Bert Baumgaernter (Idaho) gives a talk at the MCMP conference "Reduction and Emergence in the Sciences" (14-16 November, 2013) titled "Agent-based models as mixed-level: lessons from E.coli". Abstract: We argue that agent-based models (ABMs) are better conceived of as multi- or intra-level models. Consequently, they are neither reductionistic nor emergent in the explanations they provide. We argue for this by first contrasting them with analytical models, which tend to focus on macro-level entities or properties. We then use an example of an ABM of group-selection from the biological sciences which is thought to be purely individualistic. However, we argue that the model is misconceived. The macro-level properties required to make the model work are not straightforwardly given by the composition of the individuals in the group.