Relational program logics are extensions of program logics for reasoning about relations between programs. Recently, they have been applied for reasoning about quantitative differences in programs, such as differences in computational costs. We extend a categorical framework for relational program logics reasoning quantitative differences to compare programs with ‘different kinds’ of computational effects. Using this framework, we construct a relational program logic that measures how much a probabilistic program deviates from a nondeterministic program.