Hi Clelia!
Thanks a lot for your feedback.
Clelia Romano wrote:How much we are expect to obtain to consider the difference valid? If we are talking of the same thing I imagine a differente of 1.8 is big enough, isnt?it?
Yes, it is quite a good result. In Geoffrey Dean's opinion, only 10 or more will do of course!

. But in my studies, if I get something like 2 or 3, I consider it quite a good result. <1 means generally nothing alone.
But the problem is that these values are not fully comparable because we usually get bigger effect sizes (and therefore bigger effect power) when we have lesser options, or a smaller degree of freedom as the statistically correct term goes. For example in the Criminal perpetrators study we were studying planets in angular-succedent-cadent houses, so there was only three possibilities, and when there's a real astrological effect, the effect sizes are easily bigger than in a ?planets in 12 houses? type of study for example.
I looked at your study too, congratulations for that! I second Philip?s opinion that the sample size was too small. Especially when you study the more rare rising signs like Aries, a small deviation in the test group turns easily into a prominent looking percent.
The main problem in how ADB calculates the percents is that it doesn?t take into consideration the sample size at all. If we had a group of 500 something, say dentists, and 250 of them happened to have, say Mars in the 1st house (this is a totally fictional example!), when the expected value would have been 500/12=41,7, that would really be very remarkable and ADB would tell us that the difference is 250/41,7=599%. But if we had instead a group of five dentists and the expected value of having Mars in a certain decanate would be 5/36=0,14, and then one of them would indeed have Mars in that very decanate (which could happen just by accident), ADB would give us an even more impressive percent of 1/0,14=714%. We must have more sophisticated tools for evaluating the real value of a study.
Clelia Romano wrote:Anyway: have you measured in the same way all the other groups as well?
Yes I have. I like this measure because it allows you to add values together. In the last few days I?ve made a similar study with the same 10 samples about 11 different house systems. Guess which house system gave generally the biggest deviations to the random data? Regiomontanus! Maybe Lilly really was onto something when he changed Alcabitius into Regiomontanus!
In that study Regiomonanus beat (in the following order) Campanus, Equal houses, Sripati (Hindu Bhava), Porphyry, Alcabitius, Placidus, sidereal whole signs, Koch, Meridian houses and tropical whole sign houses. Tropical whole signs gave the weakest results meaning their planetary positions differed generally least from the random data groups.
Best, Sari