Comparing the scatter of citing and cited literature

Using the subjects desalination and educational psychology, the scatter of periodical articles over periodical titles was compared at two levels, the second level being a random sample of periodical articles cited by the first level. Several measures were used to compare the extent of scatter at the two levels. Some methods commonly used in bibliometrics produced conflicting evidence on whether the citing literature (first-level) or the cited (second-level) was more scattered. A computer-intensive sampling procedure, known as the Bootstrap method, was then used to estimate the scatter of the total cited population from the scatter of the empirical sample. Cumulative distributions were prepared to show what percentage of periodicals accounted for various percentages of articles at each level of scatter. Only at the 90th percentile of articles did the percentage of periodical titles in the cited literature significantly exceed that of the citing literature. At the tail-end of the Bradford-type distribution, the cited literature appears to be more scattered than the literature citing it.