Combining Incomplete Information and Learning in Microeconometric Models