This Page Covers:
First, one reason standardized tests should not be used is they are an inaccurate assessment of student performance. Most experts agree that in order to write a successful, accurate test, standardized or not, a test writer must consider aspects of the test: objectivity, validity and reliability (Perrone 20) .An objective test is one that is not affected by the bias of a particular test grader (Strenio 62). Reliability refers to how close test scores from one area or another reflect tests taken at different times, or how closely results are to tests of the same type. Validity is defined as "the degree to which a test measures what it is supposed to measure. . ." (Perrone 21).
Reliability is usually easily calculated by standardized test
companies. Most popular standardized tests have reliability
coefficients of between .87 and .93. Meaning the test is reliable
87 to 93 percent of the time. Although this is high, a very poor
test that can have a high reliability rating (Perrone 20).
The strong point of most standardized tests is often advertised
as their high level of objectivity. They say that if the tests
are graded automatically, by a computer, they cannot be bias;
i.e. The test grader has no power to grade in a bias way, so the
test must be valid. What this analysis overlooks is the test
writer's ability to be subjective. The test writer can create
questions that are bias. Therefore, test results would be
inaccurate for those discriminated against.
Validity is often is not given as much attention as the other two aspects by standardized test writers. For most standardized tests, validity is determined by having an "expert" look at the test and determine whether to content of the test is valid in his opinion. Therefore, test validity is determined by opinion, and not by scientific means. (Perrone 21). A test writing company can write questions that bind the test-taker to a "narrow subset of rote memorization" (Strenio 63) that can cause a student who understands a concept that a question is supposed to test to miss the question. This would make the test invalid, and therefore inaccurate (Strenio 63). Because of standardized test's ability to be subjective and their lack of true validity, test scores should not be taken seriously.
Another reason standardized tests scores can be inaccurate is teachers and school districts can alter their curriculum to match the tests (Foster). Instead of teaching a skill and using the test to test that skill, teachers will teach about a specific example they know will be on the test (Wagner). A teacher may not even do this with the intention of being devious. For example, Lorrie Shepard, a testing expert at the University of Colorado says that if a teacher knew that the part of the tests that tested reading ability talked about kangaroos, the teacher would be foolish not to teach about kangaroos in their class. (Wagner). This means that the student would be tested not on how well they read, but on how much they learned about kangaroos. This kind of test prepping would skew test results, even though it may not even be intentional or illegal (Wagner). Also, security measures cannot be used to prevent this. A teacher cannot be forced to not teach about kangaroos simply because a reading selection about kangaroos is on the upcoming test. This kind of test prepping is therefore virtually unstoppable. If test results are skewed, they are inaccurate and therefore useless.
The accuracy level of standardized tests is also significantly dropped by widespread cheating. In a 1990 survey, 35 percent of teachers in North Carolina admitted the presence of cheating in their schools (Wagner). A national survey in 1990 revealed that almost 10 percent of all the teachers who took the survey said that they received pressure to cheat from school administration (Wagner). In 1989 12 out of 17 elementary schools in Trenton, New Jersey passed the California Achievement Test for third grade. After evidence of cheating came about, state officials monitored the tests, only three out of 17 schools met the state's standard. (Wagner). In 1988, the California State Department of education reported that 40 California elementary schools were caught cheating (Foster). The list goes on and on. If this kind of cheating is so widespread that between 10-30% of teachers in various areas report cheating, the results on standardized tests will obviously be very inaccurate. When situations like these occur, the tests become a tool to not see which students know more; they become a tool to see what administrations lay more pressure on teachers to cheat.
The solution seems easy enough. If security is so low on tests, why not simply raise the level of security? But, most school districts cannot afford the high cost of security. First, the cost of simply giving a standardized test is extremely high. Most of the time it costs a school around eight dollars per student to proctor an exam such as the Iowa Test of Basic Skills (Olds). Although it does not seem so, this cost is enormously high If school districts are only budgeted an average of around $5,200 per student (Speer). Eight dollars may seem like a small chunk of that amount, but when one considers the costs of teachers, textbooks, computers and school buildings, the value of that $5000 skyrockets. That high eight dollar cost only entails the cost to give the test.. This would not include the cost of someone from outside the school or perhaps ever the state to come in and make sure that cheating was not occurring. The cost of test proctors, tamper proof test envelops, and delivery of test booklets at the last minute, so that teachers are not tempted to take a peek at the tests and quickly alter their lesson plans, is out of the range that schools can afford to pay (Wagner). Because it would be too expensive for schools to heighten tests security, cheating on standardized tests will continue indefinitely. This cheating will continue to lead to inaccurate, misleading test results.