Fluctuations around the average scaling curve.

<p><b>A</b>, Variance of the deviation from the average scaling curve as a function of the logarithmic city size, measured in voter turnout. City sizes are binned into 20 windows of uniform sizes on logarithmic scale. In the inset, standard deviation of SAMIs (<a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0192913#pone.0192913.e006" target="_blank">6</a>) for all metropolitan areas in our study as a function <i>β</i> − 1. Best fit line parameters are in the inset. <b>B</b>, Standardized deviation of SAMIs for the last five US presidential elections. Lower panel: Scatter plot for the Democrat (left) and Republican (right) standardized deviations (horizontal axis) and logarithmic city size (vertical axis). Upper panel: Distribution of the standardized deviations. For the Democrats (left) it is a standard normal distribution (solid line). For Republicans (right) it is a skewed distribution deviating from the standard normal distribution (solid line). We used the Kolmogorov-Smirnov test to check the normality of the distributions at a significance level of 5% (see Section G in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0192913#pone.0192913.s001" target="_blank">S1 File</a> for details).