Studies of failures in organizational information gathering, learning, and decision making highlight psychological and institutional causes. However, organizations are also political coalitions that face internal contestation over strategies, policies, and goals. The decision of whether to collect information impacts both the goals that organization members try to meet and the organization’s capacity to meet them. This paper develops a formal model that introduces political conflict into a theory of organizational learning. The model has a key insight: organizations characterized by political conflict will forego learning when leaders do not need to learn to build consensus in support of change, and will learn when leaders are unable to build such a consensus without learning. As a result, political conflict leads organizations to implement changes without first learning and to frequently learn when no changes are forthcoming. These tendencies toward ignorant decision making and educated inertia will be more pronounced when environmental variability is high, when learning about existing policies does not also teach about potential policy alternatives, and when organization members are risk averse. The model offers predictions about pathological learning behaviors that are consistent with considerable prior qualitative research. These patterns cannot be produced by experiential learning models in which political conflict is not present. The e-companion is available at https://doi.org/10.1287/orsc.2017.1164 .

Empirical evidence suggests that entrepreneurs make mistakes: too many enter markets and, once there, persist too long. Although scholars have largely settled on behavioral bias as the cause, we suggest that this consensus is premature. These mistakes may also arise from a process in which entrepreneurs continually learn about their prospects and make entry and exit decisions based on what they have learned. We develop a computational model of this process that connects pre- and post-entry learning and can be directed to analyze Bayesian-rational or biased entrepreneurs. The model suggests that, to outside observers, rational entrepreneurs may appear overconfident, seem to take too long to exit, and exhibit a positive correlation between entry cost and persistence in the market. When examining confidence biases, the model suggests that entrepreneurs whose biases cause them to perform the worst after entry will be most likely to enter, that pre-entry learning induces a positive correlation between distinct confidence biases among entrants, and that exit changes the prevalence of certain biases in the surviving population of entrants over time. Our study also speaks to recent work on pre-entry experience that documents the transfer of knowledge from parent to progeny firms, suggesting that, in addition to inheritance, differential performance may also be the result of heterogeneity in the length and quality of pre-entry learning during which an opportunity is assessed.