I wrote an article doing a comparative study between statistical data results and traditional delineation.
92 cases of young people deceased before 29 years old as consequence of an accident are compared with 920 cases of the control group.
I used the tropical zodiac and I compared percentages for the sign position of the planets, their dispositors?s position, the position of the rulers of the Ascendant, MC and the other house?s rulers and finally the Modes and Elements showing highest frequency in the Experimental Group.
The conclusions pointed out for a high percentage of accidents in some configurations showed through Tables that I construct in order to facilitate reasoning.
At last I used the case of a client who was involved in a car crash and got seriously injured some years ago. I compared the traditional delineation with the numbers that the statistical study suggested and the conclusions were interesting.
We saw that a methodical analysis of the chart in the traditional way gave a similar result, but the statistical study brought some other factors to take into account : if the astrologer is confronted with a large amount of planets in fixed signs, airy signs, and if Venus is in Taurus, we can put aside the possibility of a fatal accident before age 29.
The article is on line at: http://www.astrologiahumana.com/Researc ... rology.pdf

Clelia

Hello Clelia!

Thank you for presenting your research.

I have some questions and (constructively intended) criticisms for you before I feel qualified to comment further.

1) What exactly do the numbers in the first table mean? I feel this table could be labelled more clearly.

You state '- = less incidence; + = more incidence'.

I presume you mean that positive numbers indicate a greater incidence of fatal accidents in the accident group than in the control group among people with each of the placements concerned.

But how have these differences been quantified?

It cannot be that you are indicating that that there were 122.2 more people with the Ascendant in Aries in the accident group than outside it, since the total size of the accident group is 92, and in any case a mere numerical count would make for poor data when the control group is ten times the size of the accident group.

Nor can it be that there are 122.2 times as many people with the Ascendant in Aries in the accident group than outside it, for similar reasons.

So I presume that you (or the software you use, as the case may be) have taken zero as representing parity on some undisclosed scale and then assigned points to a certain level of difference in results according to the terms of the scale, but without knowing what those terms are I have no idea how great or small the differences are.

Perhaps (I speculate) the scale is designed in such a way that the figure for the Ascendant in Aries means that there are 122.2% more people (after norming the number of people to a percentage of the total group size in both the accident group and the control group, since the control group in purely numerical terms is ten times as large) with the Ascendant in Aries in the accident group than outside it.

If so, for a given (measured) percentage of people in the control group having the Ascendant in Aries, the equivalent percentage in the accident group would be just over (11/5) times as high, since an increase of 120% represents a multiplication of 11/5, and an increase of 122.2% is slightly higher than this. Thus, if we are to suppose for the sake of example that the percentage in the control group is 4% (taking somewhat into consideration the fact that Aries is a sign of short ascension), the percentage in the accident group would in this case be (4% * 222.2/100) 9.8%.

Likewise, by this interpretation of the figures, for a given (measured) percentage of people in the control group having the Ascendant in Aquarius, there would be just over third the number in the accident group, since whereas a reduction between the percentage in the control group and that in the accident group of 190% would lead to the percentage in the control group being, nonsensically, a negative (minus) value, the figure could theorietically make sense as an indication instead of a gain of 190% between the percentage in the accident group and that in the control group. In this case, for example, if the accident group had 2% of its 92 constituent people with the Ascendant in Aries, the control group must have (2% * 290/100) 5.8% of its 920 constituent people with the same placement. [Of course, because the size of the accident group is 92, the figure of 2% in the accident group that I employed purely for the sake of this worked example is in fact impossible in practice - the nearest it could be is [(2/92)*100] 2.16%.]

I think having gone this far I'd better stop working through the percentage increase / decrease scenario as a speculated explanation for what the figures in the table actually mean until notified of the reality of the situation!

However, I feel the need to report that I have some moderate statistical misgivings at the outset about the final column and the general conclusions related to planets that you infer from it.

You write, in inference from your final column: 'there are more protective and less protective planets'.

On inspecting the results for the 12 placements of the Moon, I see that the results in five of the 12 signs, namely Aries, Taurus, Gemini, Virgo and Capricorn show that there are very significantly more fatal accidents where the Moon is placed in those planets than shown in the control group. Yet for Cancer, Libra, Scorpio, Sagittarius, Aquarius and Pisces Moons there are fewer than in the control group, and in Libra, Aquarius and Pisces the most extreme differences are noticed.

I would conclude from this not that the Moon has a protective effect overall, but that the Moon in three particular signs (providing that the study would still bear out with a much larger accident sample size, since 92 is unfortunately unlikely to be deemed adequate when divided by 12 signs to provide for conclusive accuracy) appears to have a very strongly protective effect; in three others appears to have a slightly protective effect, and in five others appears to have a slightly harmful effect, on the risk of fatal accidents before the age of 29.

I feel that to simply add up the results for a particular planet in all the signs is a mistake since it tends to gloss over the very different effects of that planet in the different signs, and also for another reason, on statistical principle, as I'll go on to try my best to explain briefly:

If as I am presuming in the worked examples given previously, the figures in the hundreds and tens and units represent the percentage of difference between the percentage incidences of particular sign placements for particular planets in the accident and control groups, then a reported negative difference of 300% in one may represent the difference between 2 individuals in the accident group and 6 in the control group, while a reported positive difference of 66% may represent the difference between 3 individuals in the accident group and 5 in the control group. Between these two scenarios, while the numbers look very different, there is only a very slight (unitary, in fact) difference between the actual numbers in the accident group and control group, differences that could be explained purely by chance, and even if not, differences that are small. So to add a result of 300% to one of 66% and claim a figure of 366 could be very misleading. These percentages of difference do not meaningfully lend themselves to addition, full stop. And if you then go on and use the overall additions on one side and subtractions on the other to try to demonstrate that a particular planet is protective of harmful overall, it makes less sense still.

I have to cut this short but hope you don't take this post the wrong way - I am all in favour of statistical astrological research but am trying to help ensure that it is both methodologically and analytically rigorous, as well as being clearly labelled and presented.

Hope to hear back from you later,

Philip

Hello Clelia!

Hi Philip:

Thank you for presenting your research.

And thank you for reading it and take the time to comment it!

I have some questions and (constructively intended) criticisms for you before I feel qualified to comment further.

1) What exactly do the numbers in the first table mean? I feel this table could be abeled more clearly.

You state ?- = less incidence; + = more incidence?.

I presume you mean that positive numbers indicate a greater incidence of fatal accidents in the accident group than in the control group among people with each of the placements concerned.

But how have these differences been quantified?


I used the Astrodatabank data. The Factor Analysis Report appeared in form way and with this kind of sign, minus or more, after the number of each factor.
Comparing the results of the Experimental group with the control group I saw that some factors appeared in the Experimentation group 3 times more. I gave credit to this factor incidence.
Anyway, I put all the incidences as the software gave to me: even the not important scores.I copied the data putting them into tables.

It cannot be that you are indicating that that there were 122.2 more people with the Ascendant in Aries in the accident group than outside it, since the total size of the accident group is 92, and in any case a mere numerical count would make for poor data when the control group is ten times the size of the accident group.


I?m sorry that the thing didn?t become clear! 122.2 was the percentage, not the total of people with problems!
It means, according to the software tutorial, that we have something like 1.22 ( meaning that the double more 22 percent more people appeared with the ASC in Aries). But this is not the more important number in the study!


Nor can it be that there are 122.2 times as many people with the Ascendant in Aries in the accident group than outside it, for similar reasons.

So I presume that you (or the software you use, as the case may be) have taken zero as representing parity on some undisclosed scale and then assigned points to a certain level of difference in results according to the terms of the scale, but without knowing what those terms are I have no idea how great or small the differences are.


I was able to chose giving one or more points to some incidence: I chose to give one point to the presence of a planet in a sign, as well as for any other configuration.

Perhaps (I speculate) the scale is designed in such a way that the figure for the Ascendant in Aries means that there are 122.2% more people (after norming the number of people to a percentage of the total group size in both the accident group and the control group, since the control group in purely numerical terms is ten times as large) with the Ascendant in Aries in the accident group than outside it.
Yes, so you understood what I meant!


If so, for a given (measured) percentage of people in the control group having the Ascendant in Aries, the equivalent percentage in the accident group would be just over (11/5) times as high, since an increase of 120% represents a multiplication of 11/5, and an increase of 122.2% is slightly higher than this. Thus, if we are to suppose for the sake of example that the percentage in the control group is 4% (taking somewhat into consideration the fact that Aries is a sign of short ascension), the percentage in the accident group would in this case be (4% * 222.2/100) 9.8%.


Maybe something like that, or more, because the software takes into account the fact of some numerical occurrences are more frequent than others, so must have a lower differentiated score being Aries for example a sign of short ascension. I noted that some times a high numerical difference is considered less important(received a not so big percentage) than another one, lower.

Likewise, by this interpretation of the figures, for a given (measured) percentage of people in the control group having the Ascendant in Aquarius, there would be just over third the number in the accident group, since whereas a reduction between the percentage in the control group and that in the accident group of 190% would lead to the percentage in the control group being, nonsensically, a negative (minus) value, the figure could theorietically make sense as an indication instead of a gain of 190% between the percentage in the accident group and that in the control group. In this case, for example, if the accident group had 2% of its 92 constituent people with the Ascendant in Aries, the control group must have (2% * 290/100) 5.8% of its 920 constituent people with the same placement. [Of course, because the size of the accident group is 92, the figure of 2% in the accident group that I employed purely for the sake of this worked example is in fact impossible in practice - the nearest it could be is [(2/92)*100] 2.16%.]

I think having gone this far I'd better stop working through the percentage increase / decrease scenario as a speculated explanation for what the figures in the table actually mean until notified of the reality of the situation!

I don?t know if I got you now: I had the impression you was trying to cross data, and I didn?t go so far in the article. I think, though, that data can be summed, doing a table of probabilities for determined person.

However, I feel the need to report that I have some moderate statistical misgivings at the outset about the final column and the general conclusions related to planets that you infer from it.

You write, in inference from your final column: 'there are more protective and less protective planets'.

On inspecting the results for the 12 placements of the Moon, I see that the results in five of the 12 signs, namely Aries, Taurus, Gemini, Virgo and Capricorn show that there are very significantly more fatal accidents where the Moon is placed in those planets than shown in the control group. Yet for Cancer, Libra, Scorpio, Sagittarius, Aquarius and Pisces Moons there are fewer than in the control group, and in Libra, Aquarius and Pisces the most extreme differences are noticed.

I would conclude from this not that the Moon has a protective effect overall, but that the Moon in three particular signs (providing that the study would still bear out with a much larger accident sample size, since 92 is unfortunately unlikely to be deemed adequate when divided by 12 signs to provide for conclusive accuracy) appears to have a very strongly protective effect; in three others appears to have a slightly protective effect, and in five others appears to have a slightly harmful effect, on the risk of fatal accidents before the age of 29.

It?s true. I provided the Tables exactly to give the possibility of people to evaluate the highest and lowest scores meaning more and less protective strength coming from determined placement and you are putting in words what was showed in numbers!

I feel that to simply add up the results for a particular planet in all the signs is a mistake since it tends to gloss over the very different effects of that planet in the different signs, and also for another reason, on statistical principle, as I'll go on to try my best to explain briefly:

If as I am presuming in the worked examples given previously, the figures in the hundreds and tens and units represent the percentage of difference between the percentage incidences of particular sign placements for particular planets in the accident and control groups, then a reported negative difference of 300% in one may represent the difference between 2 individuals in the accident group and 6 in the control group, while a reported positive difference of 66% may represent the difference between 3 individuals in the accident group and 5 in the control group. Between these two scenarios, while the numbers look very different, there is only a very slight (unitary, in fact) difference between the actual numbers in the accident group and control group, differences that could be explained purely by chance, and even if not, differences that are small. So to add a result of 300% to one of 66% and claim a figure of 366 could be very misleading. These percentages of difference do not meaningfully lend themselves to addition, full stop. And if you then go on and use the overall additions on one side and subtractions on the other to try to demonstrate that a particular planet is protective of harmful overall, it makes less sense still.

I understood your point and I have to say, as I stressed in the article, that my study was just a preliminary consideration that can be used as an initial point to talk and share ideas with colleagues or with someone who knows statistics and astrology. I?m aware how much study it is necessary to find an astrosignature p. > 0,5 or even, in human issues the ideal being p. < 0,001!
To conclude in a real statistical way, finding an astrosignature , like the placement of the dispositor of the Moon in Pisces, Mercury in Pisces and Sun In Pisces,( the most important factors I founded inclining people to accidents,) it?s necessary to check this aspects in a global population.
Doing a scientific proof on this matter would be the objective of a monograph to present to a PHD degree. This is a simple article focused on awaking interest in the area and comparing the method with traditional delineation. It?s vision of the future : statistics working with traditional delineation!

I have to cut this short but hope you don't take this post the wrong way - I am all in favor of statistical astrological research but am trying to help ensure that it is both methodologically and analytically rigorous, as well as being clearly labelled and presented.

I didn?t take you wrong at all, and I have to thank you the collaboration. But one thing I have to make it clear: my intention at any moment wasn?t so ambitious as to give the last word in a tough matter as the statistic study, but, just the opposite, to open the discussions about the matter. So, I appreciated your intelligent remarks!


Clelia

Hi Clelia!

Thank you for your gracious response and all the clarifications.

I don't have Astrodatabank and have no experience of using anyone else's copy, so this was all new to me but helpful to know.

It does sound from what you are saying as though my speculation about the percentage differences in the percentage incidences between the two groups was at least approximately correct, though until I've seen the Astrodatabank program in action on a similar task I do not feel 100% sure of this!

One thing that would be really helpful to eliminate all uncertainty about what the figures in the table mean would be to include in your report (as you come to expand it into a full-length report, and I acknowledge what you say about it being only a starting-point for now!) the original raw data of the number of people in each group having each sign placement, as well as the percentage of people in each group having each sign placement. Then the calculation that was used by the software to obtain the numbers in your table could be deduced definitively!

Right after I posted my first reply to you yesterday, I suffered the loss of my Internet modem for the rest of the day. Knowing that this was about to occur, I rushed the last part slightly, and felt afterwards that I really should and could have explained my statistical objection to the addition of the figures in the table for each of the planets in turn to create the figures in the right-hand column more thoroughly and clearly.

Lacking the modem, I set to work in a new Outlook Express draft message with an imaginary set of data from a study similar to yours, and then worked through it to obtain the kinds of percentage difference calculations that we have both acknowledged to be the meaning of the numbers in your table. I was then able to demonstrate my point more clearly, as I shall attempt to go on to explain:

Moon:

.............Ari......Tau.....Gem.....Can......Leo......Vir.....Lib......Sco.....Sag.....Cap......Aqu.....Pis....TOTAL
AGN:....10.......11.......12........10.........8.........9........7.........7.........4.........8..........3.........3........92
AG%:...10.8....12.0....13.0.....10.9......8.7......9.8.....7.6......7.6......4.3......8.7.......3.3......3.3...100

CGN:....72.......75.......74........77.......73.......71......78.......82.......80.......77........78.......83......920
CG%:.....7.83....8.15....8.04.....8.37....7.93....7.72...8.48....8.91....8.70....8.37.....8.48...9.02..100

DIFN:...+2.97..+3.85..+4.96..+2.53..+0.77..+2.08..-0.88..-1.31...-4.40..+0.33...-5.18...-5.72.....0
DIF%:+37.9..+47.2..+61.7..+30.2..+09.7..+26.9..-11.5..-17.2..-102.3.+03.9.-157.0.-200.7.-271.2

This really should be displayed in an even-spaced font such as Courier, but I don't think the default font for these forums can be overridden (I tried with HTML but it didn't work!). I tried to tidy it up by editing afterwards to add more spaces where necessary, etc., but it didn't work, collapsing all the spaces into a single space in each case. So in the end I laboriously went through the entire thing adding dots by hand until the spacing was as even as possible! I hope it is now at least readable!

Anyhow, the meaning of the six rows:
AGN = Accident Group Number (absolute incidence of each of the placements in the accident group)
AG% = Accident Group number as a Percentage of the total number of placements (ie 92) in the accident group
CGN = Control Group Number (absolute incidence of each of the placements in the control group)
CG% = Control Group number as a Percentage of the total number of placements (ie 920) in the control group
DIFN = Numerical difference between the AG% and the CG% figures for each placement
DIF% = Percentage difference between the AG% and CG% figures for each placement

You should be able to notice that the DIF% figures are equivalent by their method of calculative derivation to the figures in your table! The only differences in their numerical values compared with yours stem from the fact that my source data has been made up to approximate to what yours might theoretically have been if using the same sizes of control group and accident group and giving similar effect sizes.

The key to my demonstration is found in the difference between the totals of the DIFN row and the DIF% row. The DIFN row's figures add up to zero. The DIF% one's do not.

I have now returned to complete this post, so to resume from where I had got to:

The reason why the figures in the DIFN row add up to zero is that they are obeying a basic law of statistics always observed when control groups are compared with effect groups, which is that any positive effect caused by (or at least correlating to) one particular cause (or situation) must be statistically balanced by an equal negative effect correlating to all others combined.

In my example table above using the made-up data, for there to be a sharply negative effect caused by the Moon in Aquarius and the Moon in Pisces, there must be a correspondingly positive (ie opposite) effect of the same value averaged out over the Moon in all the other signs combined. This is borne out by the figures, which show that the strongly negative effect on the incidence of fatal accidents before the age of 29 (in my example table) of the Moon in Aquarius or Pisces is balanced by a moderately positive one in several other signs, as measured by the numerical difference between the percentage of individuals in the control group having the Moon in each particular sign and the percentage of individuals in the accident group having the Moon in the same sign.

The effects being measured are always effects by comparison with an average. The average percentage of individuals in the accident group having the Moon in each sign is always going to be (100/12) 8.33%. The fact that there is a smaller percentage than this in certain signs means that there must be a correspondingly greater percentage than this in others.

The average percentage of individuals in the control group having the Moon in each sign is also always going to be (100/12) 8.33%. In my example table, I have presumed that in a Control Group of 920 there is still going to be some random deviation from parity in the incidence of each sign placement occurring. Thus I have presented a slightly uneven distribution of the Moon in the 12 signs shared among the 920 individuals that results in percentages ranging in practice from 7.83% to 9.02%, but the total is still 100% and the average is invariably 8.33%.

The fact that the accident group is being compared with a control group that exhibits some slight natural variations will result in some corresponding variability to the results produced. It is beyond the scope of this particular post to go into the mathematics of how to take this into account and also to take into account random variations in the accident group in assessing the reliability or otherwise of the apparent effect size demonstrated, but these are important issues that also need to be borne in mind.

But to return to the point at hand, the control group is essentially an approximation to a natural average of 8.33% for the incidence of the Moon in each of the 12 signs, and it is around this average that the variable incidence of the Moon in each sign in the accident group is essentially being measured by the experiment. Again, because it is the natural average, any extreme incidences to either side of that average in the test population (ie the accident group) must be balanced by an equal and opposite incidence on the other side.

This is to say that a protective effect (compared with the average) of the Moon in certain signs must always be balanced by a hazardous (compared with the average) effect of the Moon in the remaining signs. There might for example be a strongly protective effect in two or three signs balanced by a more mildly hazardous effect in the remaining nine signs, or there might be an even balance in the levels of protectiveness in six signs and the levels of hazardousness in the other six. Any number of distributions is theoretically possible, but the average will always hold true, and the balance of the positive effects around the negative ones will always be perfect and exact.

Thus, the Moon overall cannot be statistically demonstrated to have a generally protective function unless it is being compared with the lack of the Moon, and since the Moon has always been there so long as any human being has been born on the Earth, such a comparison is impossible unless someone first removes the Moon (which I would not personally recommend).

If you were to apply your raw data to the same calculations, creating your own DIFN line in your table by the same procedure, you would find that the figures for the 12 sign placements equally add up to zero for every single factor being tested (Ascendant, Sun, Moon, Mercury, Venus, etc., etc.) in turn. Such is the operation of the law of averages.

Thus, none of the planets can be found to have an overall protective or harmful effect by this kind of statistical experiment.

The reason your figures do not add up to zero is that they have been produced by a means of calculation that does not lend itself logically to the additive sum of the end-results of those calculations having any meaningful mathematical use. They have instead been produced by the method used in my example table above to create the DIF% row, showing the percentage difference between two sets of percentage figures, as opposed to the numerical (additive) difference between the same. They each show one percentage figure (eg the % of control group individuals having the Moon in a Aries) as a percentage of another percentage figure (eg the % of the accident group individuals having the Moon in Aries). They thus have their own internal logic as a demonstration of the effect size for each individual Moon placement, but cannot be meaningfully added up. Because of the way they (ie the figures) have been derived, the larger effect sizes will always be represented as disproportionately great figures (whether positive or negative) and the smaller ones as disproportionately small figures, so that an uneven distribution of large negative effect sizes in a small number of signs balanced by moderate positive effect sizes in a larger number of others will appear, if the figures are all added up, to demonstrate an overall (net) negative effect for the planet concerned, and conversely, an uneven distribution of large positive effect sizes in a small number of signs balanced by moderate negative effect sizes in a larger number of others will appear, if the figures are all added up, to demonstrate an overall (net) positive effect for the planet concerned; but in both such scenarios, the appearance is an illusion, a mere artefact from the misapplication of the additive process to figures that cannot (because of their derivation) meaningfully be added up to draw any such conclusions.

To recap, the DIFN line demonstrates how figures derived from the same raw data but using processes that are compatible with their summation always add up to zero because of the law of averages. The DIF% line demonstrates how figures derived from the same raw data but using a geometrical (as opposed to arithmetic) process incompatible with their summation may add up to anything but zero, and that nothing can be meaningfully inferred from this, except for the fact that the distribution of positive and negative effect sizes among the twelve signs for the planet in question was uneven, although the overall distribution was perfectly balanced.

Final note: please take a look at the Gemini and Aquarius columns in turn in the table above. You will see that for each of these sign placements, the numerical difference (DIFN) between the percentage incidence of control group individuals having the lunar placement concerned and the percentage incidence of accident group individuals having the same placement is approximately 5 - though slightly below 5 in one case and slightly above in the other. Yet the percentage differences (DIF%) are very much greater, because, in the case of the Moon in Aquarius, 3.3% compared with 8.48%, though a numerical difference of 5.18%, is a percentage difference of 157%, whereas in the case of the Moon in Gemini, 13.0% compared with 8.04%, though a numerical difference of 4.96% (almost as high as the difference for the Moon in Aquarius), is a percentage difference of only 61.7%. This comparison serves, I think, as a fairly useful example of how the derivation of the DIF% figures, though internally meaningful for the particular sign placement in question in showing the percentage difference between the control group's and study group's respective percentage incidences of that placement, massively skews the appearance of the overall data for the planet under study in favour of highlighting the larger individual percentage effect sizes in certain sign placements, while diminshing the appearance of the balancing numerical effect sizes in other sign placements that do not have such high individual percentage effect values.

Best wishes,

Philip

You wrote:

'To conclude in a real statistical way, finding an astrosignature , like the placement of the dispositor of the Moon in Pisces, Mercury in Pisces and Sun In Pisces,( the most important factors I founded inclining people to accidents,) it?s necessary to check this aspects in a global population.
Doing a scientific proof on this matter would be the objective of a monograph to present to a PHD degree. This is a simple article focused on awaking interest in the area and comparing the method with traditional delineation. It?s vision of the future : statistics working with traditional delineation!'

My reply: Agreed, and I very much admire your forthright presentation of your work in progress to the end of productively stimulating interest in the cause! I've said before elsewhere that I think that the tip of the iceberg of possibilities with regard to the full scope of statistical astrological research projects that might productively be carried out has barely yet been scratched. In application to traditional, modern and purely experimental astrological concepts, it can only help to place natural astrology on a firmer scientific footing ultimately, so long as the data that is obtained is interpreted correctly, and neither misused by astrologically worshipful astrologers to claim proof where this has not been adequately demonstrated, nor misused by devout, belligerent sceptics to claim disproof of a broader range of assumptions than the one that was specifically tested in each case. By the nature of divination, it would strike me as somewhat dubious to seek to apply statistical research to divinatory forms of astrology such as horary, though ultimately I would prefer to leave that judgement to horary practitioners, since I lack the requisite experience to be qualified to hold an opinion in that area.

Because of the lack of funding for statistical research into astrology, I suspect that the pace of research being published will remain slow for a very long time, but that a corpus of useful results will gradually be built up. The statistical astrological researchers of the 20th century such as Choisnard, Krafft, the Gauquelins, John Addey, Donald Bradley and Lois Rodden will be looked back upon as early pioneers, and those of the 21st century as their adventurous successors persisting in spite of a strong late 20th century tide of negative opinion within both the scientific and the astrological communities about the potential value of such research.

My expectation in any case!

Philip

[quote="Philip Graves"]Hi Clelia!

Hi Philip!

Thank you for your gracious response and all the clarifications.

I don't have Astrodatabank and have no experience of using anyone else's copy, so this was all new to me but helpful to know.

It does sound from what you are saying as though my speculation about the percentage differences in the percentage incidences between the two groups was at least approximately correct, though until I've seen the Astrodatabank program in action on a similar task I do not feel 100% sure of this!

One thing that would be really helpful to eliminate all uncertainty about what the figures in the table mean would be to include in your report (as you come to expand it into a full-length report, and I acknowledge what you say about it being only a starting-point for now!) the original raw data of the number of people in each group having each sign placement, as well as the percentage of people in each group having each sign placement. Then the calculation that was used by the software to obtain the numbers in your table could be deduced definitively!


Your post yesterday thrilled me so much that I went back to check my former deductions, and guess what? I did what you suggested: I tried to construct an astrosignature of the group subject to accidents before age 29.
I can see you have a good expertise on statistics, and this is not my case: I had 3 years of Statistics in the university, but I?m not able to work with this tough matter. Astrodatabank spared me the hard job, but it?s a difficult software to work with when you want to do research and don?t remember the basic foundations.
I was thinking in publishing the Astrosignature report, but this would be a second article, for those interested in the scientific basis of my conclusions. You, for example: P
What I did: I selected a factor where %NZ score in Experimental group was > 10% AND Abs ( Diff) for % NZ score is 100% or more.
The group was reduced, of course. But I founded some factors appearing with high scores; mostly the dispositor of the Moon is Pisces.

Now, I know that the ideal would be to have another collection of people who deceased because accidents in several countries and check if the same factors appears.
And this is a work for a life, isn?t it?




Right after I posted my first reply to you yesterday, I suffered the loss of my Internet modem for the rest of the day. Knowing that this was about to occur, I rushed the last part slightly, and felt afterwards that I really should and could have explained my statistical objection to the addition of the figures in the table for each of the planets in turn to create the figures in the right-hand column more thoroughly and clearly.

Lacking the modem, I set to work in a new Outlook Express draft message with an imaginary set of data from a study similar to yours, and then worked through it to obtain the kinds of percentage difference calculations that we have both acknowledged to be the meaning of the numbers in your table. I was then able to demonstrate my point more clearly, as I shall attempt to go on to explain:

Moon:

.............Ari......Tau.....Gem.....Can......Leo......Vir.....Lib......Sco.....Sag.....Cap......Aqu.....Pis....TOTAL
AGN:....10.......11.......12........10.........8.........9........7.........7.........4.........8..........3.........3........92
AG%:...10.8....12.0....13.0.....10.9......8.7......9.8.....7.6......7.6......4.3......8.7.......3.3......3.3...100

CGN:....72.......75.......74........77.......73.......71......78.......82.......80.......77........78.......83......920
CG%:.....7.83....8.15....8.04.....8.37....7.93....7.72...8.48....8.91....8.70....8.37.....8.48...9.02..100

DIFN:...+2.97..+3.85..+4.96..+2.53..+0.77..+2.08..-0.88..-1.31...-4.40..+0.33...-5.18...-5.72.....0
DIF%:+37.9..+47.2..+61.7..+30.2..+09.7..+26.9..-11.5..-17.2..-102.3.+03.9.-157.0.-200.7.-271.2

This really should be displayed in an even-spaced font such as Courier, but I don't think the default font for these forums can be overridden (I tried with HTML but it didn't work!). I tried to tidy it up by editing afterwards to add more spaces where necessary, etc., but it didn't work, collapsing all the spaces into a single space in each case. So in the end I laboriously went through the entire thing adding dots by hand until the spacing was as even as possible! I hope it is now at least readable!


Yes, it?s okay, Philip!

A
nyhow, the meaning of the six rows:
AGN = Accident Group Number (absolute incidence of each of the placements in the accident group)
AG% = Accident Group number as a Percentage of the total number of placements (ie 92) in the accident group
CGN = Control Group Number (absolute incidence of each of the placements in the control group)
CG% = Control Group number as a Percentage of the total number of placements (ie 920) in the control group
DIFN = Numerical difference between the AG% and the CG% figures for each placement
DIF% = Percentage difference between the AG% and CG% figures for each placement

You should be able to notice that the DIF% figures are equivalent by their method of calculative derivation to the figures in your table! The only differences in their numerical values compared with yours stem from the fact that my source data has been made up to approximate to what yours might theoretically have been if using the same sizes of control group and accident group and giving similar effect sizes.

The key to my demonstration is found in the difference between the totals of the DIFN row and the DIF% row. The DIFN row's figures add up to zero. The DIF% one's do not.

I have now returned to complete this post, so to resume from where I had got to:

The reason why the figures in the DIFN row add up to zero is that they are obeying a basic law of statistics always observed when control groups are compared with effect groups, which is that any positive effect caused by (or at least correlating to) one particular cause (or situation) must be statistically balanced by an equal negative effect correlating to all others combined.

I think I got what you mean. But as I wrote above the software take this into account.


In my example table above using the made-up data, for there to be a sharply negative effect caused by the Moon in Aquarius and the Moon in Pisces, there must be a correspondingly positive (ie opposite) effect of the same value averaged out over the Moon in all the other signs combined. This is borne out by the figures, which show that the strongly negative effect on the incidence of fatal accidents before the age of 29 (in my example table) of the Moon in Aquarius or Pisces is balanced by a moderately positive one in several other signs, as measured by the numerical difference between the percentage of individuals in the control group having the Moon in each particular sign and the percentage of individuals in the accident group having the Moon in the same sign.

Okay.

The effects being measured are always effects by comparison with an average. The average percentage of individuals in the accident group having the Moon in each sign is always going to be (100/12) 8.5%. The fact that there is a smaller percentage than this in certain signs means that there must be a correspondingly greater percentage than this in others.

The average percentage of individuals in the control group having the Moon in each sign is also always going to be (100/12) 8.5%. In my example table, I have presumed that in a Control Group of 920 there is still going to be some random deviation from parity in the incidence of each sign placement occurring. Thus I have presented a slightly uneven distribution of the Moon in the 12 signs shared among the 920 individuals that results in percentages ranging in practice from 7.83% to 9.02%, but the total is still 100% and the average is invariably 8.5%.

Let?s see if I understood: a result of 8,5 is not impressive at all: we need more to considerate an important fact.

The fact that the accident group is being compared with a control group that exhibits some slight natural variations will result in some corresponding variability to the results produced. It is beyond the scope of this particular post to go into the mathematics of how to take this into account and also to take into account random variations in the accident group in assessing the reliability or otherwise of the apparent effect size demonstrated, but these are important issues that also need to be borne in mind.

Of course they are, when doing a scientific statistical job!
But to return to the point at hand, the control group is essentially an approximation to a natural average of 8.5% for the incidence of the Moon in each of the 12 signs, and it is around this average that the variable incidence of the Moon in each sign in the accident group is essentially being measured by the experiment. Again, because it is the natural average, any extreme incidences to either side of that average in the test population (ie the accident group) must be balanced by an equal and opposite incidence on the other side.

Agreed!

This is to say that a protective effect (compared with the average) of the Moon in certain signs must always be balanced by a hazardous (compared with the average) effect of the Moon in the remaining signs. There might for example be a strongly protective effect in two or three signs balanced by a more mildly hazardous effect in the remaining nine signs, or there might be an even balance in the levels of protectiveness in six signs and the levels of hazardousness in the other six. Any number of distributions is theoretically possible, but the average will always hold true, and the balance of the positive effects around the negative ones will always be perfect and exact.

This is something I can?t prove using the software, since I don?t have a population specially protected against accidents, this category doesn?t exists
But I imagine the control group has an average protection since lived more than 29 years, isn?t it?

Thus, the Moon overall cannot be statistically demonstrated to have a generally protective function unless it is being compared with the lack of the Moon, and since the Moon has always been there so long as any human being has been born on the Earth, such a comparison is impossible unless someone first removes the Moon (which I would not personally recommend).

O, Philip, this is true respecting all planets! I said the dispositor of the Moon had the most protective influence because it appears more than the Sun or any planet with the highest score in the population that was protect against or inclined to accidents.

Look at this result:

Dispositor of the Moon in Fixed 0,12 0,31 -1,57 0,12 0,31 -1,57 12,24 31,43 -156,67
Dispositor of the Moon in Pisces 0,22 0,07 2,19 0,22 0,07 2,19 22,45 7,04 218,84

Now I?ll give you what is written above each column::

WtAveEx WtAveCtr WtAveDif AveExp AveCtrl AveDiff %NZExp %NZCtrl %NZDiff


That?s my way of saying: I expect it?s readable!


If you were to apply your raw data to the same calculations, creating your own DIFN line in your table by the same procedure, you would find that the figures for the 12 sign placements equally add up to zero for every single factor being tested (Ascendant, Sun, Moon, Mercury, Venus, etc., etc.) in turn. Such is the operation of the law of averages.

Thus, none of the planets can be found to have an overall protective or harmful effect by this kind of statistical experiment.

May be I had to explain it better. What I was trying to say is that the Moon?s dispositor has the highest score inclining people to be killed by an accident, and the highest negative score not inclining people to have this kind of problem. So, the Moon?s position is important in this study!

The reason your figures do not add up to zero is that they have been produced by a means of calculation that does not lend itself logically to the additive sum of the end-results of those calculations having any meaningful mathematical use. They have instead been produced by the method used in my example table above to create the DIF% row, showing the percentage difference between two sets of percentage figures, as opposed to the numerical (additive) difference between the same. They each show one percentage figure (eg the % of control group individuals having the Moon in a Aries) as a percentage of another percentage figure (eg the % of the accident group individuals having the Moon in Aries). They thus have their own internal logic as a demonstration of the effect size for each individual Moon placement, but cannot be meaningfully added up.

Do you mean the sum I did in each column is not valid? If you?re referring to it, I totally agree. And I did the sum to work as a red light to people pay attention to the cells in the column! The cells have importance, not the sum. On the other hand if you find a large incidence of Pisces, let?s suppose, because, the dispositors of the Moon in Pisces have a high score to accidents, Mercury and Sun in Pisces as well, you can observe (summing this factors in your mind) that you must pay attention to Pisces as a potentially dangerous sign, more than the other ones. And when I added this factors to construct an astrosignature I had a significant score.
If you want I can sent you by e-mail or put it in my web site and give you the link, because to put another table here I?m afraid it?s not easy to the visualization.
I create a page in html in order you can see the results more clearly.
http://www.astrologiahumana.com/untitled-1.htm


Tell me please if you think they make sense for you as they did for me!


Best wishes

Clelia

Hi Clelia!

Many thanks for your further response.

Since having an active, lively two-year-old dance-music-loving baby around in the evenings is highly incompatible with concentrated thought, I'm going to leave responding further until tomorrow or whenever I next get a reasonably free stretch of time!

Meanwhile, please note that just now as I was away from the computer and reflecting on the contents of this strand in my head I suddenly realised that I had made a glaring mathematical error in my own previous post, as a consequence of being in too much of a hurry to use a calculator - not that I should have need for one to have got this simple arithmetical operation correct. The point I refer to was my false assertion that the average incidence of the placement of the Moon in each sign for any population (control or experimental) should be (100/12) 8.5%, when what I of course should have written is '...(100/12) 8.33(R)%'. I made this error by thinking '12 eights are 96; and half an eight makes 100', but of course the division in question was by twelve, not by eight, so I should have been thinking '8 twelves are 96; and a third of a 12 makes 100'.

These mistakes happen, I suppose, in moments of hasty thought! Fortunately this mistake was of no consequence to the train of my argument, as the precise figure that is the natural average was not critical to the point, but still I wanted to correct it for the record!

Briefly, in some parts of your reply I get the impression that we are talking about two different things, and therefore at cross-purposes, which is why I really want some clear thinking time to go back and point out the differences!

By the way, I don't have any recent training in statistics either, so I have forgotten virtually all of the technical jargon I may once have picked up. This perhaps is a good thing for the purposes of this strand since essentially I want to talk in plain, logical English, even if it means I have to explain some things at elaborate length that to a statistician would be easily identifiable under a standard concise term or abbreviation that to many reading this would be opaque, rendering the whole discussion of relatively narrow accessibility. I certainly don't consider myself a statistical expert, but if nothing else, I do believe I have a reasonably keen eye and brain for logic, including mathematical logic, and that is really the only qualification I pretend to bring to this discussion. Please let my contributions therefore stand or fall on their own inherent merits of reason or otherwise, and I should be delighted in turn if an advanced statistician could point out where I have erred in my own thinking.

Best wishes until tomorrow, or soon after,

Philip

[quote="Philip Graves"]Hi Clelia!

Many thanks for your further response.

Since having an active, lively two-year-old dance-music-loving baby around in the evenings is highly incompatible with concentrated thought, I'm going to leave responding further until tomorrow or whenever I next get a reasonably free stretch of time!

That is what I call a nice duty! Take your time Philip and enjoy your child, they grow so fast!

Meanwhile, please note that just now as I was away from the computer and reflecting on the contents of this strand in my head I suddenly realised that I had made a glaring mathematical error in my own previous post, as a consequence of being in too much of a hurry to use a calculator - not that I should have need for one to have got this simple arithmetical operation correct. The point I refer to was my false assertion that the average incidence of the placement of the Moon in each sign for any population (control or experimental) should be (100/12) 8.5%, when what I of course should have written is '...(100/12) 8.33(R)%'. I made this error by thinking '12 eights are 96; and half an eight makes 100', but of course the division in question was by twelve, not by eight, so I should have been thinking '8 twelves are 96; and a third of a 12 makes 100'.

These mistakes happen, I suppose, in moments of hasty thought! Fortunately this mistake was of no consequence to the train of my argument, as the precise figure that is the natural average was not critical to the point, but still I wanted to correct it for the record!

Okay, indeed this little mistake had not a consequence of changing your reasoning.


Briefly, in some parts of your reply I get the impression that we are talking about two different things, and therefore at cross-purposes, which is why I really want some clear thinking time to go back and point out the differences!

I would like you to do this, please, because we have also the language problem, since English is not my first language!

By the way, I don't have any recent training in statistics either, so I have forgotten virtually all of the technical jargon I may once have picked up. This perhaps is a good thing for the purposes of this strand since essentially I want to talk in plain, logical English, even if it means I have to explain some things at elaborate length that to a statistician would be easily identifiable under a standard concise term or abbreviation that to many reading this would be opaque, rendering the whole discussion of relatively narrow accessibility. I certainly don't consider myself a statistical expert, but if nothing else, I do believe I have a reasonably keen eye and brain for logic, including mathematical logic, and that is really the only qualification I pretend to bring to this discussion. Please let my contributions therefore stand or fall on their own inherent merits of reason or otherwise, and I should be delighted in turn if an advanced statistician could point out where I have erred in my own thinking.

Good to me you?re not an statistician, since for me the words are more understandable than the numbers;-) I can?t imagine talking with you using the statistic jargon!l
My mathematical expertise as well was never so good as my verbal thought,in the school that was my big problem, but I?m becoming better with the years.

Best wishes until tomorrow, or soon after,

Best wishes and enjoy your family:-)

Clelia

Hi Philip!

I just found an interesting statistical study at http://cura.online.fr:80/xx/18cas3en.html

I remind of our converstion and decided to send you )and to all interested in this thread) the link!

best wishes

Clelia

Hi Clelia!

Thank you for the link.

I may not have time to reply to everything you have lately raised tonight but I'll make a start!

Firstly I would like to clarify one of my own points a little.

Where I wrote:

'The reason why the figures in the DIFN row add up to zero is that they are obeying a basic law of statistics always observed when control groups are compared with effect groups, which is that any positive effect caused by (or at least correlating to) one particular cause (or situation) must be statistically balanced by an equal negative effect correlating to all others combined'...

I should clarify my point since not all studies comparing control groups to effect groups are conducted in the same way as this one. It would perhaps be best for me to more narrowly delimit the context in which my point was meant to be applied and to elucidate it further. To this end I have a series of points to make:

1. This particular study has taken a known effect as its starting point (accidental death before the age of 29) and examined the distribution of causes (or at least correlative factors, depending on one's philosophical point of view with regard to the doctrine of astrological causation versus that of acausal correlation, but for the sake of simplicity I shall hereafter refer to them as causes!) of various natal astrological types in the known-effect group in comparison to the distribution of the same causes in a control group that has been prepared randomly from the general population without regard to any particular grouping criterion, and which therefore should not evidence any abnormal distribution of the same causes in its constituent personal make-up, which is to say none of an order of magnitude greater than would be statistically expected by pure chance.

2. In this particular study, because it is looking into possible natal astrological causes for the known effect that is the defining basis for the composition of the study group, it is considering causes that, rather than being merely generically present or absent (eg smoker / non-smoker, as might be a cause considered in a hypothetical comparable epidemiological medical study), must always be present (since there is always a Moon in one sign at birth, always a dispositor of the Moon in one house at birth, and so on!), but in a variety of different mutually exclusive configurations, typically twelve for sign and house placements of many of the causes under consideration.

3. From the nature of the causes under study as outlined above, it stands to reason that for every hypothetical cause of the known effect being considered by the study (eg the tropical zodiacal sign in which the natal Moon of each subject is placed), any numerical surfeit in the percentage incidence of that cause found in any particular individual configuration (eg, hypothetically, a greater than average incidence of the Moon in Virgo) compared with the average (which should in the case of the Moon be equivalent to the incidence expected for every sign placement by chance, though for other causes such as the Ascendant having less even frequency distributions among the signs this would not be so) percentage incidence of the same cause in each possible individual configuration must be statistically balanced out by an equal and opposite numerical deficiency in the percentage incidence of that cause found in all other configurations of the same cause combined (in this example, the Moon in all the other eleven signs summed together as an arbitrary group) compared to the sum of the number of averages that corresponds to the number of other possible configurations of the same cause (in this example, eleven, since there are eleven other signs than Virgo where the Moon may be placed if and only if it is not in Virgo; and since the average incidence of the population of each of 12 possible sign placements in the study group is (100/12) 8.3R%, eleven times that average makes 91.6R% as the percentage incidence of the Moon in all the signs apart from Virgo combined that would be expected if in accordance with the average per placement).

Indeed, and similarly, any numerical surfeit in the percentage incidence of that cause found in any arbitrarily selected grouping of configurations considered together (eg, hypothetically, a greater than average incidence of the Moon in either Virgo, Libra or Capricorn as a human-selected group, regardless of the incidence in each individual sign, which might on analysis for example be slightly below-average in two and strongly above in the third) compared to the average incidence per configuration must be balanced out by an equal and opposite numerical deficiency in the percentage incidence of that cause found in all other configurations combined (in this example, the other nine signs averaged together as a group) compared to the average per configuration (ie sign, in this example).

4. The above holds true mathematically because we are dealing with percentages, or, more fundamentally speaking, mutually competitive statistical shares or rations, of a completely considered whole that embraces all mutually exclusive possibilities, for every astrological cause we may be considering, whether it be for example the Moon's sign placement, the Moon's house placement, the Moon's dispositor's sign placement, the Moon's dispositor's house placement, or any other like category of possible natal astrological cause considered in your study.

5. To illustrate this even more clearly by analogy (though I don't suppose this will be necessary for you, personally, Clelia, but it might perhaps be of help to some other readers struggling with our statistical discussion), one may consider the case of three siblings with similar names sharing a cake. If Joe eats 50% of the cake, and Joanne eats 40%, there will be only 10% left for John. The average eaten per child is always 33.3R%, or one third, because there are three children and the size of the cake is always 100%. This then we could call each child's fair share. But the percentage consumptive excesses compared with that fair share of any one child or group of children must numerically be balanced by the consumptive deficiencies of the other(s) in equal and opposite measure. So if we take Joe for example, who has eaten 16.6R% more of the overall cake than was his fair share, it stands to reason that Joanne and John combined must eat between them 16.6R% less of the overall cake than was their fair share, or on average 8.3R% less of the overall cake per person. But because Joanne also exceeds her fair share, by the time the cake reaches John, Joe and Joanne combined have eaten (16.6R+6.6R) 23.3R% more of the overall cake than was their fair share, so there is only 10% of John's due 33.3R% share left for him - exactly 23.3R% less than was his fair share. Joanne and John as a group are still consuming, between them, 16.6R% less of the overall cake than was their fair share, to equally balance greedy Joe's 16.6R% excess, but this deficiency has been divided unevenly between a 6.6R% excess and a 23.3R% deficiency among the two remaining children concerned, instead of being split between two 8.3R% deficiencies.

6. In the above analogy for the purposes of statistical illustration, the differently sized portions of cake consumed by the three children have exactly the same statistical attributes as the different sign placements for the Moon (etc., etc.) in your accident study group. Just as the cake is a certain finite size (100%), so is your study group's population (100%). And just as Joe's excesses in consuming 16.6R% more of the overall cake than was his fair share reduce the available amount for the others by the same measure, so the presence of a certain numerical amount more than 8.3R% (100% divided by 12 signs, in this case, not to be confused with the entirely separately derived 8.3R% that cropped up in the cake example earlier!) of individuals in your study group (for example, let us say 3% of the total number more, ie 11.3R% in total) having the Moon in one particular sign must be balanced by all the remaining Moon signs being shared among the same amount less than 91.6R% (ie 100%-8.3R%) of the total study population [in our example, this would give 88.6R%, which divided by eleven signs produces an average of just over 8% of the population per sign, or just over 0.3% less than would be expected by chance per sign, though again this average may be unevenly distributed among the remaining 11 signs, just as the average deficiency of 8.3R% (ie 16.6R%/2) of the overall cake per remaining sibling after Joe took 16.6R% more of it than was his fair share was unevenly distributed between Joanne and John].

7. The same law of averages equally applies to studies in which the distribution of individuals in a known effect group according to criteria answerable by a simple 'Yes/No' split is considered, such as the 'Smoking / Non-smoking' criterion alluded to previously in the cases of medical studies into causes of particular diseases, though in such cases, because there is merely a two-way split in the distribution of individuals, the mathematics are very much simpler. Thus, if for example 60% of individuals in a known mortality group being studied were found to be smokers, the smoking component of the group has exceeded the share allowable to it by chance by 10% of the total population in the group, which much therefore be equally balanced by a 10% reduction (as a proportion of the overall group popuation) in the non-smoking component of the group, this standing consequently at 40% of the total population.

8. In your study, just as in the simpler disease study considered immediately above, we are looking at possible causes of a known effect, and because the distribution of the population in the study group for any causal criterion examined will always be averaged out around chance (as I have sought in the foregoing points to demonstrate), it is very, very important to realise that a negative 'effect' (ie a reduction in the percentage of the overall study group population compared with chance) in some of the particular mutually exclusive configurations of each generic astrological cause being studied (eg the Moon in the signs) must always be balanced by a numerically equivalent positive 'effect' (ie a corresponding increase in the percentage of the overall study group population compared with chance) in the average of all the others combined.

9. In the known-effect group (ie the accident mortality group) in your study, as a consequence of this invariable distribution around a mathematical average of the population of any set of individuals in response to any criterion by which it may be divided, it is not immediately easy to judge whether any individual apparent (positive or negative) statistical effect observed in one or more of the mutually exclusive configurations of a possible causal factor under consideration is a direct causal effect of that particular configuration or a mere statistical artefact of the averaging process consequent upon a causal relationship that resulted in one or more of the other mutually exclusive configurations of the same causal factor accounting for more (or less) than an average share of the population in the group. To illustrate, for example, if the Moon in Pisces has a protective effect that reduces the prevalence of lunar Pisceans in the study group by 5% of the overall study group population compared with the average prevalence of individuals in the study group per lunar sign placement, then the prevalence of those born under all the other Moon signs combined must correspondingly be increased by 5% of the overall study group size, which is to say by an average of (5/11)% of the overall study group size per sign. The Moon may (in this hypothetical scenario) truly have no action at all in those other signs, yet because of the workings of averages a weak positive effect appears in them (taken on average). Conversely, yet still with the exact same distribution of the study group population applying in response to the exact same distribution criterion, the Moon might truly have a markedly hazardous effect in all those other signs, and no effect at all in Pisces, yet because of the workings of averages, the effect in the eleven signs other than Pisces on average appears weak. Which is truly the case we simply cannot know, because we can never take the Moon away altogether. We can only compare distributions to the average; but it is very important at the analytical and interpretative stage of this kind of study to remain aware of how percentage deviations to one side or the other of an average for certain portions of a population may be mere artefacts of opposite effects within other portions of that same population!

Alas, this is as far as I have time to go tonight, but I really wanted to underline those basics first, even at the risk of stating the obvious at undue length here and there, because a clear mental grasp of them is essential to understanding any statistical study of the type that you have presented, as well as to the particular points I made previously!

I'm sorry this is all proving so long and drawn-out, but I do hope it will be somewhat educational and preferably also inspiring to one or two others reading these forums for whom statistical research into astrology might previously have presented a few mental stumbling blocks, which is really nothing to be ashamed of if so.

I am finding that it is nonetheless quite a challenge to convey even some of these elementary points through writing using clearly defined and consistently applied language, and virtually impossible to convey some of them in short, simply structured sentences. I have already edited this post quite extensively in an attempt to enhance clarity at various points, but I must apologise for the fact that not all the points lend themselves to short sentences, and especially not where I have repeatedly referenced them to examples as I wrote!

Best wishes,

Philip

Hi Philip!

All this conversation was very inspiring to me, and I thank you for taking the time to answer and explain so well what you meant.
As a matter of fact I understood what you pointed out in your last message, but I was not sure that I got you right.
I agree with you, we must have always on mind the average rule that you explained so well and not take for granted the plain numbers!
Your posts were really important to help me to think about the matter in which I?m giving my first steps: and what you said is a sensible thing to keep in mind.
Anyway, as we can?t do the negative proof, I particularly think that only very high scores can be taken into account as a real possibility linking the studied fact with the presence of an ?astrosignature?. (I?m not saying that one cause the other, only that they are linked.)
So, as I put online to all in http://www.astrologiahumana.com/untitled-1.htm, there are significant scores in my study, and it?s worthy to observe in further situations if they are frequent in cases of accidental death.

Have a good weekend!

Clelia

Thanks, Clelia! And the same to you!

Looking at the printed table on Page 2 of your 'Factor Analysis Report Summary' pdf file, I am trying to work out what is meant by the abbreviations in each of the columns. Not having ever used Astrodatabank forces me to infer from appearances, so I'm going to type what it looks like to me, and if I'm wrong, please correct me:

WtAveEx = Weighted (?) Average in Experimental Group - seems to represent the incidence of the placement shown at the start of the row in question in the experimental group, expressed as a decimal fragment of the totality of the population in the same group, this totality being taken arbitrarily as having the value 1. In other words, it is equivalent to a percentage divided by 100.
WtAveCt = Weighted (?) Average in Control Group - likewise a percentage divided by 100, but for the control group.
WtAveDi = Weighted (?) Average Difference - this, as we had both previously surmised from the appearance of comparable figures throughout the cells of the first table in your report that you originally linked to, is the percentage difference between the values in the two foregoing columns for the row in question. This data is precisely equivalent to the data I labelled in the row in my table 'DIF%'. The fact that the values in the two foregoing columns have in this instance been expressed as decimal fractions of 1 instead of as percentages makes absolutely no difference to the outcome of the calculation in this third column.
AveExp / AveCtrl / AveDiff: the cells in these columns contain exactly the same results as those in the first three described above, in the same respective order, indicating that whatever purpose the weighted equivalents may sometimes be used for has not been put into use in this case.
%NZExp: I have no idea what the 'NZ' stands for, but regardless, this column contains the percentage equivalent of the data in WtAveEx, ie the same data multiplied by 100, only in this instance it is expressed to four significant figures instead of 2, so it is more accurately recorded. Personally I prefer my simple designation 'AG%' in description of the equivalent data for the Accident Group in your study, but EX% would do just as well, and both would be simpler than '%NZExp'.
%NZCtrl: Again, I have no idea what the 'NZ' stands for, but this data is the percentage equivalent expressed to four significant figures of the data in WtAveCt, and is equivalent to the data in the CG% row in my table.

Thanks for linking to this table as it's really helped confirm how the program is working as well as to understand the abbreviations it uses!

I have much more to write still in response to your earlier points at a later stage (looking forward to a nice, relaxing family outing to the park this afternoon!), but I'm really glad this conversation has been inspiring to you.

Enjoy Sunday equally, and the same to anyone and everyone else reading this!

Philip