Three alternative approaches to assessment of exam responses were applied in an undergraduate biochemistry course. First, phenomenography was used to categorize written exam responses into an inclusive hierarchy. Second, responses to the same question were similarly categorized according to the Structure of Observed Learning Outcome (SOLO) taxonomy. Third, all questions on the midterm and final exams were ranked according to their level of Bloom's taxonomy. Student responses across phenomenographic and SOLO taxonomy hierarchies, and the accuracy of responses at each level of Bloom's taxonomy, were analyzed in relation to student learning approach as measured by scores on the Revised Student Process Questionnaire 2-Factor. Deep learning approach score predicted successful responses to questions ranked at the application level of Bloom's taxonomy and at the top hierarchical category of phenomenographic analysis. Results suggest that alternative assessment strategies may be used as tools to evaluate learning at increasing levels of cognitive complexity as related to a deep learning approach.