Skip to main content
Log in

Reference classes: a tool for benchmarking universities’ research

  • Published:
Scientometrics Aims and scope Submit manuscript

Abstract

Based on new comparison principles that take into account both the volume of scientific production and its impact, this paper proposes a method for defining reference classes of universities. Several tools are developed in order to enable university managers to define the value system according to which their university shall be compared to others. We apply this methodology to French universities and illustrate it using the reference classes of the best ranked universities according to several value systems.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

Similar content being viewed by others

Notes

  1. Several contributions have aimed to overcome some of its shortcomings, proposing new measures such as the g-index (Egghe 2006), the tapered h-index (Anderson et al. 2008), w-index (Woeginger 2008).

  2. These data are managed and enriched in house by the Observatoire des Sciences et Techniques through a very detailed and precise techniques that involves directly the institutions for the selection of the appropriate list of addresses mentioned in their publications.

  3. See Carayol and Lahatte (2011) for details.

  4. Thus, the study does not concern the schools that are associated to other ministries (Defense, Industry and Agriculture mainly).

  5. The function \(g_{i}^{k}\left( x\right) =\int_{x}^{\infty}sf_{i}^{k}\left( s\right) \hbox{d}s\) would be used to assess weak dominance relations.

References

  • Anderson, T. R., Hankin, R. K. S., & Killworth, P. D. (2008). Beyond the Durfee square: Enhancing the h-index to score total publication output. Scientometrics, 76, 577–588.

    Article  Google Scholar 

  • Carayol, N., Lahatte, A. (2011). Dominance relations when both quantity and quality matter, and applications to the comparison of US research universities and worldwide top departments in economics. Mimeo.

  • Egghe, L. (2006). Theory and practice of the g-index. Sientometrics, 69, 131–152.

    Article  Google Scholar 

  • Garfield, E. (1956). Citation indexes for science. Science, 123(3184), 61–62.

    Google Scholar 

  • Hirsch, J. E. (2005). An index to quantify an individual’s scientific research output. Proceedings of the National Academy of Sciences of the United States of America, 102, 16569–16572

    Article  Google Scholar 

  • Lindsay, D. (1978). The scientific publication system in social science. San Francisco, CA: Jossey-Bass

    Google Scholar 

  • Marchant, T. (2009). An axiomatic characterization of the ranking based on the h-index and some other bibliometric rankings of authors. Scientometrics, 80, 325–342.

    Article  MathSciNet  Google Scholar 

  • Merton, R. K. (1973). The sociology of science. Theoretical and empirical investigations. Chicago: University Press of Chicago.

    Google Scholar 

  • Moed, H. F. (2006). Bibliometric Rankings of World Universities, CWTS Report 2006-01, Centre for Science and Technology Studies, Leiden University.

  • Price, D. J. de Solla (1963). Little Science, Big Science. New York: Columbia University Press.

    Google Scholar 

  • Woeginger, G. J. (2008). An axiomatic characterization of the Hirsch-index. Mathematical Social Sciences, 56, 224–232.

    Article  MathSciNet  MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Nicolas Carayol.

Appendix: properties of reference classes: proofs of Theorem 1

Appendix: properties of reference classes: proofs of Theorem 1

The properties of reference classes essentially build upon the properties of dominance relations. To show this we first need to introduce a useful definition that establishes that a dominance relation is stronger than another if a dominance relation of the former type between two institutions implies a dominance of the latter over the former, for any pair of institutions.

Definition 4

A dominance relation \(\succ\) is stronger than dominance relation \(\succ^{\prime},\) noted \(\succ\gg\succ^{\prime},\) if, \(\forall i,j, i\succ j\) implies \(i\succ^{\prime}j.\)

We can now introduces some causal relations between the dominance relations.

Lemma 1

\(\forall\phi,\phi^{\prime}\in\left[ 0,1\right] \) such that \(\phi \ge \phi^{\prime}, \forall k\in K\) and \(\forall m < \left\vert K-1\right\vert\):

  1. 1.

    \(\blacktriangleright_{k}^{\phi}\gg\vartriangleright_{k}^{\phi}\gg\trianglerighteq_{k}^{\phi}\);

  2. 2.

    \(\blacktriangleright_{k}^{\phi}\gg\blacktriangleright_{k}^{\phi^{\prime}}, \vartriangleright_{k}^{\phi}\gg\vartriangleright_{k}^{\phi^{\prime}}\) and \(\trianglerighteq_{k}^{\phi}\gg\trianglerighteq_{k}^{\phi^{\prime}}\);

  3. 3.

    \(\blacktriangleright^{\phi}\gg\vartriangleright^{\phi}\gg\trianglerighteq ^{\phi}\);

  4. 4.

    \(\blacktriangleright^{\phi}\gg\blacktriangleright^{\phi^{\prime}}, \vartriangleright^{\phi}\gg\vartriangleright^{\phi^{\prime}}\) and \(\trianglerighteq^{\phi}\gg\trianglerighteq^{\phi^{\prime}}\);

  5. 5.

    \(\blacktriangleright_{k_{1},\ldots,k_{m}}^{\phi}\gg\blacktriangleright_{k_{1},\ldots,k_{m}}^{\phi^{\prime}}, \vartriangleright_{k_{1},\ldots,k_{m}}^{\phi}\gg\vartriangleright_{k_{1},\ldots,k_{m}}^{\phi^{\prime}}, \trianglerighteq_{k_{1},\ldots,k_{m}}^{\phi}\gg\trianglerighteq_{k_{1},\ldots,k_{m}}^{\phi^{\prime}}\).

  6. 6.

    \(\blacktriangleright_{k_{1},\ldots,k_{m+1}}^{\phi}\gg\blacktriangleright_{k_{1},\ldots,k_{m}}^{\phi}, \vartriangleright_{k_{1},\ldots,k_{m+1}}^{\phi}\gg\vartriangleright_{k_{1},\ldots,k_{m}}^{\phi}, \trianglerighteq_{k_{1},\ldots,k_{m+1}}^{\phi}\gg\trianglerighteq_{k_{1},\ldots,k_{m}}^{\phi}.\)

Proof

The proofs derive directly from the above definitions.\(\square\)

We now need to establish a correspondance between the relation between dominance relations and the relation between reference classes. This is the purpose of the following lemma.

Lemma 2

If \(\succ\gg\succ^{\prime}\) then \(c_{i}^{\succ^{\prime}}\subseteq c_{i}^{\succ},\forall i\in I.\)

Proof

Assume \(\succ\gg\succ^{\prime}.\) Then, \(\forall i,k \in I, i\succ j\) implies \(i\succ^{\prime}j\) which is equivalent to \(i\nsucc^{\prime}j\) then \(i\nsucc j.\) Now assume that \(j\in c_{i}^{\succ^{\prime}},\) which requires that either \(a) i\nsucc^{\prime}j\) and \(j\nsucc^{\prime}i\) or \(b) i\succ^{\prime}j\) and \(j\succ^{\prime}i.\) If (a) holds, then the definition of the “≫” relation leads to \(i\nsucc^{\prime}j\) and \(j\nsucc^{\prime}i,\) which in turn implies that \(j\in c_{i}^{\succ}.\) Now let us consider that (b) holds. When \(i\succ^{\prime}j,\) it is impossible that both \(j\succ i\) and \(i\nsucc j\) which simply means that it is impossible that \(j\notin c_{i}^{\succ}.\) Similarly, when \(j\succ^{\prime}i,\) it is impossible that both \(i\succ j\) and \(j\nsucc i,\) that is, it is impossible that \(j\notin c_{i}^{\succ}.\) Therefore, necessarily \(j\in c_{i}^{\succ}\) (and \(i\in c_{j}^{\succ}\)). So, it is possible to conclude that when \(j\in c_{i}^{\succ^{\prime}}\) then \(j\in c_{i}^{\succ},\) which implies that \(c_{i}^{\succ^{\prime}}\subseteq c_{i}^{\succ}.\)

By application of the above Lemma, the five properties of reference classes in Theorem 1 now directly derive from the five properties of dominance relation in Lemma 1. \(\square\)

Rights and permissions

Reprints and permissions

About this article

Cite this article

Carayol, N., Filliatreau, G. & Lahatte, A. Reference classes: a tool for benchmarking universities’ research. Scientometrics 93, 351–371 (2012). https://doi.org/10.1007/s11192-012-0672-2

Download citation

  • Received:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11192-012-0672-2

Keywords

Navigation