their biology teacher had taken. Monk (1994) found similar effects in mathematics and physical sciences but not in the life sciences. Goldhaber and Brewer (2000) used data from the National Education Longitudinal Study of 1988 to conduct a multiple regression analysis of 6,000 high school seniors and 2,400 mathematics and science courses. They found a relationship between teachers holding a mathematics degree and student performance, but no relationship between teachers holding a science degree and student performance. These results may have been affected by the high percentages of high school science teachers who teach out of their field, that is, a teacher with a biology degree teaching chemistry or physics.
The optimal level of subject matter training for a teacher is unclear, and there is some evidence suggesting a threshold effect—a point after which further course work provides no additional measurable impact on student learning. For example, Monk (1994) found that after a teacher had taken five college mathematics courses or four physical science courses, additional courses were not associated with additional gains in student achievement. Findings from several studies suggest that the impact on students of having a teacher with a subject matter major might vary with the level of the grade taught; the achievement of middle and high school students appears to be affected more by the amount of subject matter preparation of their teachers than that of elementary students (Rowan, Correnti, and Miller, 2002; Hawkins, Stancavage, and Dossey, 1998). Interpretation of these results, however, must consider the generally poor alignment of the content of college courses taken by teachers with the curriculum that they are expected to teach as well as by the ceiling effects in the achievement measures used in the studies. If college courses were aligned with school curriculum and if higher quality measures of student achievement were available, one might find that there are no threshold effects or that they must be higher than suggested by these studies.
There is also evidence from case studies of science teachers that teacher knowledge influences instructional practice and, in particular, that classroom discourse—an integral component of science learning environments—is sensitive to teachers’ knowledge of science (Carlsen 1988, 1992; Hashweh, 1987; Sanders, Borko, and Lockard, 1993). For example, Sanders and colleagues (1993) conducted an in-depth analysis of three secondary science teachers teaching inside and outside their areas of certification. They reported that when teachers had limited knowledge of the content, they often struggled to sustain discussions with students and found themselves trying to field student questions that they could not address.
Even more than quantity of knowledge, the qualities of teachers’ understanding of science are also important. If teachers are to help students achieve science proficiency, they too need to achieve proficiency across the four strands. Yet undergraduate science curricula, like those in K-12 science, tend to be biased toward conceptual and factual knowledge and reflect impover-