Crossing the hurdle: the determinants of individual scientific performance

An original dataset referring to a medium-sized Italian university is implemented for analyzing the determinants of scientific research production at individual level. Three different indicators, based on the number of publications and/or citations, are considered. Their distributions are highly skewed, displaying an excess of zero-valued observations, thus zero-inflated and hurdle regression models are introduced. Among them, the Hurdle Negative Binomial model exhibits a good fitting and appears to be reasonably coherent with the underlying generating data process. Indeed, the performance of active researchers is described by the count component of the model, while the odds to be in a non-active status is modelled by the zero component. Individual characteristics, teaching and administrative activities, as well as the features of the department the researcher belongs to, are considered as explanatory variables. The analysis of the results highlights that scientific productivity is lower for oldest active researchers, and that there is a significant effect of academic position on research production. Evidence of clear-cut and significant substitution or complementarity effect between teaching and research activities is not found. Indeed, a major teaching load does not apparently affect the odds to be a non-active researcher, while it has mixed and very weak effects on publication performance of active researchers. A negative relationship among productivity and administrative tasks is highlighted. Finally, the analysis points out the effects of department composition on both the odds to be non-active and the value of the production indicators of active researchers.