Mapping Science - Mapping Scientific Frontiers: The Quest for Knowledge Visualization - page 265

We illustrate the model using global citation counts of scientific publications
recorded in the Web of Science. NB models are defined as follows using log as the
link function.
Global citations Coauthors C Modularity Change Rate C Cluster Linkage C
Centrality Divergence C References C Pages
Global citations is the dependent variable. Coauthors is a factor of three levels of
1, 2, and 3. Level 3 is assigned to articles with three or more coauthors. Coauthors is
an indirect indicator of the extent to which an article synthesizes ideas from different
areas of expertise represented by each coauthor.
Three structural variation metrics are included as co-variants in generalized
linear models, namely Modularity Change Rate (MCR), Cluster Linkage (CL), and
Centrality Divergence ( C KL ). According to our theory of creativity, groundbreaking
ideas are expected to cause strong structural variations. If global citation counts
provide a reasonable proxy of recognitions of intellectual contributions in a
scientific community, we would expect that at least some of the structural variation
metrics will have statistically significant main effects on global citations.
The number of cited references and the number of pages are commonly reported
in the literature as good predictors of citations. In order to compare the effects of
structural variation with these commonly reported extrinsic properties of scientific
publications, References and Pages are included in the models. Our theory offers
a simpler explanation why the more references a paper cites, the more citations it
appears to get. Due to the boundary spanning synthetic mechanism, an article needs
to explain multiple parts and how they can be innovatively connected. This process
will result in citing more references than an article that covers a narrower range of
topics. Review papers by their nature belong to this category.
It is known that articles published earlier tend to have more citations than articles
published later. The exposure time of an article is included in the NB models in
terms of a logarithmically transformed year of publication of an article.
An intuitive way to interpret coefficients in NB models is to use incidence rate
ratios (IRRs) estimated by the models. For example, if Coauthors has an IRR of
1.5, it means that as the number of coauthors increases by one the global citation
counts would be expected to increase a factor of 1.5, i.e. increasing 1.5 times, while
holding other variables in the model constant. In our models, we will particularly
examine statistically significant IRRs of structural variation models.
Zero-inflated negative binomial models (ZINB) use the same set of variables.
The count model of ZINB is identical to the NB model described above. The zero-
inflated model of ZINB uses the same set of variables to predict the excessive zeros.
We found little in the literature about good predictors of zeros in a comparable
context. We choose to include all the six variables in the zero-inflated model to
provide a broader view of the zero-generating process. ZINBs are defined as follows:
Global citations Coauthors C Modularity Change Rate C Cluster Linkage C
Centrality Divergence C References C Pages
Zero citations Coauthors C Modularity Change Rate C Cluster Linkage C
Centrality Divergence C References C Pages