The use of concept inventories to investigate students' learning gains is common in physics education research. However, comparatively little research has compared students' learning gains on concept inventories with other more traditional assessments in the classroom. We present a study comparing second semester calculus-based physics students' performance on traditional classroom assessments including exams and homework with learning gains on SEMCO (Survey of Electricity, Magnetism, Circuits and Optics), which was previously created by combining questions on other conceptual surveys such as CSEM and DIRECT. We report on students' performance on specific items on SEMCO and corresponding traditional classroom assessments that are based on the same topic. Our results indicate that while the overall performance on SEMCO might correlate with aggregate performance on class exams, the performance on clusters of SEMCO items that assess conceptual understanding in various topical areas does not correlate as strongly with performance on corresponding traditional exams. These results raise some potentially interesting issues on the validity and usefulness of traditional classroom assessments and conceptual assessments that are often used to measure student learning in introductory physics.