[thaumasiotes]

> For science, the existence of the correlation isn't enough to argue for a predictive relationship. For science, someone would have to design a test in which the drawing was proven to anticipate later developments, rather than accompany them passively

You're confusing a "predictive" relationship with a "causative" relationship. A correlation and a predictive relationship are the same thing.

[lutusp]

> A correlation and a predictive relationship are the same thing.

Not at all. Do puddles predict rain, or accompany it? A prediction is specific, and distinct from a simple correlation. A prediction assumes the existence of a cause-effect relation between the measured property and something predicted in the future, even if (as in this case) the mechanism connecting them isn't known.

Prediction implies cause and effect. Exposure to a virus predicts infection in some of those exposed -- it's more than a simple correlation.

Inebriation predicts traffic accidents. Sexual activity predicts pregnancy and STDs. And so forth. These aren't simple correlations.

> You're confusing a "predictive" relationship with a "causative" relationship.

No, I'm asserting that that is what it means, as do all who use the word "prediction" in this context.

[thaumasiotes]

See http://www.thefreedictionary.com/predictor+variable , or http://www.theanalysisfactor.com/the-many-names-of-independe...

Pregnancy predicts sexual activity just as sexual activity predicts pregnancy (actually, better). The reference to the future is that you test the second variable in the future, not that it receives its value in the future.

From the second link:

> Predictor Variable: It does not imply causality. A predictor variable is simply useful for predicting the value of the response variable.

[lutusp]

> Pregnancy predicts sexual activity just as sexual activity predicts pregnancy (actually, better).

This merely says that the use of "predicts" in this context is meaningless, since in common usage "predicts" implies a one-way relationship.

I see from your second link that a definition has been crafted that undermines the word's common meaning. So it goes in language, an art, not a science, and one in which words mean whatever people think they mean, as with the sad case of "literally", which often means "figuratively".

http://www.merriam-webster.com/dictionary/literally

[thaumasiotes]

You're the one who specifically called out the use of 'the word "prediction" in this context'. If you're not familiar with standard statistical terminology, why try to start a fight over its use in statistics?