Pearson's r and author cocitation analysis: A commentary on the controversy

[...]when the authors from one set are added to the authors of the other set, the overall result is the following: (1) the correlations between the authors of the original set rise; (2) the author correlations within the two sets are all positive; and (3) the author correlations between the two sets are all negative. [...]Ahlgren, Jarneving, and Rousseau demonstrate, perhaps unintentionally, that Pearsons r is a very effective tool for partitioning sets.In his response, White writes, Some might argue that this sensitivity of r to the addition of unlike cases is a virtue, not a drawback, but we will not pursue that line here (p. 1251). The history of humanity stands in stark refutation of this axiom. [...]this requirement for relations remaining invariant has the potential of depriving researchers of opportunities for further analysis to understand the nature of the relationships within original set. [...]where does partition begin with a measure running only from 0 to 1, at0.5? [...]any measure satisfying their denition of a similarity measure lacks the partitioning clarity of Pearsons r. This requirement that a similarity measure only measure similarities and therefore be restricted to a range of 0 to 1 does not make sense in ACA, whose purpose is to partition authors into different sets. [...]of this, its axioms have only been logically derived and never tested against any social or other reality. [...]its entire system just hangs in mathematical space without any applicability in information science.