Our field is concerned with determining (scientific) truths about the atmospheric system. Observations provide the only means of verifying theories and models of the atmosphere, but like the prisoners in Plato’s cave, we are often limited to observational “shadows" of the variables we’re actually interested in. For example, we may wish to have single-particle properties (habit, mass, etc.) of all hydrometeors within a cloud, but we only get radar reflectivity (if we’re lucky). On the other hand, we can create models that are arbitrarily detailed, limited only by computational resources and our imagination, but these models may struggle to represent observed behavior due to approximations, errors and other uncertainties. How can we best reconcile the gaps in our model knowledge using observations that may be noisy, biased, or nonlinearly related to the quantities we are truly interested in? Bayes’ theorem provides the formalism to both reconcile these disparate sources of information and probabilistically account for attendant uncertainties. I will present some example of my research using observations to evaluate, inform, and constrain simulations of cloud microphysical processes, including analysis of polarimetric radar signatures of deep convection, constraint of ice microphysics within a mid-latitude squall line, and development of a novel physical-statistical rain microphysics parameterization. I will also provide a brief background on Bayesian inference, as well as Markov Chain Monte Carlo, a popular class of methods for solving Bayes theorem.