Background

Standardized test scores are inadequate measures of our children's learning, and one reason is that they reflect a student's performance on a limited kind of task, on a particular day, in a stressful time-limited situation. Using standardized test scores to measure our students' learning is like using a still photograph rather than a videotape. But since standardized test scores are commonly used to measure and compare students' "achievement," it is important to examine such scores in order to address the widespread myth that our students are not doing as well as they used to. We need to look deeper than the superficial data usually reported in the media.

SAT scores revisited

Overall, the average SAT scores in 1992 were about 5% lower than in 1960. However:

• About half the decline between 1960 and 1990 reflects changes in who is taking the test-and who isn't. The test is now taken by proportionately fewer students from the top 20% of their graduating class and proportionately more students from the lower 60%. Furthermore, the test is now taken by groups that have traditionally scored lower: minority students, students of lower socioeconomic status, and students for whom English is not their native language. The 1995 SAT scores were higher than they have been in years, but the significance of this is uncertain because of a change in scoring procedures.
  
  
• The SAT test today is more difficult than it was in 1975, according to the Educational Testing Service, which develops the test items (Berliner, 1992, citing the 1991 draft of the Sandia report).
  

Comparing SAT scores "then" and "now" is not a good way to measure the successes of our schools-for these and other reasons, including the fact that SAT scores say nothing about how well our educational system is succeeding with students who have not chosen to take the test (e.g., Robinson & Brandon, 1992).

Other standardized tests

Other kinds of standardized tests include so-called intelligence tests and achievement tests, both of which are commonly norm-referenced. That is, the tests are designed so that the scores of a representative sampling of students taking the test will be distributed on a bell curve. These tests are renormed about every seven years, to ensure that once again half the students taking the test will be, by definition, below average. Thus any real gains in students' learning over time will in effect be periodically wiped out. It's like resetting the odometer to 0 again and again, or making the high jump higher and higher.

Taking this re-norming into account, "We can estimate that around eighty-five percent of today's public school students score higher on standardized tests of achievement than their average parent did" (Berliner, 1992). In general, today's generation is doing better on the California Achievement Test (CAT), the Iowa Test of Basic Skills (ITBS), the Stanford Achievement Test (SAT), the Metropolitan Achievement Test (MAT), the Comprehensive Tests of Basic Skills, and the College Board Achievement Tests. Students are also doing better on intelligence tests, which measure school achievement more than native ability (Berliner, 1992; Stedman, 1994).

International comparisons

International comparisons can be made on several factors, but they must be made cautiously because of the various factors involved. For example:

• In 1989, the U.S. had the 4th highest high school completion rate of 20 major countries (NCES, 1992). The U.S.'s college graduation rate is also about double that of most other countries. We spend much more than most nations on higher education, but we are about average among industrialized nations in what we spend on K-12 education (Berliner & Biddle, 1995). Furthermore, in real dollars the per-pupil expenditures were about the same in 1990 as in 1976, for regular education instruction (Sandia, 1993). Despite claims to the contrary, the amount of money spent on instruction itself has a major impact upon the quality of education that students receive and upon their educational attainment (Berliner, 1992; Berliner & Biddle, 1995; Bracey, 1994).
  
  
• Typically a much larger percentage of our students have taken the tests on which international comparisons have been made. For example, in the first International Assessment of Educational Progress (IAEP), 75% of U.S. students were compared with the top 9% in Germany, the top 13% in the Netherlands, and the top 45% in Sweden (Rotberg, 1990, as reported in Berliner, 1992).
  
  
• Rank orderings of countries can be very misleading. For example, in the first IAEP study of reading, U.S. 9-year-olds finished second in the world among 31 nations, while our 14-year-olds finished eighth. However, the actual scores of the 14-year-olds were almost as close to first place as those of the 9-year-olds (Bracey, 1994).

International comparisons of test scores have not typically presented U.S. students in a favorable light, but national differences need to be reinterpreted in light of factors such as these. Furthermore, when the results of our higher educational system are compared with those of other countries, the U.S. looks very good indeed (Berliner, 1992; Sandia, 1993).

NAEP: Criterion-referenced tests

Standardized tests measure our students against each other, while criterion-referenced tests measure our students against the testmakers' concept of what our students "should" know and be able to do. Over the past three decades, National Assessment of Educational Progress (NAEP) scores have generally been level, or nearly so, in reading, writing, and math. However, high school students have fallen below 1970 levels in civics. Our 12-year-olds' performance in science fell considerably from 1970 to the early 1980s, then rose somewhat until 1990. A breakdown and analysis of the data shows a rather small percentage of our students achieving at the higher levels of proficiency on the NAEP assessments (Stedman, 1994).

The good news: How good is it?

So far, traditional instruction has been the norm for most of the students included in the NAEP data, the vast majority of whom are not achieving high proficiency in the school subjects tested. If indeed the NAEP adequately assesses what students ought to be learning, then schools and teachers may need to adopt more effective ways of fostering students' learning. On the other hand, the good news is that, using the scores on large-scale tests as the measure, the quality of education in our schools has not declined in the last two decades; indeed, scores on a variety of these tests seem generally to have increased (Rand, 1994).

REFERENCES

Berliner, D. C. (1992). Educational reform in an era of disinformation. Paper presented at the meeting of the American Association of Colleges for Teacher Education, San Antonio, Texas, February. ERIC: ED 348 710.

Berliner, D. C., & Biddle, B. J. (1995). The manufactured crisis: Myths, fraud, and the attack on America's public schools. Reading, MA: Addison-Wesley.

Bracey, G. W. (1994). The fourth Bracey report on the condition of public education. Phi Delta Kappan, 76, 115-127. Earlier Bracey reports were printed in the October 1993, 1992, and 1991 issues of the Phi Delta Kappan.

National Center for Education Statistics. (1992). Condition of education 1992. Washington, DC: U.S. Department of Education.

Rand Institute on Education and Training. (1994). Student achievement and the changing American family. Santa Monica, CA: Rand. (For more information, call (310) 451-7002).

Rasell, M. E., & Mishel, L. (1990). Shortchanging education: How U.S. spending on grades K-12 lags behind other industrialized nations. Washington, DC: Economic Policy Institute.

Robinson, G., & Brandon, D. (1992). Perceptions about American education: Are they based on facts? Arlington, VA: Educational Research Service.

Rotberg, I. (1990). I never promised you first place. Phi Delta Kappan, 72, 296-303.

Sandia National Laboratories. (1993). Perspectives on education in America [known as the Sandia Report]. Journal of Educational Research, 86, 259-310. A 1991 draft of this report has also been widely circulated and cited.

Stedman, L. C. (1994). The Sandia Report and U.S. achievement: An assessment. Journal of Educational Research, 87, 133-146.

Prepared for Michigan for Public Education and for the Michigan English Language Arts Framework project and © 1995 by Constance Weaver. In C. Weaver, L. Gillmeister-Krause, & G. Vento-Zogby, Creating Support for Effective Literacy Education (Heinemann, 1996). May be copied.