科学の生産性とその階層化過程 : 「マタイ効果」の基礎理論

Distribution of Scientific Productivity and the Processes of Stratification

山崎 博敏

大学論集

Daigaku ronshu: Research in higher education

[PISSN] 0302-0142

[NCID] AN00136225

This paper reports that the distribution of productivity of 124 university chemists in Japan shows the best fittness to a negative binomial distribution, and then considers the reasons for and the sociological implications of the results.

Since the inverse square law of A. J. Lotka (1926), several mathematical models on the distribution of productivity have been proposed by Williams (1944), Simon (1955), Shockley (1957), Price (1963, 1976), Allison (1976) and Rao (1980) et al. The characteristics of these models were examined in comparative perspective. The negative binominal distribution showed the best fittness to our data among these models. This result proposes the hypotheses — reinforcement in the process of research activity and heterogeneity among each scientist. It is difficult, however, to judge which hypothesis is more appropriate, mainly because both models correlate with each other. Heterogeneity of the ability and the socialization process of each scientist causes an inequality of productivity among them. And this inequality reallocates the productive scientists to research oriented-universities and strengthens their motivation to the further research on the one hand, and weakens the motivation of less productive scientists. This process increases the differences of productivity among scientists.

However these difficulties were solved by the two findings in that

(i) the distribution of the productivity in the subsample of full professors in graduate schools with doctoral programs (N = 39) shows a good fittness to the negative binomial distribution, and

(ii) the coefficient of variation of productivity increases as the age of scientists increases.

From these two facts, we can accept the reinforcement hypothesis at least. This implies that the more the scientist publishes, the more the probability to publish later increases, while the less he publishes, the more the probability decreases. This hypothesis has a significant meaning for the theory of sociology of science, because "reinforcement model" describe the Merton's Matthew Effect on the mathematical level. Our result also confirms the empirical validity and international universality of the Mertonian theory of sociology of science.

教育 [ 370 ]

日本語

広島大学大学教育研究センター

[ISSN] 0302-0142

[NCID] AN00136225