Comparison of the (A) Durkheimian model and (B) the noisy BC-model.

<p>Figures plot the size of the biggest cluster versus the number of clusters and compare it to the case of random fragmentation in all simulation runs that resulted in more than one and less than 32 clusters. Fig. 4A is based on the simulation experiment with the Durkheimian model underlying <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1000959#pcbi-1000959-g003" target="_blank">Fig. 3</a>. Fig. 4B is based on an experiment with the BC-model <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1000959#pcbi.1000959-Hegselmann1" target="_blank">[33]</a> where we varied the bounded-confidence level between 0.01 and 0.15 in steps of 0.02 and the noise level between 5 and 50 in steps of 5. We conducted 100 replications per parameter combination and measured the number of clusters and the size of the biggest cluster after 250,000 iterations. White solid lines represent the average size of the biggest cluster. The dark blue area shows the respective interquartile range and the light blue area the complete value range. For comparison, we generated <i>randomly fragmented opinion distributions</i> of agents where agents hold random opinions () and the remaining agents hold opinion and form one big cluster. We varied the value of between 0 and 100 in steps of 1 and generated 1000 distributions per condition. The average size of the biggest cluster of the resulting distributions is shown by the thin yellow-black line. (The curve stops at 22, since this is the highest number of clusters generated.) The bold yellow-black lines represent the related interquartile range. We find that the value range of the Durkheimian model (blue area) hardly overlaps with the interquartile range of the fragmented distributions (yellow area). This demonstrates that the Durkheimian model shows clustering rather than fragmentation. In contrast, Fig. 4B illustrates that the distributions of the noisy BC-model and the results for random fragmentation overlap.