Statistical model selection methods have been developed and applied to compare the performance of two deep convection parameterizations in the U.S. Dept. of Energy's ACME climate model. The comparison is based on information criterion scores that summarize the goodness-of-fit to observations of tropical precipitation and penalize against overfitting. The scores are derived using an ensemble of 400 Latin hypercube simulations of the global ACME model (1 degree, 30 levels) that perturb parameters in the Zhang-Mcfarlane and Chikira deep convection schemes. The ensemble is used to train statistical models, identify influential model parameters, determine maximum likelihood estimates, and compute information criterion scores. The scores can be used to provide guidance on the selection of a deep convection parameterization for a climate model. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344, and is supported by the DOE Office of Science through the Scientific Discovery Through Advanced Computing (SciDAC).