According to the standard definition in Sect. 11.4.2, two observational units \(u, u^{\prime}\) belong to the same experimental unit if the treatment assignment probabilities given the baseline configuration satisfy \(P\left(T_{u}=T_{u^{\prime}}ight)=1\). Section 5.1 .3 makes the argument that each louse is one observational unit, and that each lineage is one experimental unit. But the author subsequently pivots to aviary as the experimental unit, hedging his bets by stating that 'both seem to be relevant'. Discuss the arguments pro and con of louse-lineage versus louse-aviary versus lineage-aviary as the observational-experimental units. In connection with the models in Sect. 5.2, what are the substantive implications of one choice versus another?