This research report use a replication data from Clark And Golder (2006) article to test the conditional hypothesis with interaction variable from Duverger’s theory. The argument of his theory state that increasing the number of ethnic groups within a country will only to an increase in the effective number of electoral party if the electoral system is proportional or permissive.
• Multivariate Regression Analysis
Table~ ef{tab:example1} shows that when there are ethnic groups in a society there is a slightly increase in effective number of electoral parties in a multivariate regression model, for which there is control of other confounding variables as logged district magnitude, upper ties seats, number of presidential candidate, proximity …show more content…
Also, the model accounts for about 28\% of the variation in effective number of electoral parties. With all this information, it is possible to say that the model does not fit well, because the R2 value is small for what we are expecting.
• Multicollinearity
The diagnostic test for multicollinearity was run by VIF command on Stata. The Table~ ef{tab:example2} shows that for ethnic group and logged district magnitude the values are higher, over 10, if we compare with the rest of the variable. This values are worrisome because they show a moderate multicollinearity. As Multicollinearity is a not a statistical problem but a data issue cite Kellstedt, it will be easy to resolve with more data, but in this case, it is not possible, because the theory is about countries with specific characteristics, advance democracies, and it is impossible to have more.
• Outliers
Figure~ ef{fig:ev1} shows that there are two outliers in the data, Philippines and Austria. Philippines has a high leverage but a small residual squared. Austria, has a moderate high leverage and a moderate high residual …show more content…
• Marginal Effect
As shows Figure~ ef{fig:ev2} increase of number of ethnic groups and logged district magnitude increase the effective number of electoral parties. The marginal effect value is positive, but logged district magnitude only is statistical significance between 2 and 3.5, even if the cases in that range are few. The more cases of Logged District Magnitude between 0 and 2, but those cases are not statistical significance.
With this information is possible to say that the theory of Duverger, state that in a more ethnic diversity society will only translate to more political parties competing at elections when the electoral system produces a fairer translation of votes into seats, it is correct only when there is a moderate logged district magnitude, between 2 and 3,5, but not as it expected with higher
• Multivariate Regression Analysis
Table~ ef{tab:example1} shows that when there are ethnic groups in a society there is a slightly increase in effective number of electoral parties in a multivariate regression model, for which there is control of other confounding variables as logged district magnitude, upper ties seats, number of presidential candidate, proximity …show more content…
Also, the model accounts for about 28\% of the variation in effective number of electoral parties. With all this information, it is possible to say that the model does not fit well, because the R2 value is small for what we are expecting.
• Multicollinearity
The diagnostic test for multicollinearity was run by VIF command on Stata. The Table~ ef{tab:example2} shows that for ethnic group and logged district magnitude the values are higher, over 10, if we compare with the rest of the variable. This values are worrisome because they show a moderate multicollinearity. As Multicollinearity is a not a statistical problem but a data issue cite Kellstedt, it will be easy to resolve with more data, but in this case, it is not possible, because the theory is about countries with specific characteristics, advance democracies, and it is impossible to have more.
• Outliers
Figure~ ef{fig:ev1} shows that there are two outliers in the data, Philippines and Austria. Philippines has a high leverage but a small residual squared. Austria, has a moderate high leverage and a moderate high residual …show more content…
• Marginal Effect
As shows Figure~ ef{fig:ev2} increase of number of ethnic groups and logged district magnitude increase the effective number of electoral parties. The marginal effect value is positive, but logged district magnitude only is statistical significance between 2 and 3.5, even if the cases in that range are few. The more cases of Logged District Magnitude between 0 and 2, but those cases are not statistical significance.
With this information is possible to say that the theory of Duverger, state that in a more ethnic diversity society will only translate to more political parties competing at elections when the electoral system produces a fairer translation of votes into seats, it is correct only when there is a moderate logged district magnitude, between 2 and 3,5, but not as it expected with higher