Modelling continuous data

Gilbert Welch and colleagues have written an interesting essay contrasting continuous and categorical approaches to modelling the relation between exposure and health effects.[1][1] They suggest that assuming a continuous relation between exposure and outcome may produce misleading results and state