xkcd – Correlation

Comments (13)

1. I didn’t forget. It’s bad manners to copy everything of someone else’s work. There has to be a reason for people to go visit.

2. This “correlation does not imply causation” thing has got out of hand. At least in the form it’s normally quoted. Because correlation does imply causation; just not the source of causation.

3. You forgot to copy the alt-text. Every xkcd comic has an alt-text that adds to the joke.

4. You forgot to copy the alt-text. Every xkcd comic has an alt-text that adds to the joke.

5. You forgot to copy the alt-text. Every xkcd comic has an alt-text that adds to the joke.

6. John Conway @ 1
   The number of fridges sold correlates with the number of washing machines sold. This does not show that one causes the other.

7. John Conway @ 1
   The number of fridges sold correlates with the number of washing machines sold. This does not show that one causes the other.

8. Because correlation does imply causation; just not the source of causation.

   No it doesn’t. I can pick two time series, say temperature in Sydney, and the price of an unrelated stock (e.g. the Southern Norwegian Peanut Pickling Company) that are strongly correlated, but there need not be any causation. I just need to make enough comparisons until I find one.
   KATZ – there is probably a causative element there: general wealth causes both fridge and washing machine sales to increase. That’s John Conway’s point, which is correct, but not sufficient to make the general claim he’s making.

9. Because correlation does imply causation; just not the source of causation.

   No it doesn’t. I can pick two time series, say temperature in Sydney, and the price of an unrelated stock (e.g. the Southern Norwegian Peanut Pickling Company) that are strongly correlated, but there need not be any causation. I just need to make enough comparisons until I find one.
   KATZ – there is probably a causative element there: general wealth causes both fridge and washing machine sales to increase. That’s John Conway’s point, which is correct, but not sufficient to make the general claim he’s making.

10. Because correlation does imply causation; just not the source of causation.

    No it doesn’t. I can pick two time series, say temperature in Sydney, and the price of an unrelated stock (e.g. the Southern Norwegian Peanut Pickling Company) that are strongly correlated, but there need not be any causation. I just need to make enough comparisons until I find one.
    KATZ – there is probably a causative element there: general wealth causes both fridge and washing machine sales to increase. That’s John Conway’s point, which is correct, but not sufficient to make the general claim he’s making.

11. Because correlation does imply causation; just not the source of causation.

    No it doesn’t. I can pick two time series, say temperature in Sydney, and the price of an unrelated stock (e.g. the Southern Norwegian Peanut Pickling Company) that are strongly correlated, but there need not be any causation. I just need to make enough comparisons until I find one.
    KATZ – there is probably a causative element there: general wealth causes both fridge and washing machine sales to increase. That’s John Conway’s point, which is correct, but not sufficient to make the general claim he’s making.

12. Because correlation does imply causation; just not the source of causation.

    No it doesn’t. I can pick two time series, say temperature in Sydney, and the price of an unrelated stock (e.g. the Southern Norwegian Peanut Pickling Company) that are strongly correlated, but there need not be any causation. I just need to make enough comparisons until I find one.
    KATZ – there is probably a causative element there: general wealth causes both fridge and washing machine sales to increase. That’s John Conway’s point, which is correct, but not sufficient to make the general claim he’s making.

13. While it’s true that correlation doesn’t imply causation, one of my gripes is that people throw around “correlation doesn’t imply causation” to undermine reaching the obvious conclusion. For example, smoking companies couldn’t deny the fact that smokers got lung cancer at a higher rate than the general population. So, they threw-out the “correlation doesn’t imply causation” meme to drag their feet, avoid all responsibility, and undermine the belief that smoking causes cancer. That’s my gripe with people throwing out the “correlation doesn’t imply causation” as a knee-jerk reaction – people often use it to drag their feet and avoid admitting the obvious.

Comments are closed.