Meta-Analysis of Astrology Accuracy: What the Research Shows

The short answer

Meta-analyses of astrology accuracy consistently return the same finding: astrology does not outperform chance under controlled conditions. This holds across different traditions, different methodologies, and different decades of research. The effect sizes are negligible. The conclusion is not contested among researchers who have examined the evidence.

That is the honest answer. What follows is the longer one — because what the studies actually tested matters considerably, and the gap between what they measured and what most people use astrology for is larger than either side of the debate usually acknowledges.

What is a meta-analysis?

A meta-analysis pools the results of multiple independent studies on the same question, applies statistical methods to combine their findings, and produces an overall estimate of the effect size. The advantage over a single study is statistical power: small effects that individual studies lack the sample size to detect become visible in aggregate. If astrology had any predictive validity — even a modest one — a well-executed meta-analysis of dozens of studies would find it.

Several research teams have done exactly this. The results are consistent.

Dean and Kelly (2003): the most comprehensive review

The most cited meta-analysis in the astrology accuracy literature is Geoffrey Dean and Ivan Kelly's 2003 review, published in the Journal of Consciousness Studies under the title "Is Astrology Relevant to Consciousness and Psi?" It is not a conventional meta-analysis in the narrow statistical sense — it is a systematic review of over 40 controlled studies that subjected astrological claims to empirical testing.

Dean and Kelly organised the evidence into categories: sun-sign effects on personality, astrological matching (can astrologers identify individuals from birth charts), time twins (do people born at nearly identical times share similar fates?), and comparative studies of astrologers versus chance. The findings across all categories were consistent:

• Sun-sign effects on personality: no reliable correlation. Studies using standardised personality measures (the California Psychological Inventory, the Eysenck Personality Questionnaire, and others) found that personality scores did not differ between sun signs at rates above chance.
• Astrological matching: under blind conditions, astrologers could not match birth charts to personality profiles, life outcomes, or psychological test results at better than chance. This held across multiple independent test designs.
• Time twins: a large-scale study examined outcomes for people born within minutes of each other in London hospitals. Despite sharing nearly identical natal charts, these individuals showed no more similarity in personality, intelligence, or life events than would be expected by chance.
• Astrologers versus chance: when astrologers were asked to predict specific outcomes — medical conditions, professional choice, relationship status — their accuracy did not exceed what a random guess would produce.

Dean and Kelly's conclusion was unambiguous: the evidence does not support astrological claims at any level of testing. Effect sizes across all reviewed studies were indistinguishable from zero.

Eysenck and Nias (1982): the earliest systematic review

Hans Eysenck, one of the twentieth century's most prominent personality psychologists, co-authored Astrology: Science or Superstition? in 1982 with David Nias. The book reviewed the available empirical literature and concluded, with notable exceptions, that controlled research found no support for sun-sign astrology or astrological personality profiles.

The notable exception Eysenck and Nias engaged with — cautiously — was Michel Gauquelin's statistical work on planetary positions at birth, discussed below. They did not endorse Gauquelin's findings as settled, but they considered them more methodologically serious than the sun-sign literature and worth continued investigation. Subsequent research resolved that question in the negative.

For the mainstream astrological claims — personality from sun signs, character from natal charts, prediction from transits — Eysenck and Nias found no reliable evidence. This was a striking conclusion from a psychologist who took the empirical question seriously rather than dismissing it out of hand.

Shawn Carlson (1985): the double-blind experiment

Carlson's 1985 study, published in Nature, is not a meta-analysis — it is a single controlled experiment — but it appears in essentially every meta-analysis of astrology accuracy because its design is considered the most rigorous single test ever conducted.

Carlson recruited 28 astrologers, selected by the National Council for Geocosmic Research as representative of professional competence. He gave each astrologer three real California Psychological Inventory profiles and asked them to identify which one matched a provided natal chart. If astrologers could read charts, they should identify the correct profile at a rate significantly above 33% (one-in-three chance). They did not: the hit rate was approximately 34%, statistically indistinguishable from chance.

Carlson also tested whether CPI profiles predicted astrological self-assessments: they did not. The astrologers participated in designing the study to ensure they considered it a fair test. The study has been questioned on methodological grounds — most notably by Ertel and by Currey — but the critiques have not overturned its central finding, and no subsequent controlled study has demonstrated astrologer accuracy at above-chance levels.

Hartmann, Reuter, and Nyborg (2006): the largest sun-sign dataset

For those who find double-blind studies with small samples unconvincing, Hartmann, Reuter, and Nyborg's 2006 study in Personality and Individual Differences provides a different scale of evidence. The study examined 15,198 individuals, making it the largest dataset ever assembled specifically to test whether sun-sign position correlates with personality.

The researchers tested whether sun signs, time of year of birth, or planetary positions correlated with personality measures and general intelligence. They found no reliable relationship. Effect sizes were negligible across all measures. The study's scale makes the null finding unusually robust: if a sun-sign effect existed at even modest magnitude, a sample of over fifteen thousand would detect it.

The Hartmann et al. study is significant because it removes the sample-size objection. With N = 15,198, a small real effect would be statistically detectable. The absence of a detectable effect is not an absence of data — it is the data.

The Gauquelin anomaly

The one genuine anomaly in the research literature is Michel Gauquelin's work on the "Mars Effect." Gauquelin, a French psychologist and statistician, reported in the 1950s and 1960s that professional sports champions were born disproportionately when Mars was rising or at the upper meridian. His dataset eventually included over 2,000 European sports champions, and the statistical deviation — while small — appeared consistent across multiple samples.

The Mars Effect generated more sustained scientific attention than any other astrological claim because Gauquelin was himself a skeptic of mainstream astrology, his methods were statistically literate, and the anomaly proved surprisingly persistent across several of his own replication attempts.

Subsequent independent replication attempts, however, produced null results. The Committee for the Scientific Investigation of Claims of the Paranormal (CSICOP) conducted a systematic test using a fresh sample of American sports champions; the Mars Effect did not replicate. Belgian and German researchers attempted independent tests with similar outcomes. The current scientific consensus is that the Mars Effect is not a real phenomenon — most likely a product of data selection biases in Gauquelin's original samples combined with the file-drawer problem (negative replications are less likely to be published than positive findings).

The Gauquelin anomaly is worth knowing about because it is sometimes cited as evidence that "a meta-analysis found positive results for astrology." That framing is misleading. The anomaly did not survive independent replication, and it concerned only one narrow claim — a statistical correlation between a specific planetary position and athletic achievement — rather than the broader astrological system of personality typing, prediction, or timing guidance that most people mean when they ask whether astrology works.

What the meta-analyses did not test

Here is the gap that neither skeptics nor astrology advocates discuss clearly enough: every major meta-analysis of astrology accuracy tested either personality matching or single-event prediction. These are not the primary ways most people actually use astrology.

Carlson's astrologers were asked to match a chart to a personality profile. Dean and Kelly's time-twin study compared people born minutes apart to see if they shared similar life outcomes. Hartmann's dataset correlated sun signs with personality questionnaire scores. These are coherent empirical questions. They are also not the question that drives most actual astrological practice.

In everyday use, astrology functions as a timing framework. The question is not "given your birth chart, what is your personality type?" — it is "given your current transits and planetary cycles, is this a good week for the conversation you need to have, the contract you need to sign, or the project you want to launch?" These are iterative, personal, outcome-tracked questions. The astrologer offers timing guidance; the person acts; the outcome is observed; the guidance is either confirmed or disconfirmed by what happened. Over months and years, patterns either emerge or they do not.

No meta-analysis in the literature has tested this use pattern. The reason is methodological: testing iterative, personalised timing guidance across individual decision cycles requires a different infrastructure than a double-blind group study. It requires a tool that records both the guidance and the outcomes, over a long enough period to generate meaningful aggregate data.

That infrastructure is what Zodiacs was designed to build. Not to validate astrology in the abstract — the meta-analytic literature has given a clear answer on several abstract questions — but to run the experiment the literature has not run: does personalised timing guidance, tracked honestly over many decisions, correlate with better outcomes for the individual who follows it?

What this means for how to use the research

The honest summary: if someone asks whether meta-analyses support astrology as a system for identifying personality from birth data or predicting events, the answer is no. The research is consistent, the effect sizes are negligible, and no serious researcher has argued otherwise in decades.

If someone asks whether meta-analyses have tested the efficacy of personalised astrological timing guidance tracked over individual decision cycles, the answer is: they have not. The tools to do so did not exist until recently.

These are different questions. Conflating them produces confusion in both directions — skeptics who dismiss all astrological use as refuted by Carlson, and practitioners who cite unrelated findings to sidestep the Carlson result entirely.

What the research establishes is a floor: astrology is not a validated predictive system for personality or single-event outcomes. What it leaves open is a different question — one about iterative timing, individual feedback, and long-run pattern recognition that has not yet been studied at scale. The former is settled. The latter is the experiment still running.

Track your own outcomes — free to start