Testing for the Network Small-World Property

Researchers have long observed that the “small-world” property, which combines the concepts of high transitivity or clustering with a low average path length, is ubiquitous for networks obtained from a variety of disciplines including social sciences, biology, neuroscience, and ecology. However, we find three shortcomings of the currently popular definition and detection methods rendering the concept less powerful. First, the classical definition combines high transitivity with a low average path length in a rather ad-hoc fashion which confounds the two separate aspects. We find that in several cases, networks get flagged as “small world” by the current methodology solely because of their high transitivity. Second, the detection methods lack a formal statistical inference, and third, the comparison is typically performed against simplistic random graph models as the baseline which ignores well-known network characteristics. We propose three innovations to address these issues. First, we decouple the properties of high transitivity and low average path length as separate events to test for. Second, we define the property as a statistical test between a suitable null model and a superimposed alternative model. Third, the test is performed using parametric bootstrap with several null models to allow a wide range of background structures in the network. In addition to the bootstrap tests, we also propose an asymptotic test under the Erdös-Renýi null model for which we provide theoretical guarantees on the asymptotic level and power. Applying the proposed methods on a large number of network datasets, we uncover new insights about their small-world property. 1 ar X iv :2 10 3. 08 03 5v 1 [ st at .M E ] 1 4 M ar 2 02 1