Reliability for Bradford's line in small size statistics

Abstract

Aplicability of Bradford's lines was investigated with small size statistics using probability theory. Samples were taken at random and nonrepeatedly from 7,726 citations in 805 journals, which were cited by scientists of Japan Atornic Energy Research Institute during 1976 to 1978. As the results, it has become clear that expected value of ratios of the leading journals in the population decreases as the number of samples decreases. Therefore, in the case of selecting journals based on the Bradford's line made with smaller statistics, selection error may be larger than with larger samples.