[Methodological aspects of the preparation of measurements in animal experiment studies of metabolism].

Methodical aspects are discussed concerning an analysis of time series which result from experimental investigations of metabolic parameters in rats. Especially there are 2 questions. Should the measured temporal variations of the metabolic parameters be recorded model-free by empirical regression or by functional expression? Before treatment, a time-period was designed in which the rats could become a stable group. Should be chosen a steady description of the temporal parameter changes for the whole time-period of the experiment or should they be described separately for the periods of acclimating and of treatment? Especially due to the discrete time points of measurements, the 1st approach also allows the description of possible jumps in the parameter changing at the begin of the treatment. As consequence of the 2nd mode of description always a jump results. Further parametric and non-parametric statistical procedures are used for analysis of the time series and for comparisons of the time courses. In detail following statistical problems are discussed: Homogeneity of dispersion along a single-measurements course or along a mean value course, equality resp. parallelism of 2 courses, and local or global level differences between 2 courses. Finally it is pointed out that besides the characterization of the mean temporal variations of a parameter also the description of correlation between 2 parameters by empirical regression curve may give information on the mechanisms which give rise to the observed phenomena.