Consider a decision maker wishing to maximize a given objective function. The distribution of the covariates of the environment and the causal relationship between them is referred to as a context. The decision maker is uncertain about their own context but has access to two databases. First, aggregate statistics about their context. Second, a description of the aggregate statistics, action choices and observed outcomes in a possibly different context.  The decision maker uses the first database to draw an inference about their own context and to judge the usefulness of the second database. This inference process guides the decision maker's optimal action. In a simple setup, I study the impact of varying depths of uncertainty on the inference and consequent decisions. As an application, I discuss the argument of ‘structured speculation' (Banerjee, et al.(2017)).