Political science has long had debates over methodology – i.e., ways of knowing about the world – but has had fewer over ontology – i.e. what exists in the world. This was noted by Peter Hall in his 2003 book chapter, “Aligning Ontology and Methodology in Comparative Research,” but other authors like Colin Hay and Liam Stanley have made the same critique.  

Why is this a problem? Two examples, one personal and one not:

1. When I was a graduate student, it wasn’t until I reached India for my first fieldwork trip that I realized it was possible the very thing I wanted to study — ethnic violence — may not actually exist. Of course, ethnic groups can be violent. That much was well documented. But was the violence about ethnicity? One of the most prominent scholars of Indian politics, Paul Brass, had suggested otherwise. Was the language of ethnicity, as used in academic discourse, even important to Indians? What was the ‘ontology’ of political violence?
   
2. Consider also the above story by Sartori, from his 1991 article “Comparing and Miscomparing,“ which is about the problems that deal with taking things that do exist and combining them together haphazardly. This results in (p. 243):  

“nonexistent aggregates which are bound to defy…any and all attempts at law-like generalizations.”

Are we conflating a concept with a distinct measurement that applies to one country (a cat) and another concept that is quite different and measured quite differently in another country (a dog)?

It is scary to consider the possibility that some of the things we study either 1) don’t exist, or (more likely) 2) don’t exist in a way that is at all relevant to our concepts, theories, or measurements. 

What to do? I would suggest a few tacks:

1. “Less Theory. More Description.” — One obvious solution to ontological misspecification is to devote more time (and journal space) to description and descriptive studies. 
2. Fieldwork — There is a poster in the Washington Post headquarters that says, “Go talk to people” (attributed to Marty Baron). Doing this in scientific ways is no easy task, but this is generally good advice. Get your hands dirty. Build ontology from the ground up. 
3. Concept Formation – As Gary Goertz (2006: 5) notes:

“Concepts are theories about ontology; they are theories about the fundamental constitutive elements of a phenomenon.”

Often we start with data without spending enough time thinking about the concepts to which the data are (supposedly) linked.      

There will, of course, never be complete agreement on ontological issues – on what exists in the world. That is normal. But investing in more description, more time in the field, and more time thinking about concepts can help us to close the gap between our theories and the world.