6 Rationale: What to Measure



The chapter describes the rationale behind what kinds of data to look for, data sources and their reliability and the various statistical and computational methods used to analyse them. Three types of data are sought for this purpose: data on research output, data on research funding and data on research collaboration. Data sources are various: national publication and patent databases are the sources for research output. The databases of government funding agencies are the sources of research funding. A mix of institutional repositories and university human resources databases serve as the source for research collaboration.

Introduction

Why is the journal Geology not classified under Geology?

There are proximate causes and ultimate causes. This dissertation will try to uncover some of the factors determining the ultimate causes of such deviances.

A distinction needs to be made between "indicators" of quality (proxy measures) and "determinants" of quality

1 A Resource-based view of the Environemnt

Three kinds of resources: knowledge resources, economic resources, political resources.A triad of resources:
- Research output
- Research funding
- Research interactions
Research Collaboration is helpful for acquiring contracts and funding (Nieminen& Kaukonen, 2001; Harman, 2001), enhances scientists’ productivity (e.g. Lee & Bozeman, 2005; Landry et al., 1996; Harman, 1999)

2 Data Sources


Research
Statistics (Coarse-grained: ABS, university rankings)
Bibliometrics (Institutional repositories)
Patents (Software is not accounted for)
Funding
Competitive grants/projects (Grants/projects databases)
Research Centres

Collaborations
Research is labour-intensive
Why is collaboration a good thing?
Social Networking Analysis (Human resources databases)

Caveats
The validity of bibliometric studies rests on the crucial assumption that co-authors are identical to co-operators. Empirical research has shown that this is not always the case (Laudel 2002) and (Martin, 1997)

If rankings are no good for measuring performances of individuals, organisational and disciplinary units, what are they good for?

Not all properties of the research ecosystem are being measured (or measurable) but this is a general problem that need to be address by everyone.

3 Overview of the Data


x