Friday, April 8, 2011

Overstating the Fetish for Methods

A Princeton grad student, who has now successfully undermined his career, posted a letter at the school's newspaper, complaining that he had been tricked.  He complains there is "hegemony in the politics department".  I am shocked.  Yes, there has been a fetish for advanced mathematical methods (because political scientists want to be paid like/respected like economists).  This is hardly news.  Even in a department that has a name such as "politics" rather than political science. 

This debate has been going on for a long time.  The folks who feel burned by this fetish for methods have been sharing the letter via facebook and twitter.  But I have to say: spare me. 

You see, I am living on the divide.  I do mostly case studies these days--because the datasets I need to do quantitative work are being built/revised/etc.  So, my perspective has evolved over time.  I do think that there are folks that emphasize advanced methods too much.  The reality of a tight job market is that it is better to have more skills than less, that to do well in a job talk you must appeal to those who do the numbers stuff (especially because they want other people to relieve them of the responsibility of teaching these inherently, ahem, boring classes to teach).  Grad programs have learned that their students need to have the most advanced skills to maximize their competitiveness.  BUT advanced skills with no decent ideas is not going get one very far either, except if one is really, really special on the skills (like one or two a year). 

One can take solace that the last several years, at least in international relations (I have no clue as to how the Americanist or theorist job markets work), we have seen a number of "hot" candidates who have been successful because of the quality of their ideas, not their methods.  That the folks who have done the poking and soaking that this Princeton grad student has said has gone out of fashion have done very well on the job market. 

To generalize from one example (which, of course, is so wrong from the standpoint of methods), I know of one student who went to a grad school with a good but not great reputation, who had some methods skills but did not deploy them in her dissertation, but had learned Arabic, spent much time in the Middle East, developed a really cool project that focused on a hot topic (non-state actors using violence), and was a fantastic teaching assistant.  She got seven phone interviews, had four flyouts to interview (and would have had two more had she not accepted a good job), and got a good job (with a hell of a better starting package than I would have dreamt of).  And her adviser was hardly a huge name in the field--just an indiscreet blogger who knows very little of the Mideast.


Moreover, there is plenty of qualitative stuff published in most of the leading journals (except those dominated by Americanists--you know who you are).  Plus it is easier to get case study stuff published in the major presses than quant stuff since figures and tables cost the publishers more money.   

This letter also is just contradictory:
Take a very bright, young graduate student. If someone that fresh has spent the first two years of the Ph.D. program in politics taking six classes in quantitative and formal methods as well as compulsory seminar classes on the “canon,” she has little time for substantive classes that give her a theoretical background on the questions she will ultimately answer in her dissertation.
Um, isn't the compulsory seminar class on the canon where the theoretical background is supposed to developed?  Plus it has always been the case that we expect students to learn outside of the classroom--that comprehensive exams test a student's understanding of the major works in their field and much of that reading is in addition to that stuff assigned in the class. 

Moreover, when speaking of "answering" the question in one's dissertation, one of the big questions is: how do you know what you think you know?  That is methods.  Contemporary political science mostly requires folks to be able to defend their arguments--that how they do the research is important, that the choices they make must be justified.  There must be a method to the madness.  And the really big issue is this: until you know what the question is, you do not know what method you will need.  So, it is better to have mastery over the major methods before one starts asking the questions, so that one's pursuit of the answer is not limited by one's skill set, right? 

The reality is that one's understanding of the advanced methods will always be in decline after grad school, unless one spends heaps of time on that.  So, grad students then become a convenient means by which profs can do advanced work--by co-authoring with the folks with the latest training. 

Finally, what truly demonstrates how clueless this kid is about contemporary political science in the US is this:
Added to the department’s preference for certain quantitative and formal methods is the unsaid belief that empirical political science must be divorced from normative concerns. In this view, the study of politics is a purely descriptive exercise. Questions of “good” and “bad” are parceled into “political theory,” a subfield hermetically sealed from the rest of the discipline.
Forget that the student does not know what descriptive means and focus on the frustration that today's political scientists are mostly focused on understanding how the world works, not whether how it works is good or bad.  This is old, old, old news.  If this student finds this surprising, then he clearly did not do any research before going to grad school.  He should have gone to a school in Canada (not McGill) or Europe where there is more of a conflation of normative and analytical (where analytical means pursuing the why questions).  Plus we do get to sneak in the normative stuff--we call that the implications/conclusion part of the research.

I do get the frustration that math matters when political science should just be about politics.  I know that I have lost jobs in the past to folks who had better quant skills.  But I am a realist--to get a job in an American political science department, you have to get some votes from the Americanists (who tend to be a large component of most departments), and they tend to focus on methods rather than the ideas.  But if you have a really cool project that you can articulate well, you can make it past the Americanists.  And yes, there are fetishists among the Comparativists and IR types as well. 

More importantly, I fundamentally believe that curiosity should drive the research, so that one should get the most complete training in methods possible, so that one's quest is not limited by one's skills.  One reason why I co-author is to get help from those who are more skilled than I.  I learned my stats on the streets, not the best place to learn such stuff.  I know I would have been better off, not just in the competition for jobs, but in my ability to do my work if I had the kind of training that this kid is complaining about.  My grad school (UCSD) got far more systematic about how to train the methods stuff shortly after I left. And the UCSD students that I have met since then are hardly methods drones--they ask really interesting questions, come up with some fun answers, and test them using all the methods that are relevant, including case studies. 

2 comments:

Dan Nexon said...

Great post. I found that odd as well. I see the main inflection point of that debate as now precisely over whether students need to to know the cannon or can just do "scientific" work.

Steve Saideman said...

There is another point that I forgot to raise: that one cannot really evaluate the theoretical work of others unless one has a familiarity with their methods. The hidden part of this is that just as qual people cannot evaluate the work of quant people as well as they could, quant people often cannot evaluate qual work well because they do not have the skills to assess qualitative work (it ain't as easy as they think it is).