Introduction

All three major international assessments in science were given during 2003, an event that occurs only once every 12 years and offers a rare opportunity to compare performance across different grade and age levels. On the grade 4 Trends in International Mathematics and Science Study (TIMSS-4), U.S. students ranked in the top quarter of all participating countries (6th of 25). On the grade 8 assessment (TIMSS-8), the U.S. ranking slightly improved to reach the top fifth of all participants (9th of 45). On the Program for International Student Assessment (PISA), which is administered to 15-year-olds, the U.S. rank against all participating countries falls precipitously below the international average (22nd of 40). On the face of it, these results suggest that U.S. science rankings exhibit a sharp falloff beginning at secondary school and that U.S. efforts to strengthen international science performance should focus on improving secondary school science.

However, the rankings of U.S. international science performance against all the countries participating on each of the three international assessments have two important limitations. First, U.S. performance is compared against the performance of countries that represent a mix of both industrialized and developing economies. A better gauge of U.S. students’ science success would be to compare U.S. students’ science performance to only the industrialized countries that represent U.S. economic competitors. Second, the mix of industrialized nations changes across the three international assessments. Thus, the U.S. science rankings may reflect the changing mix of countries participating in each assessment and not reflect real changes in U.S. relative performance.

The National Science Foundation discussion of the TIMSS 2003 science results ignored the comparison group issue when it focused on U.S. 2003 science score improvement compared with the scores of all countries that participated in the prior 1995 TIMSS assessments: U.S. “fourth grade students remained fifth among 15 countries that participated in both 1995 and 2003” and “Eighth-grade students…raised their overall standing among the 21 countries that participated in TIMSS in both 1995 and 2003.” However, these U.S. comparisons are based on two different country sets and include different mixes of industrialized and nonindustrialized countries.

The current study corrects for weaknesses in measuring U.S. science performance by recomputing U.S. rankings against only the common set of 11 other industrialized countries participating in all three 2003 science exams. This approach of employing a common comparison group of industrial countries was previously used to assess U.S. rankings on the international mathematics assessments. It produced significantly different U.S. results compared with the full rankings against all countries participating on each assessment.

Along with examining U.S. international science performance, this study examines several country background variables that research suggests may be important in explaining students’ science outcomes on the international assessments. One such variable is students’ mathematics performance. A recent study of U.S. college students found that students’ secondary mathematics course taking was associated with science performance in college. The current study provides independent information on which to examine the relationship between mathematics and science performance by comparing a country’s international mathematics performance with its international science performance at the elementary and secondary levels.

Along with a country’s mathematics performance, three other sets of background variables are examined: science curriculum exposure, science preparation of teachers of science, and student characteristics. Because PISA rotates the emphasis of its assessments among reading, mathematics, and science, the 2003 PISA assessment collected only a limited set of science background variables. Hence, this report relies primarily on the TIMSS background data.

A cautionary note is in order when interpreting the correlation between a country’s science performance and its background characteristics. Large-scale international assessments cover widely different international systems and offer a natural laboratory in which to identify characteristics of science systems associated with performance differences. However, the correlations are only suggestive of an association; they are not, by themselves, evidence of causation. The bivariate correlations do not simultaneously control for the many unobserved country differences that could influence the bivariate correlations. They require further studies of the effect of these characteristics within a particular country to demonstrate their applicability.