Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
523388 | Journal of Informetrics | 2014 | 10 Pages |
•We show that the test by Tol (2009) for a potential Matthew effect is quite sensitive to its assumptions.•We propose an alternative test using the Kolmogorov–Smirnov test.•This test is applied to a large data set of economists.•The Matthew effect might be a temporary phenomenon not in terms of time but in terms of the citation count.
We apply the test of Ijiri and Simon (1974) to a large data set of authors in economics. This test has been used by Tol, 2009 and Tol, 2013a to identify a (within-author) Matthew effect for authors based on citations. We show that the test is quite sensitive to its underlying assumptions and identifies too often a potential Matthew effect. We propose an alternative test based on the pure form of Gibrat's law. It states that stochastic proportionate citation growth, i.e. independent of its size, leads to a lognormal distribution. By using a one-sided Kolmogorov–Smirnov test we test for deviations from the lognormal distribution which we interpret as an indication of the Matthew effect. Using our large data set we also explore potential empirical characteristics of economists with a Matthew effect.