Scientist Research Methodology
Dear friends,
I am very pleased to participate in this forum and to share with you a preview of some results obtained, prior to their formal publication in a scientific repository. This work addresses the study of synastries and has followed a rigorous methodology, comparable to that employed in scientific publications in disciplines such as physics, medicine, or psychology.
In a first phase, an exploratory statistical analysis was conducted on an initial cohort of approximately 350 couples, with the aim of identifying which planetary aspects in synastry appeared with a frequency that clearly exceeded what would be expected by chance. Among about 180 aspects studied, Venus in trine to the Descendant and Sun in trine to the Descendant stood out significantly, both with more than 4 sigma significance. This revealing and pioneering finding was made possible thanks to the generous collaboration of the Fundación Centro Astrológico de Buenos Aires (CABA), which shared its data with me. Subsequently, I was able to independently replicate their results using alternative algorithms.
However, from a strictly scientific perspective —as a physicist or medical researcher would apply it— it is essential to consider two crucial aspects:
1. With a cohort of 350 subjects, barely twice the number of aspects analyzed, the result cannot be considered statistically significant due to the so-called "look-elsewhere effect," well known in scientific research. The apparent 4 sigmas are drastically reduced when appropriate corrections are applied.
2. Using the same sample both to generate a hypothesis and then to confirm it is not methodologically correct nor scientifically acceptable. This is a common error that invalidates the robustness of the finding.
Therefore, this analysis must be classified as **exploratory**: its function is to generate hypotheses that must later be confirmed or refuted using independent, completely virgin databases that have not been previously used.
Following this premise, a new database consisting of approximately 2,000 couples was compiled. This sample was divided into two segments: 60% (1,200 couples, equivalent to 2,400 AA/A-rated Rodden-quality horoscopes) was allocated for exploration and refinement of the hypothesis; the remaining 40% (800 couples) was kept completely virgin and reserved for future independent confirmation.
When we verified whether, in our exploratory database of 1,200 couples, the Sun-Descendant and Venus-Descendant trines remained significant, we confirmed that they did. However, since there was no prior prepublication documenting these findings, we decided to use this new database also as an exploratory tool to refine the hypothesis. Thus, we constructed a combined statistic integrating both aspects (Venus and Sun), set the optimal orb at 6°30' (obtained from the exploratory analysis with the 1,200 couples), and defined the null model and the statistical analysis software.
From that point on, following the standards of the most rigorous scientific practice, the project was preregistered in OSF, where the following were documented: the definitive hypothesis, the software to be used for confirmation, and the digital fingerprint (hash) of the data files corresponding to both the exploratory (1,200) and confirmatory (800) samples. This procedure is what lends seriousness and reliability to a project before the scientific community.
You can consult the preregistered hypothesis at the following link:
[https://doi.org/10.17605/OSF.IO/7U23H](https://doi.org/10.17605/OSF.IO/7U23H)
We didn't know the results till the moment we made the confirmatory posterior analysis. These results will be published soon...😜
I read it, it's brilliant, thank you for sharing. However, I have one comment. Which should be considered before publishing. And I know you mentioned that this is a preliminary test result. However, the astrology system allows for the writing down of all the variation possibilities. And the null model in the study only includes the studied set, if I managed to decipher it correctly. This is unfortunately misleading. Because if I only look at the configurations that occur in a studied set, and the measurement is made in these, the result can be distorted. However, if the set to be examined were examined in a more stable null model that truly covers every variation possibility, i.e. in all the variation possibilities in the entire set, then the values would probably show something different. Compared to other scientific measurements, astrology has the advantage that it can write down all the variation possibilities regardless of the sample, since it is a coherent mathematical system. Moreover, it is circular. While psychology and medicine only draw from occurrences. In summary, the argument can be made that the current null model is strongly conditioned on the properties of its own database. If all variation possibilities were indeed examined (unlimited random matching), then the "excess" would almost certainly not be significant, or would be much weaker.