Much work on probabilistic evidential reasoning for crime investigation employs probabilities that express subjective expert beliefs. This use of subjective probabilities is inevitable for several reasons, including lack of data, non-specificity of phenomena and fuzziness of concepts in this domain. Numerous representation formalisms and corresponding inference mechanisms have been developed to capture and reason with the intrinsic vagueness in subjective probabilities. In the literature, these schemes are largely presented as though they are diametrically opposed to one another. This paper critically examines what aspects of vagueness are captured by these different approaches. It demonstrates that they are concerned with different aspects of vagueness. This leads to a proposal of a method to combine the different approaches.