Effect of parameters/model extensions on the qualitative dynamics.

<p>The plots are shaded to show whether model qualitatively behaves as <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0096523#pone-0096523-g002" target="_blank">Figure 2</a>. The model either matches (solid), deviates (dense hatching) or fails (thin shading). The qualitative fit is based on quantitative scores (see <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0096523#s4" target="_blank">Methods</a>). Firstly, ‘State formation’ () is high when states are large and collapse rapidly to few factions. Secondly, ‘Periodicity’ () is high if there is periodic predictability to decisions. Thirdly, ‘State size’ () is high if state formation and collapse affect all factions. Finally, ‘Capital stability’ () is high if the leading faction does not change from the initial leader (relevant only for plots e–h). The qualitative model is matched if , , and . It deviates if , or . Otherwise the qualitative model fails. Also shown (where possible) is the parameter value from <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0096523#pone-0096523-g002" target="_blank">Figure 2</a> (vertical line).