IS THERE LONGITUDINAL EVIDENCE
THAT SCHOLARS IMPROVE THE
PERFORMANCE OF THEIR UNIVERSITIES?
Tis that of priority of time in the cause before the effect.1
BETTER SCHOLARS LEAD BETTER UNIVERSITIES. The earlier evidence has shown this in a variety of settings. A correlation does not prove causation, but for causation we must first have a correlation. What we do not know from the patterns presented in chapters 2 and 3 is whether more cited leaders are actually more effective. It may be that scholar-leaders are being picked for reasons other than their academic past as researchers. Scholarship might just be a proxy for management ability or leadership skills. Alternatively, elite universities, like those in the U.S. Ivy League, might choose distinguished faculty as leaders for reasons of status. But even if they do, it still seems interesting to try to understand why. Maybe all universities would like highly cited leaders but cannot afford them; maybe they would not.
In this chapter I adopt a longitudinal research design. Its aim is to try to establish whether universities that are led by more cited leaders go on to perform better in the future. I calculate each individual's level of scholarship, then, a number of years later I measure the performance of his or her institution. This relies on time lags to help uncover whether better scholars may actually cause research universities to improve. The issue of what is cause and what is effect is passionately debated in the social sciences. Indeed, some appear to take the extreme view that causal relationships can never be shown, and that it is almost a crime to try. Others, particularly economists, have thought a great deal in a practical way about causal associations—recently preferring to use natural or field experiments.
To truly prove causality would require leaders to be randomly assigned to universities, which would in principle make it possible to isolate the leader effect from other unobservable influences.2 This, of course, cannot be done
1A Treatise of Human Nature, by David Hume (1967, p. 76), edited by L. A. Selby-
Bigg. Thanks to Jeremy Miles for helpful discussion.
2 Economists often use “instrumental variables” to try to prove causality. To get an intu-
itive idea of this, think of the market for corn. The outcome, the price of corn, is deter-
mined by the shape of demand and supply curves and where they cut. Rainfall, say, is a
kind of instrumental variable. It shifts the supply curve of corn. As the supply curve of