Small decisions about nothing can have a large effect on DNA transfer modelling for activity level evaluation

Evaluations given activity level propositions require the expert to make a number of choices that relate to how the complexity of the world is simplified and modelled. Some of these choices relate to circumstances about the case, some to the data used and some to manner of modelling the data. In the right circumstances, the evaluation can be sensitive to any one of these categories of choice. In this paper an example is provided that demonstrates an evaluation sensitivity to the choice of how DNA transfer is modelled. Specifically, whether observations of no-DNA in a DNA transfer experiment should be considered as DNA transfer having not occurred, or having occurred at a level below the detection capability of the laboratory. An evaluation scenario is set up that demonstrates how this choice can lead to the difference between the evaluation slightly supporting one proposition to strongly supporting the other. A solution to this modelling choice is provided that shows how DNA transfer can be modelled as mixture of models, with non-DNA observations being simultaneously considered as no DNA transfer, or DNA transfer below detection capability. This is achieved using hierarchical Bayesian modelling, however could also be applied using other tools.