Fig. 2: Preferential attachment effect on Gini. | Nature Cities

Fig. 2: Preferential attachment effect on Gini.

From: Parallel scaling of elite wealth in ancient Roman and modern cities with implications for understanding urban inequality

Fig. 2

The boxes each refer to N = 100 Gini estimates for each level of preferential attachment (PA), which indicate the median and interquartile ranges (IQR). The whiskers indicate 1.5 times the IQR, and the circles are outliers beyond that range. As the plot indicates, increasing preferential attachment strength leads to increased Gini coefficients (greater inequality). The specific model settings chosen (the range of PA values explored and number of agents) were chosen so as to produce a clear image depicting the model dynamics. It should be noted that since Gini coefficients are bounded in [0,1], and PA values ultimately lead to a continuum between complete connectivity and a network where all agents are connected to only one super-connected agent, the expected relationship between the PA and Gini coefficient is actually S shaped, and increasing PA further reveals a smooth increase to an asymptote where the Gini coefficient is 1 (interested readers can explore these dynamics by running the Python code provided in the project repo in ref. 38).

Back to article page