Statistical theory aids inference in scientometrics

Conclusions

Laboratory size may affect the number of scientific publications per capita. However, no analysis I have seen of my data or of other data has demonstrated the existence of such an effect convincingly.

An analysis⁵ of patentable research among 500 leading industrial firms concluded that the number of patents filed per year divided by annual sales is independent of the annual sales, i.e., on the average, small, medium and large industrial firms are all equally inventive. If patents are viewed as a measure of research productivity in the commercial sector, and if sales are viewed as a measure of size or research effort (a company's research budget would be a better measure), then this finding that the ratio of patents to total sales is independent of total sales exactly parallels my finding that publications per capita are independent of laboratory size. The parallelism of these findings suggests the possibility that, in general, the productivity per capita of a research effort is independent of the total size of the research effort, whether the setting is academic or commercial.

Whether or not this empirical generalization is ultimately confirmed, the methodological message of this note seems to me less arguable.