I think that this subject has not been traited correctly by the scientific and academic communities. A phenomenon has been claimed, and most scientists in the world ignored it, silently delegating to skeptic groups who betrayed the most elementary exigences of science like honesty or rigor.
Finally, the question has not been studied seriously. I limited my study to the Gauquelin / Ertel / Skeptics period (1955 - 1990s), although other searchers claim to have also observed statistical effects. To summarize, my temporary answer is : I don't know ; this opinion is justified below.
To try to find an answer, I worked in two directions :
- I started to work on restoration of historical data, in order to be able to reproduce the original tests. This is done with a program, Gauquelin5, published on github.com.
I started to study the history related to this topic.
The first surprise was to find out that wikipedia contains almost no information about this subject. This is detailed in this page about wikipedia. Some information is available online, but some important sources of information are not present. This led me to adhere to the "Astrological Association" to retrieve the past issues of Correlation journal, to buy a CD containing archives of The Skeptical Inquier journal, and buy several paper books.
Do these tests show anomalies ?The statistical tests on astrology done by Michel et Françoise Gauquelin followed three main directions :
- Professional groups
- Heredity experiments
- Character traits
Effects were observed for Mars, Saturn
Yes ?Some arguments speak in favour of the existence of statistical anomalies :
- Effects have been observed by Gauquelin, replicated by Gauquelin, by the first group of skeptics (Comité Para, 1976, Belgium) and by german scholars (Suitbert Ertel and Arno Müller).
- The computations done by Gauquelin were recognized as correct, even by the most critic skeptics.
A very strong suspicion of fraud exist concerning Paul Kurtz, who possibly manipulated the data to make the effect disappear during the test done by the second group of skeptics (CSICOP, 1979, USA).
Fraud documentation comes from 3 sources :
- Patrick Curry, Research on the mars effect, a detailed investigation by an historian of science
- Dennis Rawlins' Starbaby
Suitbert Ertel analysis with citation count.
Kurtz gathered the data in 3 "canvasses". Ertel shows how the proportion of famous sportsmen decrease in the 2nd and 3rd canvasses and how the percentage of athletes with mars in Gauquelin sectors decrease from 19.2 to 12.1 and 7.3% to finally show no mars effect.
He also shows that the proportion of athletes using Gauquelin 1979 US replication on 432 sportsmen to compare the structure of data.
- For the third (and last) test done by a skeptic group (CFEPP, 1996, France), the protocol signed by Gauquelin and CFEPP was not respected for the most important point : data selection was opaque, done without any external control. I have read and re-read the protocol (I have the original publication in french) ; if what I have read about the way this test was conducted is true, this experiment is a clear example of scientific cheat and should be firmly condemned by the scientific community.
- Suibert Ertel provided an independant confirmation of the effect using citation count. He counted the number of times a sportsmen is cited in a pre-defined list of books, thus obtaining a objective (though imperfect) measurement of eminence. Using all data gathered by Gauquelin and skeptics (4387), he observed that the statistical effects augments with eminence.
No ?But other arguments oblige to be more cautious :
Ertel inspected Gauquelin data in 1988 ; he could access to data not published by Gauquelin (less eminent persons), and showed a selection bias using citation count. Skeptics wonder if selection bias exists in all Gauquelin work. I think that this is a legitimate and serious question. Gauquelin was extremely precise and rigorous, I personnaly believe that he was honestly trying to select the most famous persons.
But the fact that he knew the position of mars before deciding if a person was eminent enough to be included in a group, and the lack of objective criterium to estimate eminence can lead to selection bias. If this is the case, this would invalidate both Gauquelin tests and the Comité Para skeptic test of 1976 (because Gauquelin helped them to bulild their group).
- Ertel in 1987 reproduced eminence grading on painters and musicians and did not reproduce a clear relation between eminence and importance of the effect.
- Ertel has been criticized by skeptics during the 1990s (in particular by Jan Nienhuys) ; to know if these critics are meaningful, a precise examination of the arguments need to be done. I have not studied the details, so I don't know. See Ertel - Nienhuys arguments.