Talk:Correlation does not imply causation

From Wikipedia, the free encyclopedia
Jump to: navigation, search


ATTENTION: This page was moved after a vote at Talk:Correlation implies causation/Page title.

Bells Theorem[edit]

Bells Theorem does not disprove local causality. This is the mainstream theory right now. What it does is 'disprove' the existence of a Theory exhibiting locality AND counterfactual definiteness.

Missing source for paragraph 2 General pattern[edit]

It is necessary to provide a source for the 6 relationships between 2 events. A possible source could be the following book: Causality: Models, Reasoning and Inference by Judea Pearl. As I don't have a copy in my possession, it would be nice if someone cold have a look into this book. I will delete the paragraph in 4 days, if there is no source provided. — Preceding unsigned comment added by Da Vinci Nanjing (talkcontribs) 17:56, 5 February 2017 (UTC)

I believe this section as a whole does not require a source as nobody would deny the fact that correlation can be due to multiple things. However, the question of whether this list is exhaustive could require a source, so I have restored the paragraphed but rephrased it in a way that does not suggest the list is exhaustive. If you think some specific elements of the list would require a source, I am happy to look for one, but I do not think that the general statement requires a source. It seems like an important section of this article, so I think it is important to keep it 7804j (talk) 21:48, 12 February 2017 (UTC)
All material on Wikipedia must be attributable to reliable, published sources, including any analysisis, interpretation, or synthesis such as describing a "general pattern". See § Original research, below. —Sangdeboeuf (talk) 20:59, 29 March 2017 (UTC)

Can we put in a PSA that Correlation does imply causation, just not necessarily so?[edit]

If I had a dime every time someone told me that correlation does not equal causation as a means to dismiss correlation as evidence... Perhaps a section labelled: "Improper use as an argument" ?

Akiva.avraham (talk) 16:47, 15 March 2017 (UTC)

Only if a reliable, published source makes that distinction. Wikipedia does not publish original thought. See § Original research, below. —Sangdeboeuf (talk) 21:06, 29 March 2017 (UTC)
Correlation does imply causation, but it doesn't mean causation. I think replacing 'imply' in the title by 'mean' would be better. Nederlandse Leeuw (talk) 01:15, 25 April 2017 (UTC)
An implication in this context is the propositional sense of an implication, see Implicational propositional calculus.GliderMaven (talk) 03:47, 25 April 2017 (UTC)
Yes, but encyclopædia article titles are expected to be in natural language, not propositional logic. Michael R Bax (talk) 20:59, 16 November 2017 (UTC)
The two are not mutually exclusive, as here, where the natural language seems to have incorporated a propositional logic term.GliderMaven (talk) 21:33, 16 November 2017 (UTC)
A better title could be "Correlation is not causation". Karlpoppery (talk) 23:47, 25 April 2017 (UTC)
* "Correlation does not show causation"
* "Correlation does not necessitate causation"
* "Correlation does not prove causation"
* "Correlation does not demonstrate causation"
Michael R Bax (talk) 20:59, 16 November 2017 (UTC)

Is it possible to prove causation, if correlation is negative sometimes? (talk) 09:48, 15 June 2017 (UTC)

The title is fine if one uses the term "imply" according to standard use in the field of logic (wikt:imply verb use 1 and 2), whereas a natural language use (wikt:imply verb use 3) isn't right. The title "Correlation is not causation" has other problems, because the nature of 'equivalence' here is unclear. Given this, I suggest no page move. Klbrain (talk) 17:08, 28 June 2017 (UTC)
Yes, but encyclopædia article titles are expected to be in natural language, not propositional logic. As such this title is misleading. In fact I was sent to this page by someone arguing that (in general rather than in logic) correlation does not imply causation! Michael R Bax (talk) 20:59, 16 November 2017 (UTC)

Oxford Living Dictionaries: "imply, verb: … 1.1 (of a fact or occurrence) suggest (something) as a logical consequence." Using this definition, correlation does imply causation. Michael R Bax (talk) 20:59, 16 November 2017 (UTC)

Original research[edit]

This article appears to consist of a great deal of original research – most of the general statements about the topic are unsourced, and appear to synthesize the various sourced examples. I'd suggest at a minimum adding the {{original research}} tag until the article has reliable sources to support the interpretation, analysisis, and synthesis given. —Sangdeboeuf (talk) 20:56, 29 March 2017 (UTC)

Note: I've added {{original research}} to the article. —Sangdeboeuf (talk) 14:52, 19 April 2017 (UTC)

There's a big difference between something being poorly referenced and original research. I've not noticed any original research, although there is much that is unsourced. Given that, and given it's otherwise impossible to remove a tag which is not specific, and given that you have failed to identify any specific problems here, I've removed the general tag; please retag any specific parts of the article you believe may be poorly referenced.GliderMaven (talk) 16:05, 19 April 2017 (UTC)

Move to "Correlation is not causation"[edit]

I want to open the discussion on possibly moving this article to "Correlation is not causation". The word "imply" often leads to pointless confusion when referring to this concept, as can be seen in this very talk page. "Is not" is often used and avoids any ambiguity. Does anybody object to this? Karlpoppery (talk) 15:30, 3 May 2017 (UTC)

There must be some historical value to CDNIC, and after all it's more precise in a more formal setting, but I for one support reconciling with CINC. — Preceding unsigned comment added by 2607:FCC8:620C:A000:4422:9DDB:7D2:C445 (talk) 23:39, 23 June 2017 (UTC)

Propositional logic does not belong in article titles. I support this proposal. Michael R Bax (talk) 21:02, 16 November 2017 (UTC)

I am opposed to this move:

  • First, the page is about a logical fallacy. It is completely appropriate that the word "imply" is used in its more rigorous sense, in agreement with its use in formal logic.
  • Second, the use of this meaning of "imply" is not some rare, weird, jargony usage. To complain that it amounts to "propositional logic in the title" is silly. It is a standard meaning of the word "imply." It is the most natural meaning for the word to have in the context of pointing out a logical fallacy.
  • Third, the "does not imply" version is more commonly used than the "is not" version.
  • Fourth, some of the core discussion in the article itself would become oddly disconnected from the article title if the change is made. eg: This fallacy is also known as cum hoc ergo propter hoc, Latin for "with this, therefore because of this". The word "therefore" in that version corresponds to the word "imply" in the saying. The fallacy "cum hoc ergo propter hoc" sounds like a subtly different (perhaps vitally different) fallacy than "correlation is not causation."
  • Fifth, "correlation is not causation" is imprecise. What would "correlation is causation" even mean? Correlation is something that is directly observed (for instance, in the most formal setting, by statistical analysis of data in an experiment). Causation is something that is inferred or hypothesized or described by theory, about the events that were observed. It only works if the word "is" is suitably interpreted. It is metaphorical use of language, whereas "does not imply" is a precise, logical statement.
Gpc62 (talk) 00:05, 20 November 2017 (UTC)