*Abstract: One approach to likelihood inference in the presence of a nuisance parameter is to use an integrated likelihood, in which the nuisance parameter is eliminated from the likelihood by integration with respect to a given weight function. Integrated likelihoods have the advantage that they are always available and, unlike the profile likelihood, they are based on averaging rather than maximization, an approach which has been shown to be more reliable in many models of interest. On the other hand, integrated likelihoods have the drawback that the weight function for the nuisance parameter must be chosen by the analyst. However, because the weight function may be chosen differently to achieve different goals, this perceived drawback may also be viewed as strength. In this talk, I will discuss general properties of inferences based on an integrated likelihood along with some different approaches to choosing the weight function.*

*Host: Todd Kuffner*

*Tea @ 3:45 in room 200.*