Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

When can someone "elicit a cause from a correlation"?

nathanieljschwartznathanieljschwartz Alum Member
edited September 2017 in Logical Reasoning 1723 karma

So im going deeper and deeper into concepts and working on cookie cutters. Is there any situation where it would be appropriate to assume a cause from a correlation?


  • OlamHafuchOlamHafuch Alum Member
    2326 karma

    You can never ever conclude causation from correlation, but correlation does make causation more likely. Therefore, additional correlation can strengthen an argument.

  • BirdLaw818BirdLaw818 Free Trial Member
    553 karma

    I don't think you need to do this but the only way correlation can correctly turn into cause and effect is at the veryyyyy least if you rule out the possibility that the relation isn't reversed, you've ruled out all lurking variables, and that X always causes Y. You'll never need to infer this much or consider it. Flaw questions will sometimes infer a cause and effect from correlation but it's always blatantly obvious. the other times where you deal with actual causation in an argument , you wont have to make that jump from correlation to causation.

  • LSAT Is ComingLSAT Is Coming Alum Member
    edited September 2017 530 karma

    This is a lot, but since you asked (others feel free to correct me, this is me spitballing from a Stats class I took a few years back):

    When A and B are positively (or negatively, but let's stick with positively) correlated, there are four possibilities:
    * A causes B
    * B causes A
    * Outside force C causes A or B or both (or multiple outside forces do)
    * Random occurrence

    The crux of the common fallacy occurs because people are prone to inferring a cause from a correlation alone, without outside information. To correctly infer a cause from a correlation between two things A and B necessitates identifying the following:
    * A and B are correlated to a statistically significant degree
    * This relationship between A and B recurs throughout nature (which would indicate that an outside force C doesn't cause one or the other, or the relationship between the two)
    * A precedes B (or B precedes A) -- on a deeper level, typically this temporal distance should be predictable based on other factors in the environment if there is actual causation

    It is only possible to infer the above through repeated interactions with a phenomenon in nature or through extensive experimentation. Obviously, the bar is pretty high, and rightfully so. For the sake of the LSAT, I don't think I've come across a question in which an arguer has correctly assumed causation, which is why the flaw is so cookie-cutter.

Sign In or Register to comment.