Wars over the disputed Nagorno-Karabakh region have flared between Armenia and Azerbaijan for a century, and disputes over the the identity of the region simmered long before that. With a new war between Armenia and Azerbaijan looming over the horizon, it is interesting to note that the United States Congress maintains congressional caucuses to represent both Armenian and Azerbaijani interests. Is the fissure between the two nations of the Caucasus reflected in a fissure between the two caucuses?

What if I told you that only 9 of the 149 full members of the Congressional Caucus on Armenian Issues in the 111th Congress of 2009-2010 were also members of the Congressional Azerbaijan Caucus? Members of both caucuses were:

Darrell Issa
Eddie Bernice Johnson
Peter King
Frank Lobiondo
Joseph Pitts
Allyson Schwartz
Nydia Velazquez
Diane Watson
Joe Wilson

That doesn’t sound like a lot of people. It seems like Congress is split on favor for Armenia and Azerbaijan just like the two countries are split on ownership of the Nagorno-Karabakh region.

But I could frame the statistic in a different way: 9 of the 41 Representatives in the Azerbaijan Caucus were also members of the Caucus on Armenian Issues. That’s not a majority, but it seems to be at least a significant minority. 9 out of 149 seems like a much smaller number than 9 out of 41 I could also have reported just as easily that “only” 32 members of the Azerbaijan Caucus are not members of the Armenia Caucus.

Statistics can deceive, but only when they’re presented selectively, out of full context. The full context of the statistics is presented in all four cells of the following table, along with the row totals (149 in the Armenian Issues caucus, 286 not) and column totals (41 in the Azerbaijan caucus, 394 not). This is the actual data for the 111th Congress:

When I’m asking whether Armenia vs. Azerbaijan conflict is reflected in a division of Congress, what I really mean to ask is whether membership in the Armenian Caucus makes one less likely to be a member of the Azerbaijan Caucus, and if so to what degree. The following table shows how members of Congress would behave hypothetically if membership in one caucus made one no more or less likely to be a member of another caucus:

And this table shows what the pattern of caucus memberships in the 111th Congress would have looked like hypothetically if no members of Congress were members of both caucuses (a condition of complete exclusion):

How does the actual, first table compare to the two hypothetical examples? It lies somewhere in between. There are fewer actual members of both caucuses (9) than we would expect if people entered into the caucuses independently of one another (14), but that actual number is closer to 14 than to 0. (If you want to get wonky, a Chi-Square test shows that the relationship between membership in the caucuses has a p-value of 0.08 and is therefore of marginal significance.) If there is a membership fissure between the Armenian and Azerbaijani caucuses in the United States Congress, it is minor compared to the split between the two nations themselves.