Publication rate as a function of laboratory size in three biomedical research institutions

Abstract

In three biomedical research institutions, there is no indication of a single laboratory size at which the number of publications per scientist is maximal or minimal. In a scattergram of the number of publications of a laboratory against laboratory size, the horizontal coordinate measures the number of scientists in a laboratory, the vertical axis measures the number of publications from the laboratory (counting each publication once regardless of the number of authors), and each laboratory is represented by one point. Scattergrams for the Rockefeller University (RU), New York, the National Institute for Medical Research (NIMR), London, and the National Cancer Institute (NCI), Bethesda, are each described well by a straight line through the origin. The slopes of the lines for the three institutions are not significantly different. In these laboratories, ranging in size from 1 to 46 scientists, one additional scientist increases the expected annual number of publications of a laboratory by approximately 1.1, regardless of the size of the laboratory. Although the three institutions have significantly different mean laboratory sizes, the frequency distribution of laboratory size in each institution is described well by a 0-truncated negative binomial distribution, as predicted by a simple model of laboratory population dynamics.