Mandy McConnell
Dr. Gang Guo
POL 523
29 April 2009
Final Essay
In political science we often explore questions of the utmost complexity. In contrast to the hard, physical sciences, the social sciences must attempt to quantify and operationalize variables that are illusive and multifaceted. Whether a researcher is defining democracy or measuring ideology, a host of problems is likely to arise with attempting to quantify such inherently qualitative concepts. Perceptions and subjectivity are also likely to play a role in the methodological decisions any political scientist must make. Because of these issues we must spend time carefully examining our model, sample, and variable choices to ensure the most accurate and unbiased criteria are selected with which to examine our hypotheses. Yet regardless of our statistical rigor, all sub-fields of political science are likely to face similar methodological issues. The pieces in this week’s readings are by no means immune.
As Box-Steffensmeier et al tells us, there are marked differences between the partisan preferences of men and women often referred to as the ‘partisan gender gap.’ The reasons for the existence of such differences have long been debated and range from institutional explanations to more vague conceptions of cultural and societal pressures on preference formation. The fact remains that gender is a salient to our political ideologies. Despite the extremely interesting theoretical implications of this line of research, little empirical testing has been done that takes into account the complicated nature of measuring partisanship. Here the authors simply create a dummy variable to capture whether a respondent was a democrat or republican. This fails to take into account the various other political identifications as well as the strength of partisanship that may explain the variation in the data better than the model’s simplified measure. Furthermore, I argue that the measure of ideological polarization, here ‘Conservative Macroideology’, is another over-simplified measure of the political environment that cannot reflect the real-world. Perhaps dis-aggregating the political affiliation measure into an ordinal (i.e. conservative, moderate conservative, moderate, etc.) variable would be more appropriate. And though I do not have a current recommendation for accurately capturing the ‘political climate’ I am not convinced that a measure of democratic polarization , as well as conservative polarization, should be excluded from the model. These are not issues of nuance; these are measures that make the model more theoretical and less applicable.
Departing from the American context, Gershkoff and Kushner’s paper Shaping Public Opinion also suffers from its own methodological issues. Though I do not doubt that the rhetoric of the Bush administration worked to inform the public’s opinions about the ‘war on terror’ , there is no measure that safely captures the extent of said rhetoric’s affect. Simply showing that there is correlation between speech dates, selective news coverage, and manipulation of the terrorism index does not immediately imply causation. Public support for the Iraq war is most likely based on a number of interrelated issues including economic, religious, and ideological factors. In other words, I am not convinced that Bush convinced us. Perhaps the public’s support of the Iraq war is really a proxy for the support of underlying issues that the administration worked to exploit but that are not explored in this particular paper. I realize that the scope of an individual research project must necessarily be limited; but a balance must be found between parsimony and the examination of the proper variables.
That said, there are identifiable problems with our operational concepts in and across all fields of the social sciences, including political science. I do not have all of the answers and will readily admit that I am also constantly struggling to find a better way to define the variables most useful for our studies. The purpose of this paper was to provide examples that address the common issues that belie all of our political work. The exceptional pieces discussed above are some of the best examples of new work in our field and it works to show us that the questions and methods of one sub-field are not necessarily any farther along than another. Keeping the lines of communication open across all of these schools will prove to be important for our continuing development. I firmly believe that only by working together will we finally find the most parsimonious and realistic ways to define, operationalize, and test our multitude of dependent and independent variables.