Contextual bandit algorithms seek to learn a personalized treatment assignment policy, balancing exploration against exploitation. Although a number of algorithms have been proposed, there is little guidance available for applied researchers to select among various approaches. Motivate