I don't recall who coined that phrase. However, it is frequently misused.
Sometimes it is used to put down "bad researchers" who use correlational
methods (including ordinary regression) and "good researchers" who use ANOVA
methods. Sometimes it is used to mean that if there is correlation, causation
is ruled out. These polemicists don't realize that the stat methods are the
same, that it is the design of the data gathering that supports (or doesn't
support) causal inference. Further, causal inference needs to have a model
that shows that the cause (independent, predictor) variables "go along with"
(correlate with, lead, are associated with) the effect (dependent, criterion)
variable. Using the word correlation in the broad sense, showing correlation
is an important (necessary) part of inferring causation, but it is not enough
by itself (is not sufficient) to make the inference.
I seem to recall that John Stuart (sp?) Mills had an essay something like
"joint presence and absence" that much of this kind of causal inference is
laid out on.
[Funny isn't it how only part of what one read in epistemology in 1966 remains
in 2001.]
Andrew Morse wrote:
> Who was the first to say "Correlation does not imply causation" in so many
> words? I know that the idea dates back to David Hume, but Hume did his
> work about a century before the term "correlation" acquired its modern
> statitical meaning. I've seen many sources that crdit Karl Pearson with
> banishing the idea of causation from modern statistical theory, but none
> that attribute the quote directly to him. Sewall Wright? Francis Galton?
> Ronald Fisher? Any other candidates?
>
> -- Andrew
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
