Achievement tests: multiple choice gives way to exams that also teach
Princeton, N.J. — School reforms may come and go, but after half a century of standardized tests, students can be excused if they clutch their No. 2 lead pencils with the white knuckles of a determined batter. Too often, the pitch of multiple-choice questions, like some academic bean ball -- ``e. none of the above'' -- bounces off the psyche of students.
Fill-in-the-blank exams are symptomatic of an American fixation. If it can be taught, it can be tested, or so conventional wisdom dictates. In the not too distant future, however, students and teachers can expect tests that differ strikingly from the multiple-choice, machine-scorable answer sheets.
The courtship between computer technology and new developments in learning theory is having its effect. The pending marriage, many of whose ``vows'' are being written here at the Educational Testing Service (ETS), promises a ``golden age'' in which examinations will be tailored to the individual, taken by appointment wherever a computer is accessible, and then instantly scored and reported.
``What lies ahead,'' said ETS president Gregory R. Anrig in a relaxed, wide-ranging interview at his office, ``is a new generation of assessment instruments, as well as new methods for making them readily available to individuals and institutions.''
The tests ETS is developing will move beyond the traditional determination of admissions and placement. They will also provide instruction and guidance for students and teachers, Mr. Anrig says.
``It will be possible for a youngster to sit down at a console and, in any academic area, be able to answer some questions [and then], on the basis of those answers, have the . . . program determine what the next question will be,'' Anrig continues.
``You'll now be able to get feedback -- both on the screen and in printout form -- for the student and for the teacher . . . ,'' adds this former teacher of American history, high school principal, and school superintendent. ``The cumulative effect will be truly diagnostic.''
As a result, the student will no longer get just a test score that tells ``here's what the situation is,'' as Anrig puts it. He will also get advice -- ``here's what you can do about it'' -- to improve learning.
``That's terribly exciting,'' he says. ``What this does is bring testing and instruction together, and one begins to inform the other.''
Founded in 1947, ETS has eight regional offices in addition to the corporate headquarters on a rolling, open campus here in Princeton. ETS administers college entrance examinations -- the SATs -- for millions of American students each year, as well as examinations for graduate and professional schools. It also conducts research on testing theory, learning, and educational policy.
Before taking over the helm at ETS in 1981, Anrig worked in Washington as director of the federal Equal Education Office in the late 1960s and later as executive assistant to James Allen, then US commissioner of education. He was a leading figure in the desegregation of school systems in Massachusetts, where as state education commisioner (an office he held for 10 years) he took an active role when Boston schools were under federal court order to desegregate.
Last year ETS administered its English-as-a-second-language tests to more than 440,000 students from 140 countries. Recently it signed a contract with the People's Republic of China to test all students coming from that country to study in American colleges and universities.
In conjunction with the College Board, ETS is currently developing an entry-level college placement test that uses computer ``branching'' techniques, which allow for a variety of lines of questioning, depending on how a particular question is answered. The test is programmed to identify the appropriate level of difficulty for each test-taker and then to continue questioning at that level, making adjustments automatically depending on whether answers are right or wrong. The result, says Anrig, is a test that is more individualized and yet requires possibly 80 percent fewer questions than traditional tests. Nonetheless, it is ``equally reliable and valid.''
Expected by the turn of the century (or before today's second-graders leave school) are computer-scorable tests that will allow for free responses rather than multiple-choice answers. These will rely on computer ``simulations of human thought processes, rather than just recall and recognition on multiple-choice questions,'' Anrig says.
``We can test a lot of things on a computer. The question is: What should we test on the computer? What is it we want to accomplish? And then how do we have to change the computer or software to accomplish that so we don't become captives of the technology, but rather be its master?
``I don't think the tests should determine the curriculum. . . . By and large tests should grow out of the curriculum. That is, the decisions ought to be made locally, and then that should begin to be reflected over time with the kinds of tests that are designed. . . . What's happening right now -- one of the most promising parts of the educational problem -- is that there's a growing nationwide consensus, on at least the high school curriculum, more so than there has ever been probably since 1968,'' Anrig says.
Such optimism does not mean there are not mine fields ahead for diagnostic testing, including problems related to students' privacy rights (``We sign a contract with the individual student when they take one of our tests,'' says Anrig); cultural and racial bias in tests; and misuse of tests (ETS prohibits Texas and Arkansas from using its National Teacher Exam as the sole means for dismissing already-practicing teachers).
Right now, ``one of the fastest-moving changes . . . is in teacher testing,'' Anrig says. He told educators gathered this past weekend at the American Association of Colleges of Teacher Education in Denver, ``The race towards teacher-testing is not only nationwide, it is across the board.''
Such rapid acceptance of the idea that teachers should be tested opens the potential for serious problems, he warned. Twenty-one states now require students to pass a test before entering a teacher-education program. Thirty-two states have or will have by 1988 a testing requirement for certification.
``Accountability for teacher education should rest with the entire college or university, not solely with the teacher-preparation unit,'' Anrig insisted. He pointed out that 60 to 80 percent of the training received by a ``prospective teacher is in academic departments other than the department or college of education. On the NTE Core Battery, for instance, more students tend to have difficulty qualifying on the Test of General Knowledge than on the Test of Professional Knowledge.''
He endorses the recent proposal by Albert Shanker, president of the American Federation of Teachers, for a nationwide ``National Teacher Exam'' as rigorous as a bar exam or a certified public accountant's test. Like Mr. Shanker, he does not want to see federal, state, or local government creating the tests.
Anrig cautions the people rushing to put through teacher tests to keep such testing in perspective. ``What you're measuring is the academic knowledge of the teacher,'' he explains. ``Yes, . . . the teacher has to have academic knowledge. But there's more. You must find out the other stuff. . . . I'd want to know a lot more about what kind of person that teacher is -- what kind of commitments they have, what . . . they value. It's very important to keep these tests in that perspective.
``When I ask myself, who were the great teachers that I had in my life, they weren't great because of how they did on a test.They were great because they were human, caring people, sensitive, dedicated, persevering. When I did poorly, they kept on saying, `You can do better.' Those qualities can't be measured by any test,'' he says.
An area that should be of profound concern to everyone involved in education, he says, is the effect of the recent teacher-testing movement on the access of minorities to teaching jobs.
ETS research concludes that, if current teacher-testing trends continue, the minority teaching force by the year 2000 will be less than half its current level of about 12 percent.
Anrig points out that this decline will take place ``at the same time as the proportion of minority students enrolled in American schools is increasing dramatically.'' The mismatch between the racial and ethnic compositions of students and teachers ``is a matter with serious social and education implications for the nation's schools,'' he says.
Jim Bencivenga is the Monitor's education editor.