We present the results of a national study on the teaching and learning of astronomy as taught in general education, non-science-major, introductory astronomy courses. Nearly 4000 students enrolled in 69 sections of courses taught by 36 different instructors at 31 institutions completed (pre- and post-instruction) the Light and Spectroscopy Concept Inventory (LSCI) from Fall 2006 to Fall 2007. The classes varied in size and were from all types of institutions, including 2- and [math] colleges and universities. Normalized gain scores for each class were calculated. Pre-instruction LSCI scores were clustered around [math], independent of class size and institution type, and normalized gain scores varied from about [math] to 0.50. To estimate the fraction of classroom time spent on learner-centered, active-engagement instruction we developed and administered an Interactivity Assessment Instrument (IAI). Our results suggest that the differences in gains were due to instruction in the classroom, not the type of class or institution. We also found that higher interactivity classes had the highest gains, confirming that interactive learning strategies are capable of increasing student conceptual understanding. However, the wide range of gain scores seen for both lower and higher interactivity classes suggests that the use of interactive learning strategies is not sufficient by itself to achieve high student gain.