Why are some average indicators calculated only on a small sample?

To reduce the biases of career age and sheer size of publication sample for a research er/constellation it is sometimes valid to calculate some indicators based on the 5, 10 or 15 publications with the highest indicator values for that specific indicator. For example, the indicator Citation average (C/P) can be calculated on the most highly cited publications for each person. In the bibliometric analysis toolkit the result show how highly cited your 5, 10 or 15 top publications are compared to top publications by other researchers at Karolinska Institutet/SLL. The purple bars show in which decentile you can be found based on the average number citations to your top publications. You can also see the average value for each decentile.