12.07.2015 Views

Contextual Determinants of Electoral System Choice - Åbo Akademi

Contextual Determinants of Electoral System Choice - Åbo Akademi

Contextual Determinants of Electoral System Choice - Åbo Akademi

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Double-checking with four separate binary logistic regression models confirmsthese results. The comparison <strong>of</strong> plurality systems with all other systems returnsthe highest chi-square value. All models are significant at the 0.001 level, exceptfor the comparison between majority systems and other systems, which barely failsto reach the 0.01 level <strong>of</strong> significance. British and French legacy explain theadoption <strong>of</strong> plurality systems, both effects being significant at the 0.001 level. Themodel that compares majority systems with other systems is very poor – partlybecause the majority systems constitute only 15 per cent <strong>of</strong> all cases, partlybecause majority systems do not stand out from other systems as a homogeneousgroup, at least not in terms <strong>of</strong> patterns <strong>of</strong> diffusion. Temporal diffusion during thefourth period is the most important variable in explaining the introduction <strong>of</strong> mixedsystems. Regional diffusion stands out as the principal determinant <strong>of</strong> electoralsystem choice when proportional systems are compared to all other systems.Cultural diversity and population size are added to the regression model in tables8.2 and 8.3. Comparisons between plurality systems and all other systems arepresented in table 8.2, whereas coefficients <strong>of</strong> the three remaining comparisons aregiven in table 8.3. The chi-square and the Nagelkerke values are slightly higher butthe –2 log likelihood value also exceeds the one in table 8.1, suggesting that theexplanatory model is not improved by including the two variables <strong>of</strong> the rationalperspective. Neither cultural diversity nor population size reaches a satisfying level<strong>of</strong> significance in any comparison between different categories <strong>of</strong> the dependentvariable. However, when plurality systems are compared with all other systemsrespectively, the significance <strong>of</strong> cultural diversity is not very far from the 0.05level, indicating that there is a vague pattern <strong>of</strong> plurality systems associated withculturally fragmented countries. Concerning patterns <strong>of</strong> diffusion, the tendenciesobserved in the former model remain, slightly strengthening the importance <strong>of</strong>temporal diffusion during the fourth period and weakening the role <strong>of</strong> Frenchlegacy. The likelihood ratio test returns significant chi-square values for Britishlegacy, French legacy, and the fourth period at the 0.001 level, while the effect <strong>of</strong>regional diffusion comes very close to that level.When these findings are double-checked with four separate binary logisticregressions, the difference between plurality systems and other systems is onceagain the most prominent <strong>of</strong> all, whereas the comparison <strong>of</strong> majority systems with212

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!