Scholars are expected to pursue a research program and to contribute the results of their investigations to the scholarly record of their disciplines.The best estimate of scientific impact could be gained by an analysis of a researcher's total contributions carried out by qualified peers.
This is usually not feasible, so interest has focused on devising a "second best" method.
Such a method is often referred to as a "metric", and a number of candidate metrics have been proposed.
The word itself implies measurement, quantitation and arrangement on a scale.
This tutorial discusses the advantages and drawbacks of several metrics.
Characteristics of a Useful Metric
To be useful, an evaluation metric should be:
- Objectively derived
- Simple to calculate
- Consistently applied
The metric should arise from some established body of data and the method used should be explained. Deriving the metric should not require complicated techniques or take a long time. The same metric should be applied to all, in all cases. Research traditions of various disciplines should be respected in comparisons across disciplines.
Metrics Rely on the Scientific Literature
Candidate methods for establishing research impact generally rely on some form of bibliometrics, or the mathematical or statistical analysis of a portion of the scholarly record. Among the most common methods of estimating research impact are:
- Publication count
- Citation count
- Impact factor
- h-index, and its variants
- Eigenfactor analysis
Each of these has advantages and drawbacks.