Some thoughts on stats as a poor way for thinking about astrology.
What are statistics and why do we give them credence? Statistical inquiry is a method of aggregating data (that is, looking at a massive number of cases at once) and finding common patterns in the aggregate data. When we take many many cases (e.g., billions) and see something occur (let’s call it X) most of the time in those many cases we tend to say that X happens in such cases. We give this form of inquiry credence because of the many times X seems to happen and the very few times X seems to happen. Empirical evidence.
So it’s a form of inquiry by numbers. It’s what we call
quantitative inquiry – and there has been a strong bias towards quantitative data for a long time now exactly because when we quantify things
interpretation doesn’t seem to get in the way. Interpretation is where things get messy. You and I may disagree on how to interpret the color red. What a car means. Whether cigarette smoke smalls good or bad. Etc. But if we stick to the numbers, numbers don’t lie. 5 = 5 and that’s more than 4, which is always more than 3, etc. No interpretation or disagreement needed. So it seems, at first glance, that qualitative inquiry is something we can be more certain about than anything that requires interpretation. (And by the way, let’s call the interpretive parts of a study its “qualitative” parts.)
But what greybeard is pointing out, which is absolutely right and every statistician will tell you this – it’s what you learn in Stats 101 – is that every statistical study is only as good as its
qualitative parts.
What are the qualitative parts of a statistical study?
Everything that is being tested for, in any stats study, has to be
coded for. If the point of stats is to look at a billion cases and find how many times X occurs, than the statistician has to make a decision about what X looks like and how she’s going to find it in each case.
That’s not difficult if X – the thing we’re looking for – is just a physical measurement. Let’s say for example, we want to test whether billiard balls move with force that is equal to the force they’ve been struck with. In stats, we run a billion cases of billiard balls being struck with force A. X = billiard ball moving away from the thing that struck it with the force of A. Every time we see that happen, we tick it off as a “yes,” and then we count the yeses. If the yeses are above 95% or so we say “O.k., we’ve found a statistically significant conclusion that says billiard balls move with a force equal to the force they’ve been struck with. (It’s a little more complicated but that’s the idea).
But what if X is not as simple as a physical measurement? What if the thing we’re looking for hinges on
interpretation? What if we want to prove statistically, for example, that democratizing a nation brings peace. Unlike with the billiard ball example, we have to make some interpretations about what we mean by “democracy” and what we mean by “peace.” I may run that inquiry and find that everywhere we look, wherever a country has been democratized, democracy has brought peace. But then somebody comes along and disagrees with my study. I think I have the weight of science to give credence to my study. After all, the numbers are on my side. In my study every single case of democracy = a more peaceful country. But then it turns out I defined peace very narrowly. Peace in my study = no civil war. So cases like America count as peaceful. "But
are they?" my critics ask. What about violent homicide deaths in America? Poverty? The international wars it engages in? Etc.
A quantitative study is only as good as it’s qualitative aspects.
And this, as I read it, is greybeard’s point.
What is greybeard taking issue with when it comes to this paper? One of the Xs in this paper – one of the things it’s look for in its many cases – is an individual’s career. How does the paper code for that. The 10th house. So this paper assumes that what astrology says is that a person’s career is defined by his 10th house. And so we can look to the 10th house in a billion cases and see how many careers are described by a case’s 10th house.
We must also ask how the paper is coding for how astrology indicates a career. Does Mars in the 10th house in this paper indicate a military career? Saturn in the 10th house = science?
All of these coding choices are interpretations that narrow what astrology is actually doing. Greybeard’s point: Astrology is a complex language of symbols that cannot be understood by breaking up the chart into particular, pinpoint pieces (like for example, career on the 10th). A chart must be read holistically, in a gestalt like fashion (and if anybody has read any of greybeard’s postings we all know this is his thing). Each point in the chart speaks to, and is informed by, the rest of the chart. Yet a statistical approach is going to
demand that we find more narrow ways of coding for indications. We must find something more simple like “Mars in the 10th house = military career” to code as X so that we can sift through a large number of cases and find that specific X.
Notice, importantly, that this is exactly what Gauquelin did – which is why I never understand when astrologers point to him as evidence of astrology. In order to substantiate astrology statistically he wound up throwing out 90% of what astrology is. The interpretation of signs, houses and (for the most part) planets is gone in his stats work. Interpretation made things too tricky. Instead we’re left with a Mars affect and a Saturn effect (and a few others). In distilling astrology into something that might be made statistically verifiable it actually takes all the interpretation out of astrology, which in effect guts astrology.
So, why is statistics not a method well suited for thinking about astrology?
Because astrology is a language that is too complex and essentially interpretive - with it's key symbols (planets, signs, houses) being essentially too "polymorphous," to use Robert hand's word - for stats to be a method for saying anything of genuine insight about astrology.
(Polymorphism: There's a series of articles by Hand where he deals with what he means by this:
http://www.stariq.com/astrologybyhandlib.htm Essentially he means this: "Astrological polymorphism is the fact that astrological indications can be expressed in many possible ways.")