Is It All Written in the Stars? A Methodological Commentary on Sachs's Astrology Monograph and Re-Analyses of His Crime-Statistics Data

Writing in 2003 (Volume 45, Issue 1), Alexander von Eye, Friedrich Lösel, and Roni Mayzer took up a question that most statisticians would consider settled: can the sign you were born under predict how you behave? The occasion was a 1999 book by Sachs titled Die Akte Astrologie, or The Astrology Dossier, which argued that genuine, statistically significant links tie the twelve signs of the Zodiac to patterns of human behavior.

Sachs built his case on contingency tables. He crossed each Zodiac sign against a long list of behavior categories pulled from official Swiss registers, then pointed to the spots where particular signs lined up with particular behaviors more often than you would expect by chance. Those local associations were his evidence. Among the many examples, the authors chose to examine the one Sachs treated as a strong and clear case: the connection between the signs and crime.

On the surface Sachs did not hide his reasoning. The authors note that his methods are fairly easy to follow. The problem lies underneath. When you run a large number of statistical tests across the same body of data and go looking for the comparisons that happen to come out significant, some of them will, purely by accident. That is capitalizing on chance. Sachs's approach also ignored that the many tests he performed on one dataset are not independent of each other; they share the same underlying numbers, so treating each result as a fresh, standalone finding inflates the apparent signal.

To test whether his conclusions held up, von Eye and his colleagues went back to Sachs's own figures on criminal convictions and birth signs and reanalyzed them with two different tools. The first was Configural Frequency Analysis, a method designed precisely for spotting which cells in a table deviate from expectation while keeping the multiple-comparison problem in view. The second was a two-way cluster analysis. Neither reproduced the relationship Sachs described.

The verdict is blunt: the claims in the monograph cannot be sustained. What looked like a pattern connecting the Zodiac to crime dissolves once the analysis accounts for how many tests were run and how those tests lean on one another. The paper closes by weighing both the statistical findings and what they mean for the substance of the astrology question itself.

Keywords: Zodiac, astrology, crime, human behavior, Configural Frequency Analysis. One small flourish the authors allowed themselves, listing their own birth signs beside their names (Virgo, Leo, Capricorn), reads as a quiet joke at the expense of the subject matter.