It is common in online markets for agents to learn from others' actions. Such observational learning can lead to herding or information cascades in which agents eventually ``follow the crowd''. Models for such cascades have been well studied for Bayes-rational agents faced with deciding between two possible actions - one ``good'' action and one ``bad'' action. In this paper, we consider the case when these agents instead have more than two actions, where again only one of these is good. We show that sequential observational learning in such settings has substantially different properties compared to the binary action case and further show than increasing the number of ``bad'' choices from 1 to 2, can improve the agents' learning.