In this chapter, we discussed a range of techniques that can be used to analyse certain types of forensic entomology data. By no means is this the only way to analyse these data, and one can easily imagine more complicated models using random effects of temporal correlation structures.

The experiment presented in this case study was carried out at only one temperature value. It needs to be repeated under different temperature conditions, and this will lead to more complicated GAMM models that may include temperature interactions.

Another question often asked by forensic entomologists is how many samples to take. Obviously, this question can be split in two questions: How many samples do we need for an experiment as described in Section 8.2, and how many samples should the police take at the crime scene. The first question may be answered using data and models presented in this chapter. It brings us in the world of power analysis (Zar 1999). The only problem is that conventional power analysis equations are based on linear regression, and not for models with heterogeneity, random effects, auto-correlation and smoothers. Bootstrapping may be an option here. Zuur et al. (2007) applied some simulation tools to see what happens with estimated patterns if 5%, 10% or 25% of the data are removed. A similar approach may be adopted here. As to the second question, how many samples should the police take, we cannot address this question without having access to such data and investigate the variation in such samples. This will be the next step.

0 0

Post a comment