The question that this paper attempts to address is well framed by rich literature and well researched. The author’s passion and efforts that have been put into variable specification and measurement should be highly appreciated considering the abstractness of explanatory factors and their difficulty of quantification. The empirical strategy, using multilevel models, is refreshing. But I have a question in this regard. Multilevel modeling has been chosen and employed in the paper because “it allows for partial pooling of the cabinets and the introduction of country invariant variables into the model without fear of identification issues”(p.17). As far as I get from the paper, there are two levels of variables in the model: electoral cycle and country levels. I am not familiar with multilevel modeling techniques, but it seems unusual that these two levels are treated as conceptually parallel. Don’t the variables at the electoral cycle level somehow overlap with ones at the country level? Or this might not be a problem in multilevel modeling? Another question I have with respect to the empirical test is that which independent variable is a proxy of the degree of “party institutionalization”. The third hypothesis predicts that more institutionalized parties will lead to less conflict. I can not find corresponding independent variables in the model to test this hypothesis.
You make a very good point in pointing out the lack of research and statistical testing of the types and outcomes of semi-presidential systems. I was wondering what role conflicts over power play in the establishing of the semi-presidential systems over parliamentary or presidential and if these have enduring effects on the outcomes? Can the intensity of these conflicts matter more than the specific mechanisms (veto powers, reserve policy areas) used as a compromise/agreement?
The question that this paper attempts to address is well framed by rich literature and well researched. The author’s passion and efforts that have been put into variable specification and measurement should be highly appreciated considering the abstractness of explanatory factors and their difficulty of quantification. The empirical strategy, using multilevel models, is refreshing. But I have a question in this regard. Multilevel modeling has been chosen and employed in the paper because “it allows for partial pooling of the cabinets and the introduction of country invariant variables into the model without fear of identification issues”(p.17). As far as I get from the paper, there are two levels of variables in the model: electoral cycle and country levels. I am not familiar with multilevel modeling techniques, but it seems unusual that these two levels are treated as conceptually parallel. Don’t the variables at the electoral cycle level somehow overlap with ones at the country level? Or this might not be a problem in multilevel modeling? Another question I have with respect to the empirical test is that which independent variable is a proxy of the degree of “party institutionalization”. The third hypothesis predicts that more institutionalized parties will lead to less conflict. I can not find corresponding independent variables in the model to test this hypothesis.
ReplyDeleteYou make a very good point in pointing out the lack of research and statistical testing of the types and outcomes of semi-presidential systems. I was wondering what role conflicts over power play in the establishing of the semi-presidential systems over parliamentary or presidential and if these have enduring effects on the outcomes? Can the intensity of these conflicts matter more than the specific mechanisms (veto powers, reserve policy areas) used as a compromise/agreement?
ReplyDelete