The correlation vs causation condundrum

Folks, as a scientist I live and breathe the analysis of data. And one of the cardinal rules of science and statistics is that correlation does not imply causation–in other words, we can only ever say that y and z are related, but not that one causes the other. There is always the possibility of confounding and lurking variables that could have unduly influenced results. And there are various mathematical models to ascertain the degree to which y and z are related to each other. The benefit of using this logic is that it prevents one from incorrectly pointing the finger at something as the cause of another thing (or conversely claiming that one thing doesn’t cause another).

The danger of this, on the other hand, is that sometimes it prevents people from pointing a finger at something at all. I’ve noticed that this has become a common rhetorical strategy in arguments and debates. And I’ll freely admit that I’ve used it in the past. When someone pulls out the “correlation vs causation” card it’s a surefire way to trump someone’s argument. After all, how does one address what amount to essentially invisible, hypothetical variables? It’s impossible to do so, obviously, so the argument ends with, “Well, correlation vs causation, so I guess it’s really a wash.”

The problem is that in real life there are no washes or cop outs. Practically speaking, yes, everything does have a cause. Perhaps sometimes there are multiple factors influencing the outcome of something, but all of those factors behave according to the laws of the universe, and therefore act in concert in a predictable, measurable way–even if we human beings cannot as of yet predict or measure those influences. I can’t help but feel that the “correlation vs causation” device completely undermines the scientific process.

When people use correlation vs causation outside of a scientific context, what they’re really saying boils down to,”Well it’s impossible to know for sure, so I win.” I mean, technically speaking, someone could use the correlation vs causation device to successfully argue against the statement: “If I throw a ball into the air then gravity will pull it back down to the earth.” In other words, it’s not a very practical argument. The logic of using the scientific principle as a rhetorical device is baffling. If someone wants to make the correlation vs causation point in an argument, then they’ve not only defeated their opponent’s argument, but they’ve simultaneously defeated every argument, including their own.

Blindly attacking the principles of certainty and uncertainty doesn’t make any sense because technically, using the scientific definition, everything has a degree of certainty and uncertainty associated with it. So what’s the point of using the argument at all? Cleary, z happened. We know that. We saw it. We have data confirming it. There’s no arguing whether or not an event happened. And, as we know, things just don’t happen by themselves. For every cause their is an effect and vice versa. There’s a rhetorical disconnect between knowing that something happened and then using an argument to say that it’s impossible to determine what caused said event.

It makes much more sense to attack probability than it does to attack correlation and causation. Rather than make a blanket statement about how it’s impossible to ever prove whether or not one thing causes another, it makes much more sense to make an argument about which things are more likely to have caused something. Because, as I mentioned in the previous paragraph, when something happens, there is obviously a cause. And in most cases we have evidence–data–about an event. And, as I mentioned at the beginning of this piece, there are mathematical models out there to determine the degree to which one thing is associated with something else. So doesn’t it make more sense to take all of the data, determine what the possible causes are, and then run each one through the math to determine which explanations are most probable?

Because in reality it makes no sense to declare, “Well, it’s impossible to ever be certain about anything, which makes all arguments invalid, which makes your argument invalid.” It makes much more sense to say, “Well, looking at the data, my argument is more likely than yours.” The second statement still leaves room for uncertainty, and it still leaves room for further investigation and discovery, while acknowledging that something had to have caused whatever you’re debating about.

Advertisements

4 thoughts on “The correlation vs causation condundrum

  1. My response ended up being so long here that I made it into a blog post and linked your article. As always I enjoy reading what you have to say, so I hope my post extends your thoughts adequately. Feel free to keep the conversation going if you wish! 🙂

  2. You make some very good and valid points. But I’m not ready to accept that the correlation vs. causation device completely undermines the scientific process. I do find this article extremely interesting, and I’ve had to read it several times over, but I think there’s something I’m not getting. When I’m discussing creation and evolution, for example, someone inevitably says, this caused that, or that X proves Y. The other person denies that, or they say something like, “Science deals with evidence and predictive modeling, not with proofs or certainty.” That’s all well and good until that person commits the same error, but then defends himself and explains why it’s okay in this situation to do that, but it’s not okay for someone else to commit that error.

    Further, who’s to say what argument is more likely to be correct? Isn’t that a personal opinion? I’m not trying to go off on a tangent here, but I think it’s more likely that the heavens and the earth were created in six literal days by God, but someone else may believe it’s more likely that everything came about by random chance. What’s more likely? Is it more probable that roadkill can come to life by being struck by lightning, or for a huge number of random molecules to self-replicate and become a living organism, or that God created man in his image?

    The whole correlation vs. causation, proof and certainty thing really gets me thinking about the best way to communicate with people, and I’m still trying to figure that out, and I think I’m a long ways away. Maybe you can help me out.

    1. You raise some excellent points. You’re correct in that it is exceedingly difficult to provide a uniform baseline for determining what is most likely to be correct. I’m sure that there are mathematical models for determining the probability of almost anything, but I’m also sure the math is waaaay above my head. Plus, the average person probably isn’t going to bust out the statistics in a friendly discussion.

      I think we can agree that different people require different degrees or amounts of evidence in order to change a personal belief. Obviously the amount and nature of the evidence to support the existence of a God is enough for you, but it isn’t enough for me. I’d have a hard time, however, coming up with a number or percentage indicating how much more likely it might be that mine is the correct answer, though.

      Then again, perhaps that’s just the nature of this particular argument. If we were comparing two ideas in the physical world that could both be tested by physical means, I think the math becomes a lot easier and determining the probability or likelihood of one probable cause over another becomes much simpler.

      But how do you quantify something that isn’t physical, corporeal, or otherwise measurable? You can’t, and so it’s impossible to use a scientific or mathematical model to arrive at any sort of statistical conclusion about something like God or heaven or any other spiritual or metaphysical thing.

      I feel like the more important takeaway is that it’s a very dangerous thing to try to deal in absolutes. And I personally believes that this goes for everyone, scientists and secular folks alike. We live in a dynamic universe, the majority of which we haven’t yet begun to explore and I like to think that means that anything is possible.

      So in my mind, there’s a very real difference between stating “I do not believe in a God” and “There is no God.” Saying that, “there is no God” is an unreasonable statement that cannot be tested or verified. Saying “I do not believe in God” is a statement based on some level and degree of evidence, and is not absolute. It acknowledges that we live in a dynamic universe and that information and knowledge is subject to change in the future, leaving room for reevaluation.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s