and I'm glad to see it covered in mainstream media. I don't think that K-12 testing should be highly speeded, but the necessity of time limits will guarantee that some students will not finish. The point to consider is, is speed part of the construct being measured? If it isn't, then the time limit should be set so that almost every youngster, if not all, will finish the tests. However, I'd like to see more debate on whether speed is part of the constructs that we're trying to measure, both with K-12 testing and with college admissions exams. The article seems to assume that speed is never going to be a construct of any of the skills that we want all K-12 students to have. But if we're willing to give children reading tests that allow them to take all day to read a few simple passages, are we educating them to read very well in that case? Are those reading skills going to serve them well in real life?
Frequent readers of this blog know that I often say that problematic test results are not necessarily an indicator that the test is flawed, but instead may be an indicator of something wrong with the learning process behind the skills being measured. Do we have any measure of how fast kids should be able to read? Yes, a child who needs triple time to answer a reading passage is technically "reading", but is does their extreme slowness in this skill perhaps put them at a different level of educational acheivement as youngsters who can read much faster? Shouldn't the number of children who currently can't finish tests under the time limits be a reason to discuss just why they're reading so slowly, and not an argument to abolish time limits altogether?
The article quotes a VP at NCS Pearson who thinks that time has nothing to do with what schoolchildren should know:
But it also highlighted a debate among testing professionals: Why time tests at all? What matters "is that you demonstrate what you know, tell us if you've mastered the content," argues John Twing, vice president for testing at NCS Pearson Inc., which writes and administers the annual Texas state exam, among others. "Time confounds how much you know with how quickly you recall it," he says.
I disagree. Time doesn't "confound" your knowledge. It's true that a test that is highly speeded is going to reward those who think quickly more so than those who think slowly, but speededness and timing are not the same thing, and I think it makes sense to decide how slow is too slow when it comes to academic performance. If it takes a student all day to answer a set of math questions, that could be evidence that the student was taught to solve such problems using a confusing and inefficient method. If we remove time limits, we remove any evidence that students are being taught to think inefficiently.