15.4 R Code 409 When there are repeated measures within individuals, either the repeated measures within individuals or the means across individuals might be modeled as normal distributions. Again, this hierarchical model can be easily programmed in BUGS. Whenever applying a normal likelihood function, it is important to check that the data are reasonably normally distributed (i.e., at least roughly unimodal and symmetric). Otherwise the parameter values may be mis- leading. If the data are severely non-normal, they might be transformed to normal, or else a non-normal likelihood function could be used instead. Data that have prominent outliers might be better modeled with a t dis- tribution than a normal distribution. This idea was introduced in this chapter, but examples of applying it in BUGS wait until subsequent chapters. 15.4 R CODE 15.4.1 Estimating the Mean and Precision of a Normal