A recent meeting of the Research Working Group at the Cosmobiology Institute led to a very interesting discussion.
The topic was the future direction of scientific research in astrology. Today, many researchers are doing exactly what modern science expects: collecting large datasets, applying rigorous statistical methods, reporting p-values, effect sizes, statistical power, and correcting for multiple testing. This is a significant step forward compared to the anecdotal evidence that has dominated astrological research for decades.
Some of these studies are genuinely impressive. They report highly significant p-values, adequate statistical power (often above 80%), and reproducible, albeit sometimes small, effect sizes. Such results certainly deserve attention.
However, I believe we should be careful not to overstate what these findings actually demonstrate.
[toc]
Currest State
At present, these studies establish statistical associations between planetary configurations and terrestrial events. They show that the observed relationships are unlikely to be explained by random chance alone. But they do not yet establish why these associations exist.
During the discussion, I suggested that the next stage of astrological research should focus on developing a fundamental causal model—a model that proposes a concrete mechanism through which astrological factors might influence observable events, and that generates new, testable predictions.
This proposal was met with some skepticism. Several colleagues argued that such a model is either unnecessary or premature, and that accumulating more statistical evidence should remain the primary objective.
I respectfully disagree.
In my view, statistical correlations are an essential first step, but they cannot be the final destination. Without a causal framework, even the most rigorous statistical findings remain vulnerable to alternative explanations, such as temporal confounding, and provide little guidance about which astrological principles reflect genuine underlying processes and which do not.
The purpose of this article is not to criticize the current statistical work—quite the opposite. I believe it forms an indispensable foundation. My argument is simply that the next major advance will come not from discovering ever more correlations, but from developing and testing causal models capable of explaining those correlations.
Why Causal Model is Important?
My concern is that many researchers seem to treat this step as the final goal, while I believe it should be the beginning of a much larger scientific program.
The reason is simple: correlation alone does not tell us why a relationship exists.
Imagine that we repeatedly find statistically significant correlations between planetary configurations and terrestrial events. The studies have high statistical power (>80%), very small p-values, and survive corrections for multiple testing. That is certainly interesting. But does it necessarily mean that planets influence events on Earth?
Not yet.
There is another explanation that must be taken seriously: temporal confounding.
Temporal Confounding
Planetary positions are, fundamentally, functions of time. Many terrestrial processes are also functions of time. Economic cycles, social behavior, climate, epidemics, technological development, and countless other phenomena change over time. Therefore, it is entirely possible that both planetary configurations and observed events are independently correlated with time, creating an indirect association between them.
The statistical correlation itself cannot distinguish between these competing explanations.
It is compatible with at least several causal models:
- planets influence events;
- time influences both planetary positions and the observed events;
- some unknown third factor influences both;
- or the observed correlation is simply a statistical artifact.
Exactly the same dataset can support all of these possibilities. Correlation alone cannot tell us which one is correct.
This is why I believe the next step should not be searching for more correlations, but developing a fundamental causal model of astrological influence.
Explanation
A scientific theory should not merely describe patterns. It should explain them.
Suppose, for example, we hypothesize that a planet's altitude above the horizon modifies the strength of its influence. That immediately generates a series of concrete, testable predictions. We would expect house position to matter. We could compare cases where zodiacal longitude is nearly identical but altitude above the horizon differs substantially. We could investigate whether the proposed effect changes continuously with altitude.
Notice what has happened. We have moved from asking, "Is there a correlation?" to asking, "Does reality behave as this mechanism predicts?"
That is a much stronger scientific question.
Quicker Search
A causal model also dramatically reduces the search space.
Without a model, virtually every astrological technique becomes a candidate for investigation: signs, houses, aspects, midpoints, harmonics, fixed stars, asteroids, different house systems, Arabic Parts, and so on. The number of possible variables quickly becomes enormous. The research strategy naturally turns into searching through thousands—or millions—of possible combinations, hoping that some of them produce statistically significant results.
With a model, many of these possibilities can be excluded before any statistical analysis begins.
If the proposed mechanism depends on altitude, then house position becomes scientifically relevant, while midpoint structures may become secondary or even irrelevant. If the mechanism depends on angular separation, then aspects become central. The theory tells us what should matter before we analyze the data.
This is exactly how mature sciences operate.
Predictions
Another important point is that good scientific theories generate new predictions.
A correlation explains existing data. A mechanism predicts observations that have not yet been made.
The history of science is full of examples. Kepler described planetary motion remarkably well, but Newton transformed those descriptions into a theory by proposing a mechanism. Mendel described inheritance patterns, but molecular genetics eventually explained why those patterns existed. Correlations often come first—but they are rarely the end of the scientific process.
Scientific Community Expectation
There is also a practical reason for moving in this direction.
The broader scientific community is unlikely to take astrological research seriously if it consists primarily of increasingly sophisticated correlation studies while the possibility of temporal confounding remains unresolved. Researchers from statistics, causal inference, physics, neuroscience, or systems science are much more likely to engage if they see an attempt to formulate explicit causal hypotheses that can potentially be falsified.
Finally, I think it is worth remembering one fundamental distinction.
- Correlation answers the question: "Are these variables associated?"
- Science ultimately wants to answer a different question: "Why are they associated?"
Until we can propose—and rigorously test—a plausible causal mechanism, every observed correlation remains compatible with multiple competing explanations. Statistical evidence is necessary, but it is not sufficient.
For that reason, I believe the future of astrological research lies not in discovering ever more correlations, but in building and testing explicit models of the underlying process. Once we have such models, statistics becomes far more powerful, because it is no longer searching blindly—it is evaluating specific theoretical predictions.
To me, that is the path that has the greatest chance of turning astrology from a collection of intriguing statistical observations into a genuine scientific research program.
