A Framework Built to Be Tested Against
The Big Five — openness, conscientiousness, extraversion, agreeableness, and neuroticism, often abbreviated OCEAN — emerged from decades of factor-analytic research into the structure of personality, with major contributions from researchers including Lewis Goldberg, Robert McCrae, and Paul Costa from the 1980s onward. Unlike many personality frameworks, the Big Five wasn’t designed around a particular theory of what personality should look like. It was derived empirically: researchers analyzed huge numbers of personality-descriptive words and ratings, looking for the underlying dimensions that best explained how those descriptions clustered together. Five dimensions kept emerging, robustly, across different languages, cultures, and methods of measurement.
This makes the Big Five an unusually good instrument for testing astrological claims, for a specific reason: it’s measured the same way regardless of what hypothesis is being tested. A researcher administering a Big Five inventory isn’t asking questions designed around astrological categories — they’re using a general-purpose personality measure that has been validated against a huge range of outcomes (job performance, relationship satisfaction, health behaviors, political attitudes) entirely independent of astrology. If sun sign, or any other astrological factor, correlates with personality in the way astrology claims, the Big Five should be able to detect it — it’s sensitive enough to detect plenty of other things.
What the Studies Actually Did
The basic study design, replicated many times since the 1970s, is straightforward: administer a Big Five personality inventory to a large sample, record each participant’s birth date (and sometimes birth time and location, for studies testing more than just sun sign), and check whether any of the five dimensions correlate with sun sign, or with other astrological factors, at rates exceeding what chance would predict.
Some of the largest such studies have used samples in the tens of thousands — large enough that even a very small real effect (a correlation of, say, 0.05 — tiny, but real) would likely reach statistical significance, simply because of the sample size. This is an important methodological point: with a large enough sample, you don’t need a strong effect for it to be detectable. The studies in this literature have generally had more than enough statistical power to detect even quite small effects, if those effects existed.
The Results, Consistently
The accumulated finding across this research program is about as close to a flat null result as social science produces. Correlations between sun sign and any of the Big Five dimensions, in large samples, cluster around zero — typically smaller than 0.05, which in practical terms means sun sign explains a negligible fraction of one percent of the variance in any personality dimension.
For context, it’s worth knowing what does show up in Big Five research at this kind of scale, for comparison. Birth order shows small but detectable effects on some personality measures in some studies (though birth order research itself has faced replication challenges). Cultural and national differences show up clearly. Even relatively subtle factors — certain early childhood experiences, for instance — can produce detectable, if small, effects in samples of this size. Sun sign doesn’t. The null result isn’t “we couldn’t find an effect because our methods weren’t sensitive enough” — the methods are demonstrably sensitive enough to find other small effects of comparable hypothetical size. The null result is “we looked, with tools capable of finding small effects, and there’s nothing there.”
Beyond Sun Sign: Full Chart Studies
A reasonable objection to sun-sign-only studies is that sun sign is the crudest possible astrological variable — a single factor out of dozens that a full natal chart contains, and one that astrologers themselves often describe as relatively minor compared to rising sign, moon sign, and the overall configuration of aspects.
This objection has also been tested, though less extensively than sun sign alone, because full chart analysis requires more data (accurate birth time and location, not just birth date) and more complex statistical handling (a full chart contains dozens of correlated variables, raising the multiple-comparisons issues discussed elsewhere in this series). The studies that have attempted this — comparing Big Five profiles against broader sets of chart factors, including rising signs, moon signs, and aspect patterns — have generally found the same pattern as the sun-sign-only research: no detectable signal exceeding what would be expected from chance, given how many comparisons were made.
This is a less clean result than the sun-sign findings, partly because “full chart astrology” isn’t a single, standardized hypothesis the way “sun sign correlates with personality” is — different astrological traditions and individual practitioners weight chart factors differently, and a study has to make choices about which factors to test and how to combine them, choices that aren’t dictated by astrology itself in any uniform way. But within the studies that have been conducted, the broader pattern holds: more chart complexity hasn’t produced the signal that sun sign alone didn’t show.
The Self-Report Problem
One persistent methodological wrinkle in this literature concerns who is reporting personality, and whether they know their own astrological information.
Several studies have found a curious pattern: when people are asked to rate how well their sun sign’s traditional personality description fits them, people who know astrology and know their own sign tend to rate the fit as higher than people who don’t follow astrology — even though, when their actual measured personality (via Big Five or similar instruments) is compared to their sun sign’s traditional description, there’s no difference in fit between the groups. The people who believe their sign fits them aren’t measurably more “sign-typical” in their actual personalities. They just perceive a better fit.
This is consistent with the self-fulfilling and self-concept mechanisms discussed elsewhere in this series — people who’ve absorbed a sun sign identity may describe themselves in ways that align with it, on self-report measures, even when more behaviorally-grounded measures don’t show a corresponding difference. It’s also a caution about study design: a study that relies entirely on self-report, especially self-report of how well a description “fits,” is vulnerable to this effect in a way that a study comparing independently-measured personality traits against astrological factors isn’t.
What a Positive Result Would Need to Look Like
It’s worth being precise about what kind of finding would actually constitute evidence for astrology’s personality claims, given the Big Five literature’s null results, because “no effect found yet” is compatible with several different underlying realities, not all of which are “astrology is false.”
A positive result would need to show: a correlation between some astrological factor and some Big Five (or comparably validated) personality dimension, of a magnitude detectable in samples of the size already studied (tens of thousands), that replicates across independent samples and research groups, and that corresponds — at least roughly — to what astrological theory actually predicts (a Mars-Saturn opposition correlating with something related to conscientiousness or its absence, say, rather than an arbitrary correlation that happens to reach significance but bears no relation to traditional astrological meaning).
No finding meeting this description has emerged from fifty years of this research program. This doesn’t logically prove that no such relationship could ever exist — absence of evidence isn’t proof of absence, and there’s always a hypothetical effect too small for any study yet conducted to detect. But it does mean that the relationship, if it exists, would have to be smaller than effects this research program has repeatedly demonstrated the capacity to detect — smaller, for instance, than the effect of birth order, which is itself a famously small and contested effect.
What’s Left to Test
The Big Five literature has been most thorough on exactly the kind of claim it’s best suited to test: broad, stable personality traits, measured the same way regardless of hypothesis, correlated against astrological factors in large samples. What it hasn’t addressed, and probably can’t, are the more dynamic claims some traditions make — claims about timing, about how traits express differently across life stages or circumstances, about the kind of “potential” language discussed in the philosophy of determinism elsewhere in this series.
These claims are harder to test partly because they’re harder to specify precisely enough to test — which is itself informative. A claim that can be operationalized into “does X correlate with Y in a large sample” and then tested has, in astrology’s case, mostly come back null. The claims that remain untested tend to be the ones that resist this kind of operationalization in the first place — not because they’re more sophisticated, necessarily, but because they’re less specific about what, exactly, they predict.
Fifty years and tens of thousands of participants have produced a clear answer to one question and left several other questions essentially where they started — not because the research was inadequate, but because those other questions were never quite stated precisely enough for research to address them.