Effect of lecture instruction on student performance on qualitative questions

The impact of lecture instruction on student conceptual understanding in physics has been the subject of research for several decades. Most studies have reported disappointingly small improvements in student performance on conceptual questions despite direct instruction on the relevant topics. These results have spurred a number of attempts to improve learning in physics courses through new curricula and instructional techniques. This paper contributes to the research base through a retrospective analysis of 20 randomly selected qualitative questions on topics in kinematics, dynamics, electrostatics, waves, and physical optics that have been given in introductory calculus-based physics at the University of Washington over a period of 15 years. In some classes, questions were administered after relevant lecture instruction had been completed; in others, it had yet to begin. Simple statistical tests indicate that the average performance of the “after lecture” classes was significantly better than that of the “before lecture” classes for 11 questions, significantly worse for two questions, and indistinguishable for the remaining seven. However, the classes had not been randomly assigned to be tested before or after lecture instruction. Multiple linear regression was therefore conducted with variables (such as class size) that could plausibly lead to systematic differences in performance and thus obscure (or artificially enhance) the effect of lecture instruction. The regression models support the results of the simple tests for all but four questions. In those cases, the effect of lecture instruction was reduced to a nonsignificant level, or increased to a significant, negative level when other variables were considered. Thus the results provide robust evidence that instruction in lecture can increase student ability to give correct answers to conceptual questions but does not necessarily do so; in some cases it can even lead to a decrease.

Figure 1

Scatter plots for representative questions (5, 6, 8, 10, 13, and 20 in Table 1). Each diamond represents a single class. Black diamonds represent all classes; gray and white diamonds represent classes questioned before lecture instruction and after lecture instruction, respectively.

Figure 2

Raw differences in average performance arranged in increasing order. The labels correspond to those in Tables 1 and 2. Bars shaded dark gray, medium gray, and light gray indicate effects significant at the 1%, 5%, and 10% levels, respectively.

Figure 3

Estimated coefficients for the parameter instruction in the regression models (${\beta }_{\mathrm{I}}$), arranged in the same order as in Fig. 2. Bars shaded dark gray, medium gray, and light gray indicate effects significant at the 1%, 5%, and 10% levels, respectively.