*In comparison to traditional instruction, under expert modeling instruction high school students average more than two standard deviations higher on a standard instrument for assessing conceptual understanding of physics.*

The effectiveness of modeling instruction in enhancing student learning of physics is being continuously evaluated with well-established standardized instruments. Chief among these instruments is the *Force Concept Inventory *(FCI)*. *The FCI assesses the effectiveness of mechanics courses in meeting a *minimal teaching performance standard: to teach students to reliably discriminate between the applicability of scientific concepts and naive alternatives in common physical situations. *Questions on the FCI were designed to be meaningful to students without formal training in mechanics.

The FCI has consistently shown that students bring into their physics courses a wide array of naive beliefs about the motion of physical objects that are incompatible with Newtonian theory. Figure 1 summarizes data from a nationwide sample of 7500 high school physics students involved in the Modeling Workshop Project. The average FCI pretest score is about 26%, slightly above the random guessing level of 20%, and well below the 60% score which, for empirical reasons, can be regarded as the threshold for understanding Newtonian mechanics.

Figure 1 shows that traditional high school instruction (lecture, demonstration, and standard laboratory activities) has little impact on student beliefs, with an average FCI posttest score of about 42%, still much below the Newtonian threshold[1] [2]. This failure of traditional instruction is largely *independent of the instructor’s knowledge, experience and teaching style*.

High school teachers in the Leadership Modeling Workshop Project (1995-1999) began a shift from traditional instruction to modeling instruction in their first four-week summer workshop. After their first year of teaching, posttest scores for students of these *novice modeler*s were about 10 percentage points higher, as shown in Fig. 1 for 3394 students of 66 teachers. Students of *expert modelers* do much better. For 11 teachers identified as expert modelers after two years in the Project, posttest scores of their 647 students averaged 69%. Thus student gains in understanding under *expert* modeling instruction are more than doubled (40 percentage points gained), compared to traditional instruction (16 percentage points gained) (Figure 1).

Subsequent data have confirmed all these results for a total of more than 20,000 students. Furthermore, student FCI gains for “ordinary” Arizona teachers, 80% of whom were *not* physics majors, are almost as high as those for leading teachers nationwide. Teachers who implement the Modeling Method most fully have the highest student posttest FCI mean scores.