46
37CENNED, you would definitely be on my short list of candidates for the mother of all statistical astrology studies! 8)

Just a couple of issues, though, re: your observations.

1. "prior probability ? data ? posterior probability." Sadly the views of a lot of people remain unaffected by new data. Oftentimes the new data are ignored, or simply explained away.

2. The results of your moon and soccer study are really interesting. Yet I don't see an explanation for the correlation. And what might it be like? Causal? Synchronicty? An artefact of the data that might evaporate the a larger study? Then astrologically, how do we explain the cases that don't fit the pattern?

47
Hi Waybread

I don't want to speak for my son without speaking to him, so I'll wait until I see him over Christmas. But (so you understand) I don't feel any desire or inclination to justify astrology. I have a personal conviction in astrology (as I understand it and use it) because my experience has shown it to have a multi-faceted value which cannot be doubted. But I also think its most worthwhile principles are heavily dependent upon Platonic philosophy which doesn't stand up well to purely objective analysis. Some elements are probably of more value to science than science is currently able to recognise; but such reliable principles can afford to wait for attitudes to change. So I don't feel any motivation to prove astrology; I'm just happy to have the opportunity to explore it and use it.

But I will report back later,
Deb

48
Deb, thanks for your response. I hope your son is interested! In my case, astrology is my passion, whether or not it actually works.

I would like to distinguish between (1) the natural and physical sciences and (2) the social and behavioural sciences. I am not saying you or anyone on this thread conflates them, but these are very different branches of knowledge with very different subject matters. Any statistical test of astrology's claims would belong in the latter category.

The term "behavioural science" would belong primarily (though not exclusively) to psychology. Psychology and some sub-fields of sociology, I believe, are the types of social/behavioural science most heavily invested in quantitative research methodologies of the sort that would best relate to astrology.

Other branches of the social sciences with some relevance to astrology (such as cultural/social anthropology) are increasingly moving into qualitative methodologies, in which the researchers recognize the Procrustean Bed effect of many quantitative research designs. Even with a survey instrument that has been pre-tested, the study oftentimes doesn't capture enough of the subjects' own experiences.

Coming from a qualitative background myself, this is why I think it would be important for researchers who are statisticians to talk to experienced astrologers about how they actually sit down and read a horoscope. (Books on horoscope interpretation steps should also be consulted.) And then I think a statistical research design could combine the best of both qualitative and quantitative procedures.

But I personally do not feel that simply because I am convinced that astrology "works" that it therefore makes sense to sidestep the issue of why correlations between horoscope placements and subjects' personality traits or life experiences have consistently failed, in study after study. I should think this is a particular problem for astrological techniques that appear to be objective, rather than those based upon some kind of loose pop-psych intuition.

To my way of thinking, the anti-statistics response among astrologers has been mostly one of simply ignoring the more thoughtful critics, rather than facing them openly. [I am not speaking here, of the committed antagonists like Richard Dawkins.]

Worse yet, some astrologers imagine the more outrageous critics, in a leap of logic, to stand in for the entire body of scientists and social scientists. This leads to an anti-science bias among some astrologers that totally belies our reliance upon science in our everyday lives. It also puts astrologers in the bind of effectively saying that the scientific methodologies that allow us to drive a car, take beneficial medications, or send e-mail-- necessarily and inexplicably are worthless insofar as astrology is concerned.

I don't see why. If astrologers argue that (A) there is a strong correlation between certain horoscope placements and human events, but (B) statistics are incapable of demonstrating it; then we are stuck with an exceptionalist argument for astrology--which I have not really seen astrologers make in much depth or persuasiveness, beyond Cornelius, The Moment of Astrology. And then divination of a spiritual sort is not what many astrologers think they are doing. Some even argue for explanations (like gravity) that are fundamentally scientific in character, even where they are highly implausible.

So unless an astrologer has a more spiritual or theological explanation for why astrology "works", I cannot fathom why a better research design, devised by a team of knowledgeable experts of good will, would not be worth conducting. Because otherwise, look what we are left with!

Thankfully, with a visit from your son over the holidays, you are likely to get an open-minded hearing.

49
Hi waybread!
waybread wrote:... Any statistical test of astrology's claims would belong in the latter category.
My argument would be that astrology is more a mathematical system of metaphysics.

One does not assert 2+2=4 because of a statistical inference, but because it is axiomatic true. Astrology's "truth claims" would go into this category.
waybread wrote: But I personally do not feel that simply because I am convinced that astrology "works" that it therefore makes sense to sidestep the issue of why correlations between horoscope placements and subjects' personality traits or life experiences have consistently failed, in study after study. I should think this is a particular problem for astrological techniques that appear to be objective, rather than those based upon some kind of loose pop-psych intuition.
These are often just bad studies, esp since they are being applied to something that statistics simply doesn't relate to.
waybread wrote: ... It also puts astrologers in the bind of effectively saying that the scientific methodologies that allow us to drive a car, take beneficial medications, or send e-mail-- necessarily and inexplicably are worthless insofar as astrology is concerned.
But these aren't "scientific" as much as they are technological. Just as it makes sense to sharpen a rock to make a knife. This doesn't take quantitative or qualitative studies, but a practical calculation, as it were, to determine its effectiveness.
waybread wrote: I don't see why. If astrologers argue that (A) there is a strong correlation between certain horoscope placements and human events, but (B) statistics are incapable of demonstrating it;
It isn't that there are 'strong correlations' to astrological placements, it is that astrological placements are equated to certain outcomes, e.g. Saturn in Mars' signs indicates the predeceasing of siblings, provided of course that Mars gives the topical signification for siblings. This is more akin to geometry, where theorems are deduced from postulates. Again, this is not something that lends itself to statistical analysis.
Gabe

50
In 44 matches the Moon was above the horizon at kick-off and in these games 22 of the countries with a FIFA ranking that was higher than their opponent?s rank won their games (that?s 50%). 10 of the remaining above-horizon games were drawn and in 12 the non-favored country won. These results confirm the prior probability.
I don?t think this is correct. The prior probability that was assumed was not that the favorites would won 50% of the time, but that there were not a difference between the two groups.You are using now the prior probability of the favorites winning a game. The prior probability could be assumed a priori through previous games. For instance, in previous world cups, the favourite rate would be 40% of all games, then we could see the difference with our results. It would be possible, for instance, that the moon above the earth gave no results and the moon below the earth gave concrete results.

I admit that Bayesian analysis is not my forte, however. :lala

I really don?t like the subjective component on bayesian analysis, but my vision of astrology has always been of conditional probability.
Meu blog de astrologia (em portugues) http://yuzuru.wordpress.com
My blog of astrology (in english) http://episthemologie.wordpress.com

51
GR wrote:...
It isn't that there are 'strong correlations' to astrological placements, it is that astrological placements are equated to certain outcomes, e.g. Saturn in Mars' signs indicates the predeceasing of siblings, provided of course that Mars gives the topical signification for siblings. This is more akin to geometry, where theorems are deduced from postulates. Again, this is not something that lends itself to statistical analysis.
My understanding of astrology largely coincides with Gabe's here.

- Ed

52
But these aren't "scientific" as much as they are technological. Just as it makes sense to sharpen a rock to make a knife. This doesn't take quantitative or qualitative studies, but a practical calculation, as it were, to determine its effectiveness.
I think the "technical term" is "Convergent Realism" (but I am not sure it may be pragmatical something). It is the epistemological notion that, if a technology work, then the concepts behind it must be true. This historically is not true, as many technological progresses were made with now discredited science. (There is a paper about it in JSTOR by Larry Laudan but I can?t remember his arguments if my life depended on it)
Meu blog de astrologia (em portugues) http://yuzuru.wordpress.com
My blog of astrology (in english) http://episthemologie.wordpress.com

53
yuzuru wrote:
But these aren't "scientific" as much as they are technological. Just as it makes sense to sharpen a rock to make a knife. This doesn't take quantitative or qualitative studies, but a practical calculation, as it were, to determine its effectiveness.
I think the "technical term" is "Convergent Realism" (but I am not sure it may be pragmatical something). It is the epistemological notion that, if a technology work, then the concepts behind it must be true. This historically is not true, as many technological progresses were made with now discredited science. (There is a paper about it in JSTOR by Larry Laudan but I can?t remember his arguments if my life depended on it)
I certainly do not believe that because a technology is effective or pragmatic in certain circumstances, that its stated conceptual bases are true.

- Ed

54
Great discussion, guys! And here is my take on several of your points.

GR, I wouldn't suggest that astrology's practices or truth-claims are mathematical on the level of 2+2=4. In my universe, 2+2 invariably equals 4, but astrological predictions or analyses often/sometimes are incorrect. [I'll leave the term "often/sometimes" deliberately vague, so that hopefully both the True Believers and the sceptics can take comfort in it.] Statistics (and what do I know?) seem to address relationships that are not invariably true, but that occur by a figure much greater than random chance or at a level found in a control sample of the population.

I totally agree that sometimes there are 'bad studies", if we take 'bad" merely to mean flawed research designs. This is why I would like to see better studies.

Technology of the types I mentioned is based upon science!!!!!! In fact, some scholars prefer to think of "science" and "science applied." Depending upon the technology in question, it is based in subjects like physics, anatomy, and chemistry. I guess we should distinguish between "high tech", like the science and technology used to produce the plastic in my computer keyboard that allows me to type this message; vs. "low tech", which might be sharpening a knife on a whetstone. Note that definitions of both science and technology depend upon the time and place in question. Science today is not the science of yesteryear. Mathematics are important in both science and technology.

GR and Ed: The death clock example precisely would lend itself to statistical analysis! Surely we could come up with a lot of horoscopes of people who have had siblings predecease them, and then compare their Saturn and Mars placements with those in horoscopes of people who have not had siblings predecease them. (That is, for astrologers with the stomach for "death clock" astrology, which I do not have.) If this predictive technique is invariably correct, a good statistical study should show that.

Of course, the operative paradigm or mechanism remains to be explained. It had better not be scientific if one wishes to argue that it is not amenable to research.

Yuzuru, assuming that your result on football match outcomes would still hold in a much larger study, how do you explain it?

Also, I agree that people can invent a technology without understanding the reason why it works. To me a good example would be practical herbalism: traditional herbalists knew or believed that certain herbs cured certain health problems, but later scientists were able to isolate the effective compounds in the plants and explain how they acted on the body.

55
yuzuru wrote:(There is a paper about it in JSTOR by Larry Laudan but I can?t remember his arguments if my life depended on it)
Can you name the title of the paper?

57
Could this be it? Thanks anyway.
A Confutation of Convergent Realism
Author(s): Larry LaudanSource: Philosophy of Science, Vol. 48, No. 1 (Mar., 1981), pp. 19-49.


Yet it is the charlatans that attract the media attention, because they claim ?statistical evidence? of news worthy situations. That said, they won?t (I sincerely hope) be obtaining any statistics tenure at a reputable University
Thanks for your son's input Deb. However I think it's still concentrated around critically considering statistics that are believed to prove something while in statistics on astrological research nothing is found. In the former the question is whether the statistics were performed in a proper way, while the latter hasn't been discussed actually.
This is more akin to geometry, where theorems are deduced from postulates. Again, this is not something that lends itself to statistical analysis.
The problem in astrology is that the applied postulates contradict among astrologers on fundamental issues (e.g. sidereal vs. tropical) and the deduced theorems appear (contrary to the rules) rather as new postulates deduced from postulates than theorems. A theorem must be proved and thus in some way should be repeatable, making it thus testable with statistics. Otherwise, on what basis should the astrological indication be trusted, and before all else how could the meaning have been ascribed to the astrological placement? Analogy? Although useful sometimes for explaining the unknown analogy unfortunately often has proved to be wrong, just like the medieval doctrine of signatures.