Coincidences and Correlations

by Venkat on July 1, 2013

Most people have heard the admonition, “correlation is not causation.” Few have heard the related admonition, “coincidence is not correlation.”

In common usage, a coincidence is about pairs of rare events that have a background relationship within a model. Like thinking about a friend you haven’t thought about for ten years, and then running into him the next day. But here, I mean a more banal technical sense of “coincidence” — juxtaposition. Co-incidence, as in “occurring together”

So if you and I are at a coffee shop at the same time, our mutual presence is a “co-incidence” whether we are long-lost friends or strangers who ignore each other.

But why are these two admonitions necessary at all? Why would we assume relatedness among unrelated things, or see causal relationships where there are only correlations? They are necessary because decisions enacted in the real world as opposed to inside your head, share physical time and space with other enactments in progress: situations.

Situations can be understood as little bounded regions in space and times through which many narratives flow. In the coffee shop where I am writing this right now, there are several other people doing random other things. Some I’ll never see again and have no relationship to. Others I might see repeatedly and suspect a correlation. Perhaps they live in the same neighborhood. Or their presence on many days in a row might be a string of coincidences with no reason. Maybe they just happened to have several completely different errands in the neighborhood this week.

Correlation might have  a shared proximal cause (maybe we live in the same area) or a shared distant cause (I live near here, she works near here). Or there might be actual causation. He’s here because I am here. Maybe somebody trying to steal the secret papers I have in my backpack.

But stepping back, this is a very good way to develop situation awareness of a particular place and time. What are the major rivers of causal flow passing through this time and place? Which ones are just juxtaposed? Which ones run together elsewhere but are unrelated? Which ones interact before and after? Which ones are actually part of the same flow? Which of the many rivers of causation flowing through a time and place actually dominates that situation?

If you’re bad at generally being “in the moment” as I am, this can be a useful way to force yourself out of your own head and parse your surroundings. The huge value to doing this is that you might see connections between event streams that you can make.

Internally focused thinkers often miss opportunities because they are so involved in a particular story, they don’t pay attention to other stories that are evolving alongside. Externally focused thinkers don’t make this mistake. They are able to take advantage of coincidences and correlations.

So while the admonitions may be useful for statisticians analyzing data, they are even more useful for action-oriented people looking for opportunities. You should remember: coincidences can be turned into correlations. Correlations can be turned into causations.


Previous post:

Next post: