9 thoughts on “Correlation Does Not Mean Causation”

  1. You missed something that John Lott did not, when he titled his famous book “More guns, less crime”. He showed negative correlation, but in addition he showed temporal order, which is one way to support the hypothesis that you’re dealing with causation rather than merely correlation.

  2. Negative correlation can imply causation, but in the opposite direction — i.e. if the hypothesis is “more guns causes more crime” but there’s a negative correlation, causation could be:
    – “more guns causes less crime”, or
    – “less crime causes more guns”, or
    – “more crime causes less guns”, or
    – “less guns causes more crime”.

    (Given the “ratcheting” effect and circular logic of urban Democrat leaders’ anti-gun policies, those last two in tandem seem particularly likely.)

    In any case, negative correlation flatly disproves a “more guns, more crime” causative relationship. And as pkoning points out, John Lott’s temporal analysis further disproves it; cause must precede effect, not the other way around. Loosening gun laws cannot be shown to cause more crime if a rise in crime came before less-restrictive gun laws went into effect.

  3. Have you looked into what countries are actually counting when it comes to homicide rates? Definitions are not as consistent as you may expect.
    For example, both Australia and the UK don’t count murders unless they have a suspect in mind or charged – no suspect = not a murder.
    This is why Australia claims to have had no mass shootings since they tightened gun laws in 1996: in mass shootings since then, the perpetrator has either killed himself or been killed by police, so their crimes don’t count towards official statistics.

    I’m sure other countries have differing definitions also. If you look up infant mortality, you’ll see a similar chart where the US appears worse than many third world countries – because of differing definitions, some of which appear designed purely to make the government look better.

  4. +1 for Jonathan. Data definitions are hugely important.

    Also I bet El Salvador looks a lot different with new data after they locked all the bangers up.

  5. Good comments all!
    Richard gets the bingo! Cause the whole gun=crime=homicide was a communist canard to begin with.
    Like the term “gun-violence” they always start a discussions with. (Because they’re masters at making sure the argument starts the way they want it to. And as Vox points out, it’s aways a lie.)
    There is no such thing as gun-violence. Only humans doing the wrong kind of violence with guns.
    Guns=homicide is a lie because a causation/correlation can’t happen without human intent.
    The streets could be paved with guns to no one’s hurt without that intent.
    And as Richard points out, when you lock up the criminals? One’s homicide rates drop to nothing.
    Detroit has a homicide problem. And in truth only a portion of Detroit as the problem. And guess who lives in that part?
    Should we consider it coincidence that the #1 taboo to discuss in America is race? (It’s more acceptable in America to talk about incest, and banging your sister when you were going up, than to discuss the differences in races.)
    That’s the causation/correlation and the reason it can never be fixed.
    Let’s just say the problem isn’t guns, it’s people with little or no impulse control.
    No matter, no one wants to fix homicide “problem” anyway. They just want to make sure it’s all one-sided.
    It’s simple. Communist want power. And that comes from the barrel of a gun…..That you don’t own.

  6. Clearly the USA is a statistical outlier.

    Moreover, from this chart we can deduce what the per capita firearms ownership in this country “should” be, which is roughly 20% based on global averages any the current homicide rate.

    Therefore the USA has “too many” guns, and additional gun control is clearly needed.

    See how this works?

    • Let’s put it another way:

      They are arguing from different premises.

  7. Correlation does not equal causation.

    But causation *REQUIRES* correlation.

    • Indeed, but that non-zero correlation might come with either sign.

Comments are closed.