Originally published in U.S. News and World Report, April 19, 1992, with co-author Betsy Wagner.
A U.S. News inquiry finds widespread cheating on standardized tests.
Cherokee Elementary School in Lake Forest, Ill, was riding high in the late 1980s. It was named one of the nation’s outstanding elementary schools by the U.S. Department of Education. Principal Linda Chase won kudos as a leading Illinois principal. And Lake Forest school Superintendent Allen Klingenberg was honored as one of the nation’s top 100 education administrators and named Illinois Superintendent of the Year. But Cherokee’s success was a house of cards. In recent months, the embarrassed Lake Forest school board stripped Chase of her principalship for urging teachers to doctor their students’ standardized-test scores in order to inflate Cherokee’s reputation. The board also bought out Klingenberg’s contract.
The scandal in Lake Forest is not an isolated incident in American education. Intensifying demand that the nation’s $228 billion annual investment in public education pay greater scholastic dividends has put tremendous pressure on teachers and school administrators nationwide to raise standardized-test scores, the most quantifiable measure of achievement. Coupled with astonishingly lax security among the nation’s leading standardized basic-skills tests, this pressure has produced a school testing system that is rife with abuse–and consequently less and less useful as a true measure of educational success.
Satisfied customers. Blatant cheating–ranging from supplying students with test answers to actually tampering with answer sheets–is widespread. And in the highly lucrative but intensely competitive testing market, companies are acutely aware that customers are looking for high scores, and they frequently encourage a variety of practices that artificially drive up the numbers. The reality, laments Bonnie Bracey, an award-winning teacher in the Arlington, Va., school system, is that teachers are under the gun to hike test scores in order to “protect themselves.” Consider:
Thirty-five percent of the participants in a 1990 survey of North Carolina teachers reported that they were aware of or involved in test tampering. Forty-three percent said the number of teachers cheating on tests was increasing.
In a national survey of educators in 1990, 1 in 11 teachers reported pressure from administrators to cheat on standardized tests.
In 1989, 12 of 17 Trenton, N.J., elementary schools met the state’s standards for third graders on the California Achievement Test. In 1990, after allegations of test tampering, state officials monitored Trenton’s standardized testing; only three of the city’s schools met the state’s minimum performance standards.
A 1989 survey of 3,000 Memphis teachers produced charges of extensive cheating on the California Achievement Test–including a case of a teacher’s displaying correctly filled-out answer sheets on the walls of her classroom.
As a result of the widespread abuses in standardized testing, parents are denied a true picture of their kids’ academic achievement, and taxpayers are misled about the performance of the schools they fund. What’s more, say educators, standardized-test results typically are not used to individualize instruction and fill in the gaps in students’ skills, as they were originally intended to do. Instead, they are filed or tossed, as teachers begin prepping for the next round of testing. Perhaps most important, at a time when experts are calling on schools to teach more advanced skills and to enliven their classrooms, today’s standardized skills tests may in fact be driving down the level of instruction in many schools.
Growing industry. In the face of demands for greater accountability and higher standards, the amount of standardized testing in the nation’s public schools has skyrocketed in the past decade, to more than 100 million tests annually–at a cost of $700 million to $900 million. The tests are required in virtually all of the nation’s 15,367 public school systems. And more testing may be on the way. In January, a committee convened by Congress, the nation’s governors and the U.S. Department of Education urged the establishment of a new system of national exams. Shortly afterward, the U.S. Senate passed legislation that would set such a system in motion.
Any national testing system will have to confront the many problems plaguing current standardized tests. About 80 percent of these are commercially published and measure students’ performance against that of a national student sample. Five tests dominate the market: the Stanford Achievement Test (the test administered in Lake Forest) and the Metropolitan Achievement Test, both published by the Psychological Corp., a subsidiary of Harcourt Brace Jovanovich; the California Achievement Test and the Comprehensive Test of Basic Skills, published by CTB Macmillan/McGraw-Hill, a division of Macmillan/McGraw-Hill School Publishing Co., and the Iowa Tests of Basic Skills, published by Riverside Publishing Co., a subsidiary of Houghton Mifflin Co. All five tests are largely multiple-choice measures of rudimentary skills. All five are flagrantly abused.
The reason is clear: The stakes in standardized testing have become tremendously high, for everyone from administrators and teachers to parentswho own property within a school system’s borders. To school superintendents, for instance, high scores frequently mean national reputations and big contracts. Allen Klingenberg earned nearly $150,000 in salary and benefits at 2,000-student Lake Forest. School award programs, such as the U.S. Department of Education’s Blue Ribbon Schools program, rely heavily on standardized-test scores in selecting winners. And increasingly, schools are eligible to win state-financed funding bonuses by boosting students’ test scores.
Punitive acts. Educators can also be punished for poor performance on standardized tests. Arkansas lawmakers passed a School Report Card Act in 1989, one of a number of recent “accountability” initiatives that rank schools’ performance, largely on the basis of testing. In Oklahoma, under a 1989 law that instituted testing statewide in grades 3, 5, 7, 9, 10 and 11, low-scoring schools risk sanctions that include takeover by state authorities. At the local level, a number of school boards are drafting superintendent contracts requiring test-score improvements.
The pressure to produce on standardized tests moves quickly down the educational chain of command. To make sure his principals understood his priorities, John Murphy, the former superintendent of schools in Prince Georges County, Md., hung charts with the California Achievement Test scores of each of the county’s 171 schools on the walls of the conference room adjoining his office–where his predecessor had hung student art. In Lake Forest, Superintendent Klingenberg sent congratulatory letters to teachers with high-scoring students. One way or the other, says Don Barfield, a former director of testing in the San Francisco school system, “the message from superintendents is ‘perform, perform, perform on standardized tests.’ It puts the fear of God in principals.”
Tempting the teachers. In the wake of new incentive programs, there’s also greater temptation for teachers to tamper with tests. In 1989, a Greenville, S.C., social studies teacher was fired for slipping students answers to the Comprehensive Test of Basic Skills prior to their taking the test. Under the South Carolina School Incentive Reward Program, the teacher, a former Teacher of the Year at Greenville High School, had received bonuses totaling $5,000 in the two years prior to her dismissal.
Standardized-test scores also go a long way toward defining a community’s identity–and marketability. In Lake Forest, a parent who helped uncover the cheating scandal at Cherokee Elementary School was confronted in a parking lot and told to stop “making trouble.” One reason: community concern about property values. “You’re paying $200,000 of premium for guaranteed good news [on test scores] here,” says Lake Forest resident Robin Mueller.
But despite the premium placed on standardized-test scores, test security in the nation’s school systems is extraordinarily weak. Teachers have ready access to the tests and answers, since the same tests are often used year after year. And they generally administer the tests to their own students in their own classrooms, frequently scoring them by hand–all conditions that make cheating easy. The most benign result of such laxness is that teachers are sorely tempted to teach precisely what they know the test will cover–often to the exclusion of higher skills. Says Lorrie Shepard, a testing expert at the University of Colorado: “As a teacher, you would be foolish if you knew a standardized test had a reading passage about kangaroos and didn’t teach your kids about kangaroos.”
Some states are toughening their test security. Under a 1989 Mississippi law passed in the wake of a testing scandal that year, school systems must have a security plan approved by the state. Community proctors administer tests jointly with teachers. The state’s special-education rolls are cross-checked with local testing lists as a way of ensuring that less talented students are tested. Answer sheets are electronically scanned for suspicious erasures. The state department of education randomly audits test administrations. And test tampering is now a misdemeanor with a maximum penalty of a $1,000 fine and 90 days in jail.
But Mississippi is the exception among the 40 states that require standardized testing, says testing expert Susan Phillips of Michigan State University. Indeed, a 1989 survey by a watchdog group called Friends for Education revealed that only 7 states require test proctors; only 17 do random monitoring of test administrations, and only 6scan answer sheets for irregularities.
The high price tag for rigorous test security is a major reason for its absence in many states, say Phillips and other experts. Safeguards like tamperproof packaging, last-minute delivery of test booklets to schools and optical scanning of answer sheets add significantly to testing budgets. The most effective deterrents to test tampering–proctors, frequent in-person audits of test administration and the regular use of new tests–are still more expensive.
As a result, the administration of standardized tests is left largely in the hands of local teachers and school officials, who increasingly are being judged by the very tests that they are relied on to safeguard. “It’s the worst role conflict I can imagine,” says Dean Nafziger, head of the Far West Laboratory for Educational Research and Development and a former testing director in the San Diego schools.
Price of security. Test publishers bear much responsibility for the flaws in the nation’s standardized testing system. First, they simply don’t see the security of their products as their responsibility. “It’s up to the individual customer,” says Michael Kean, head of public and governmental affairs at CTB Macmillan. Testing companies typically ship test booklets that are unsealed, and answer keys are readily available. All three leading test makers publish “secure” forms of their tests. But the tests are more costly and can be obtained only through special contracts. As a result, says Kean, use of publishers’ secure tests is “rare.” Indeed, many publishers’ practices inflate scores artificially–making their products more attractive to potential customers.
One practice that is not illegal but is no less destructive to the integrity of testing is simply to keep old tests in circulation for years. In Maryland, for example, students were still taking a 1977 version of the California Achievement Test in 1989. Since the average score on each test is determined by a national sample of students who have taken the test without practice, students of teachers who have administered a particular test year after year are at a distinct advantage. Their teachers are able to familiarize them with test questions and answers, and the students are thus likely to score higher than the national norm.
The impact of the widespread use of old tests is tremendous, as John Jacob Cannell, a Montana physician, documented in a 1987 survey. Cannell found that the majority of the 32 states using national tests that year reported above-average scores at every grade level in every subject. He called it the “Lake Wobegon Effect,” after humorist Garrison Keillor’s fictional Minnesota township where “all the children are above average.” A 1989 study commissioned by the U.S. Department of Education corroborated Cannell’s findings. It concluded that “with respect to national averages, [school] districts and states are presenting inflated results and misleading the public.”
School systems say they use old tests because it is too expensive to buy new ones regularly. Publishers maintain that it is too expensive to write new tests more often than they do now–every eight years or so–and that they sell old tests to help educators gauge the performance of their schools over the years. But in the promotional material for its 1987 edition of the Stanford Achievement Test, the Psychological Corp. alludes to the public-relations value of using old standards. “There are many political and administrative reasons to continue using the 1982 norms,” the corporation tells potential customers.
In addition, a sister company of CTB Macmillan sells widely used test preparation materials that CTB’s own testing specialists say are glorified crib sheets. American School Publishers, like CTB a subsidiary of Macmillan/McGraw-Hill School Publishing, markets the study guide “Scoring High” for several of the major standardized tests. It is used by more than 1 in 5 teachers, according to a recent survey. In a 1990 document obtained by U.S. News, CTB officials write: “Despite the fact that CTB and American School Publishers are part of the same parent company, CTB opposes the use of ‘Scoring High’ materials …. [They] are examples of [test] coaching activities that are inappropriate.” The document also criticizes another study guide, “Learning Materials,” published by CTB itself, as “inappropriate” test preparation. Both guides, they write, include practice tests that resemble tests currently on the market and thus “can increase test scores without any real and lasting increase in the skills being measured.”
By rote. That seems to be exactly what is happening in school systems nationwide. In a 1991 study by the University of Colorado’s Shepard and three co-authors, scores dropped dramatically when students at 36 schools were retested on the same topics a few weeks after they took their school systems’ regular standardized tests. Questions were asked slightly differently the second time around. “Teachers are not teaching students skills and concepts,” Shepard said. “They are teaching specific examples by rote memorization.”
Publishers also make their tests more marketable by making them easier. According to Roger Farr, a co-author of the Psychological Corp.’s Metropolitan Achievement Test, “word recognition” items, which measure the lowest level of literacy, were removed from the reading-comprehension section of the test when it was revised in 1977. But when the test was revised again in 1986, those low-level items were reintroduced, Farr says, because in the interim many school systems, eager to keep their scores as high as possible, had dropped the MAT in favor of rival exams that contain the easier word-recognition questions. “They’ve forced us to not have anything that’s too difficult on the tests,” says Farr. “For a superintendent, the criterion is often, ‘Give me a test that makes me look good.”‘
In light of the abuses, it is perhaps not surprising that many standardized tests prove to be poor measures of what students actually learn. In 1989, 56 percent of Virginia’s eighth graders scored above the national average in math on the Iowa Tests of Basic Skills. But in 1990, on a federally funded test, only 15 percent of the state’s eighth graders showed a grasp of fractions, percentages and other math skills generally introduced by the seventh grade. Fully 67 percent of Indiana’s eighth graders outscored the national average on the California Achievement Test in 1988, but only 14 percent had mastered the seventh-grade skills on the federally funded exam in 1990.
Many educators believe that the tests themselves are contributing to the poor performance of the nation’s students, in part because test preparation consumes tremendous amounts of classroom time. A 1990 study by Mary Lee Smith of Arizona State University found that the nation’s elementary schools spend an average of 100 hours–three to four weeks of school–on standardized testing annually. In another study of 100 schools by the University of Colorado’s Shepard, 68 percent of teachers reported doing test prep activities “regularly, throughout the year.”
Critics also charge that test-based teaching drives down the level of instruction in the nation’s schools because the tests themselves focus on low-level skills. The basic California Achievement tests for grades 6 through 12, for example, consist of 55 multiple-choice questions on vocabulary, 55 on reading comprehension, 35 on spelling, 35 on capitalization and punctuation, 55 on usage and sentence structure, 50 on arithmetic, 55 on math concepts and applications, and 25 on “study skills.” Only reading comprehension and math concepts–a third of the test–tap into students’ advanced-thinking skills.
Sticking to basics. Advanced topics that are not tested are ignored in many classrooms, critics charge. Half the teachers in a 1990 study by Janie Hall and Paul Kleine of the University of Oklahoma said they had shifted their teaching strategy to match particular questions on standardized tests. And in her study, Smith found that teachers gave short shrift to science, social studies and writing in favor of reading, vocabulary lists, spelling, punctuation and arithmetic. There is also less student participation in classrooms where tests drive teaching, less debate and fewer opportunities for students to develop original solutions to problems. Says Arlington, Va., teacher Bracey: “There are teachers who would do things hands-on in their classroom but stick to workbooks with tests hanging over their h
Advocates of standardized testing say that kids must master basic skills before tackling more advanced work. But critics counter that teaching to the tests through endless workbook drills is counterproductive, that reading and writing are best taught by having kids read and write a lot, rather than by having them practice punctuation ad nauseam. “Skill drill by itself is the worst possible teaching,” says Shepard. “It’s boring for both students and teachers, and students learn best when skills are taught in context.” Ironically, low-achieving students are hurt the most by test-driven teaching, testing critics charge, because they are consigned to skill drills for long periods.
The lesson of the corruption of standardized testing in recent years seems to be that the nation mistakenly has tried to ratchet up accountability in public education on the cheap. Multiple-choice tests of low-level skills may be relatively inexpensive to administer, particularly in the absence of tough test security. But as they are used today, many are of dubious educational value. This should serve as a warning to the advocates of national testing. If the nation is to build a new national examination system, as seems increasingly likely, it needs to invest the resources necessary to build tests with high standards and rigorous security. If it doesn’t, the testing debacle in America’s schools may only get worse.
OLD-FASHIONED CRIB SHEETS In 1990, Rosa Walker, a veteran teacher in North Carolina’s Winston-Salem school system, was suspended without pay for giving five English classes a list of 35 spelling words that later turned up on the California Achievement Test. In the wake of the incident, Winston-Salem Superintendent Larry Coble received calls from a number of other North Carolina school officials detailing similar test-security violations in their districts. An investigation of the Walker case turned up state records of other teachers and administrators throughout North Carolina who had been punished for giving students advance copies of tests, helping students during tests and similar transgressions.
THE VALUE OF AN ERASER Public School 5 on Staten Island, N.Y., had the highest reading scores in the borough for five straight years, an accomplishment that earned the school lavish praise. But an official investigation revealed that between 1986 and 1990 the school’s principal, Murray Brenner, had systematically changed his students’ incorrect answers. The investigation was launched after the Staten Island Advance, a local newspaper, reported that a student with learning disabilities had scored in the 90th percentile on the reading test in 1987. The student’s mother had twice raised questions about the inflated test score with her local school board, but her complaints had been ignored.
THE BEST AND BRIGHTEST Under a 1989 Oklahoma law, schools are put on probation by the state if their students’ average scores on the Iowa Tests of Basic Skills rank in the bottom fourth of schools statewide. In 1991, 121 Oklahoma schools worked their way off the state’s probationary list by raising their test scores. Forest Reece, a member of the state board of education, conducted a study of the 121 schools. He found that a number of the schools had tested only their brightest students, by abusing a provision in the 1989 testing law that exempts special-education students from state testing with their parents’ permission. In 20 of the schools surveyed by Reece, 2 of every 3 students had been exempted from standardized testing using this maneuver.
eads.”