Wednesday, November 6, 2019

Medium N Analyses: Know A Little About A Lot?

When I was discussing my work with a pal yesterday at Tufts (Kelly Greenhill rocks mightily), she was surprised to find that I was the one in the Dave/Phil/Steve team doing the Latin American countries.  She didn't think I had any expertise there, and she is right.  But I asked: where do you think I have expertise?   I don't think she answered, but it gets to a downside of my method: medium N analysis.

Small n is where one studies a few cases, developing a deep knowledge about them.
Large n is where one studies many, many cases/observations using statistical techniques to determine which factors seem to matter more than others.

Folks tend to learn one, the other, or both in grad school.  Me?  Besides a few exceptions where I did some quantitative work, I have mostly been studying more than a few but less than many:
  • My dissertation: I studied the international politics of three secessionist crises and asked how a variety of countries responded to them.  So, who took sides in the Congo Crisis and why?  Who supported Nigeria vs Biafra or Biafra vs Nigeria and why?  Who supported Bangladesh's secession and why?  I ultimately studied something like forty cases.
  • My first book dropped the Bangladesh case since it was really all about India, plus I wanted to get hip and current, so why did countries support the Serbs or the Croats or the Bosnians in the wars of Yugoslavia's demise.
  • My second book, with Bill Ayres, considered why some countries engaged in irredentism while others did not in the 1990s.  This involved bigger case studies of Croatia, Serbia, and Armenia vs. Hungary, Romania, and, yes, Russia while considering shorter case studies of other inconsistent irredentism: Albania, Ireland, Cyprus/Greece/Turkey, Pakistan/Kashmir.
  • My third book, with David Auerswald, on NATO in Afghanistan studied Presidential systems (US, France, Poland), Single-Party Parliamentary Systems (UK, Canada plus Spain and Turkey), coalition governments (Denmark, Netherlands, Germany plus Belgian, Italy, and Norway), a few partners (countries who were not members of NATO--Australia and New Zealand), and then many countries reacting to the Libya campaign.
  • My fourth book is an exception and an accident--Canada in Afghanistan--because it could not all fit into the NATO book.
  • The fifth book, very much in progress, compares fifteen or so democracies.  We wanted to see how oversight of the armed forces varies among countries, focusing on the role of legislatures. So, we have three types of democracies (Presidential, Westminster, European), so we need some cases of each AND we want enough cases within each to understand the sources of within-category variation.  So... medium n.
There are a few patterns here--I like to have cases from all the various combinations of variables, I like to have more than a few cases per category, and I don't stick to the same cases.  I am not an expert on any particular country or region as my theory tells me which cases I need to study--which variables need to be varied to see what effects they may have.

Getting back to medium N analysis, I do it for several reasons but at some cost.  The upsides include:
  • Getting more variation than I would if I only studied a couple of cases.  For me (and not just for me), variation means leverage.  That with more variation, I can get at the casual mechanisms that are at work.  
  • Getting a fuller sample may not mean getting a full sample, but I have a greater chance of having representation of the thing I am studying.  For NATO in Afghanistan, I couldn't study all 20 plus members in the same level of intensity, but I could study nine closely and then a few more from a distance.  We covered most of the major actors as well as sampling some others to have non-members, to have smaller countries, and so on so that we could claim to have a representative sample of the countries involved in the enterprise.   
  • Simply learning more.  More is more, and the kind of stuff I am seeking to understand is just very interesting. 
There are challenges and costs:
  • I can do this kind of work if I am funded.  Going to four or five or six countries requires money.  It is probably no accident that I developed more ambitious projects when I moved to Canada where I have more access to more money
  • It tends to require co-authorship.  I don't have the time to go to fifteen countries, but the three of us together do.  I don't mind co-authoring--I enjoy it and get much from it--but there are folks out there who discount co-authored work.  And, to be sure, not all co-authoring relationships go well.
  • I am linguistically lame, so going to five countries with five different languages means having to find translators/interpreters.  This post is partially inspired by my copy-editing of a chapter in a book on fieldwork where I discuss this challenge directly.  
  • Oh, and back to the start, this method means I know a little bit about a lot of places but not that much about any one of them. 
I need to do more thinking to explain why I do this, what the advantages are, and so forth, as I think there is more medium N analysis out there, but we don't have guides/playbooks/etc.  Any suggestions would be most appreciated.

1 comment:

Rob Chasen said...

As a curious person, I'm curious to understand lots and lots of things, generally at a medium level of detail (because it takes a longer time to learn something in great depth, and at great depth, I often find things much less interesting. I have two bachelors degrees. When I considered, at several points in my 20s and early 30s, whether to get an advanced degree, I realized that there are half a dozen subjects I'd be interested to another bachelors in, but I didn't identify anything I really wanted to do a master's degree, especially a "professional" degree like an MBA or a JD (which might have been logical in my career in business).