Figure 4. Causal hypotheses and test result.

(a) Schematic diagrams for hypotheses H1 and H2. H1: Mean degree is determined by activity through function μ_k = f(A). Then degree is random distributed according to the conditional probability distribution P(k|μ_k). H2 is the other way around. (b) and (c) Results of Monte-Carlo simulation with 10⁵ samples following H1 and H2 for the Spanish Wikipedia data. The vertical red lines show the goodness-of-fit χ² of the actual data to H1 and H2, respectively. The empirical analysis clearly favors H1 over H2.