Quality and Analysis of Small Data Sets – A Statistical Point of View

We are often dealing with veterinary studies with very limited number of experimental units e.g. few animals. This might be due to complicated experiments, high cost for each animal, time demanding experiments, etc.

All other things being equal, when the data set is small, it is difficult to demonstrate significant treatment effects. It is therefore of interest to improve the quality of data in order to demonstrate significant effects. On the other hand the choice of experimental design is also very important. Furthermore, the choice of analytical methods might also have some influence on whether or not an effect can be seen.

Examples of small data sets will be given, illustrating the effect of high data quality. Different experimental designs will be discussed and the influence will be illustrated. Choosing exact methods for the statistical analysis, evaluating outliers or strange observations by influential statistics may improve the result as well.