Can Astrology Predict Workplace Accidents?

[Malerin]

Hi, all. First time posting here. I found this interesting doctoral thesis from 1992. Don't know if anyone's taken a look at it.

The researcher's hypothesis is that, if astrology is accurate, accidents are more likely to occur during "hard" aspects (opposition, conjunction, square) in a person's chart. Since the only information she had were the dates of workplace injuries, and the birthdates of the persons injured, she went by "hard" aspects of the sun sign. In other words, if astrology is accurate, people should be getting injured more on their birthday, three months prior and after their birthday, and six months after their birthday.

Her sample size was 1005 documented cases of workplace injuries. I'll quote her conclusion:

"The hard aspect categories (1, 4, 7, 10) are well above the expected value, while the two lowest injury categories are 3 (55) and 12 (55). The Chi-Square test value for this sample with 11 degrees of freedom is 107.728 and the statistical significance of this distribution is P< .0000000001."

http : // safire.net/sara/results. html

There are obviously several conclusions that can explain this:

1. Her math is wrong. Very unlikely, if you read the paper. She earned her doctorate from this study and a basic statisitical error would have been spotted immediately.
2. She "massaged" the data to get the results she wanted. She was biased going in and deliberately excluded cases that didn't fit her hypothesis (or chose cases that did).
3. She got lucky with the sample size. Not very likely with a sample size over 1,000 and such an extreme probability result.
4. Astrology has predictive value regarding work place accidents.

I'm betting a lot of you are going to with (2). That's skeptical, right? I mean, how do we know the sample she worked with was really random? But that begs the following question: If this were a dissertation on a mundane subject (e.g., the affect of soft music on standardized testing), would anyone be questioning her integrity?

Should she get the benefit of the doubt as an honest researcher? Or is her conclusion so radical that her integrity is immediately called into question? But if that's so, if any researcher who "proves" an extraordinary claim falls under the microscope for intellectual (dis)honesty, doesn't that imply a very strong bias towards materialism and science? Shouldn't we be skeptical about things we take for granted? How far down the skeptical rabbit hole should you go?

I contacted the rearcher recently and asked her if any follow-up studies were ever done. She knows of one done years ago for an astrology journal with inconclusive results. Her paper might as well have been swallowed by a black hole.

[SezMe]

Welcome to the fora, Malerin. Here is your link:

http://safire.net/sara/results.html

Now I'll go take a look.

ETA: Responding to your post alone, not the information in the link, does it not appear preposterously obvious to you to assert that most accidents occur over a nine-month(!) period around a person's birthday. How could that NOT be the case?

[Malerin]

Welcome to the fora, Malerin. Here is your link:



ETA: Responding to your post alone, not the information in the link, does it not appear preposterously obvious to you to assert that most accidents occur over a nine-month(!) period around a person's birthday. How could that NOT be the case?

I don't think I was clear. Her hypothesis is that accidents would occur ON the 3rd, 6th and 9th month, and birthday. Four months out of the year.

[AntiTelharsic]

She earned her doctorate from this study

And from the "University for Humanistic Studies", no less!

[Malerin]

And from the "University for Humanistic Studies", no less!

Does that really have any bearing? There's a link to her study, which you can read, and she obviously is well versed in in-depth statistical research. You should respond to the validity (or non-validity) of her research.

If you like, I can trot out renowned Harvard-trained neurosurgeon Dr. Allan j. Hamilton ("The Scalpel and the Soul"), who claims to have been healed by an Indian Shaman, and also claims to be able to predict when people are going to die. Are those claims MORE or LESS valid because he went to Harvard?

[Arkayik]

Welcome to the forum Malerin, and as such, it bears much resemblance to its ancient Roman counterpart. New posters are seen as new blood in a sort of cranial-blood-sport...

Having said that, you can be assured that amidst the bramble of useless replies there will be nuggets of shining glory.

In any case, what are the odds that there will be a lot of accidents within 4/12ths of the year? Be much more interesting if it predicted the day, then the not-so-hapless victim could stay home...

Cheers,

Ark

[AntiTelharsic]

If you like, I can trot out renowned Harvard-trained neurosurgeon Dr. Allan j. Hamilton ("The Scalpel and the Soul"), who claims to have been healed by an Indian Shaman, and also claims to be able to predict when people are going to die. Are those claims MORE or LESS valid because he went to Harvard?

Whether or not he went to Harvard is irrelevant to those particular claims. The fact that he went to Harvard means that I can have some faith that his doctoral thesis was reasonably supervised and challenged, without needing to review it myself, if I were even qualified to, which I'm not.

But since the PhD in question was awarded by a now-defunct Californian woo-degree mill I can have no such faith, and since I'm not going to take the time to review the thesis myself I'm just going to write it off. If an objective and capable person were to analyze her results I might have something to go on, but I can't find anything about this thesis apart from on pro-astrology sites.

[Malerin]

Welcome to the forum Malerin, and as such, it bears much resemblance to its ancient Roman counterpart. New posters are seen as new blood in a sort of cranial-blood-sport...

Having said that, you can be assured that amidst the bramble of useless replies there will be nuggets of shining glory.

In any case, what are the odds that there will be a lot of accidents within 4/12ths of the year? Be much more interesting if it predicted the day, then the not-so-hapless victim could stay home...

Cheers,

Ark


You would need to do a natal chart for each person in the sample, and be able to interpret it to a high degree of confidence. For that, you need location and time of birth for each person (not just day), and a hell of a lot of time. Lacking that information, the researcher went with the only thing she had: date of birth.

Bring on the Nuggets!

[Foolmewunz]

Malerin,
We have our traditions, here.

So, firstly, Welcome to the JREF Forums.

Secondly, we sort of pride ourselves in being skeptics around here, and the blood sport (we all lettered in it at the Penn Jillette Holistic University of Sitting By the Fountain) is woo-bashing. So expect to be flayed if you can't provide anything better than what you've offered thus far.

Thirdly, toss it out on the stoop and see if the cat licks it up, okay? We've seen lots of newbies with ye olde favorite tactic,.... "Say, what do you guys think of this?" Then within two pages, it's blatantly obvious that this isn't an innocent question, it's a lead-in from a woo who's determined to show us ornery cynics a thing or two.

Fourthly,... Don't trifle with us, Malerin. Show us your beliefs and your interpretations. No one's reading that entire "dissertation" (there are rules in the public areas that prevent me from referring to it as what it is), so since you're so mightily impressed, show us your figures and not just a link to a Tree Hugging Crystal Tuning Woo Diploma Mill. Summarize the findings. That's the traditional call-down for someone who keeps blathering, "so read the article", "so watch the video", "so you didn't buy his book, huh".....

Don't ask us again to read it. It's incomprehensible. Doctoral thesis, my arse!

[Malerin]

If an objective and capable person were to analyze her results I might have something to go on, but I can't find anything about this thesis apart from on pro-astrology sites.

You're not an "objective and capable" person? You're unable to evaluate basic statistical research? You don't seem to have much of an opinion of yourself...

[AntiTelharsic]

You're not an "objective and capable" person? You're unable to evaluate basic statistical research? You don't seem to have much of an opinion of yourself...

I have better things to do.

[Malerin]

Malerin,
We have our traditions, here.

So, firstly, Welcome to the JREF Forums.

Secondly, we sort of pride ourselves in being skeptics around here, and the blood sport (we all lettered in it at the Penn Jillette Holistic University of Sitting By the Fountain) is woo-bashing. So expect to be flayed if you can't provide anything better than what you've offered thus far.

Thirdly, toss it out on the stoop and see if the cat licks it up, okay? We've seen lots of newbies with ye olde favorite tactic,.... "Say, what do you guys think of this?" Then within two pages, it's blatantly obvious that this isn't an innocent question, it's a lead-in from a woo who's determined to show us ornery cynics a thing or two.

Fourthly,... Don't trifle with us, Malerin. Show us your beliefs and your interpretations. No one's reading that entire "dissertation" (there are rules in the public areas that prevent me from referring to it as what it is), so since you're so mightily impressed, show us your figures and not just a link to a Tree Hugging Crystal Tuning Woo Diploma Mill. Summarize the findings. That's the traditional call-down for someone who keeps blathering, "so read the article", "so watch the video", "so you didn't buy his book, huh".....

Don't ask us again to read it. It's incomprehensible. Doctoral thesis, my arse!


Um, I believe I DID summarize the findings (even included a direct quote from teh "results" section of the paper). Do you want me to spoon-feed you? The "Methodology" and "Results" sections are about 5 pages long, total. If you can't bother to spend 10 minutes on it, I don't think you're going to be able to contribute much.

[Malerin]

I have better things to do.

Don't we all. But this is a skeptics forum, and I have presented evidence that points to a rather bizzare conclusion.

Is it a common practice here to dismiss evidence without even a cursory glance at it?

[AntiTelharsic]

Don't we all. But this is a skeptics forum, and I have presented evidence that points to a rather bizzare conclusion.

Good for you. That obliges me and my time how?

Is it a common practice here to dismiss evidence without even a cursory glance at it?

I can't speak for everyone. Look, if this had anything going for it -- the thesis coming from a reputable school, a study published in a peer-reviewed journal, someone else repeating the results, mention anywhere other than some dozen pro-astrology sites -- then I might be interested. But it doesn't seem to have anything going for it, so I'm not interested in spending my time trying to convince myself that some random piece of apparent woo has merit. If someone else is willing to do it, however, someone objective and qualified, then I'll be happy to consider what that person has to say. In the meantime, I have no more reason to take it any more seriously than I do any of hundreds of crackpot websites.

[Foolmewunz]

Um, I believe I DID summarize the findings (even included a direct quote from teh "results" section of the paper). Do you want me to spoon-feed you? The "Methodology" and "Results" sections are about 5 pages long, total. If you can't bother to spend 10 minutes on it, I don't think you're going to be able to contribute much.

You're right, I can't be bothered with it. If this is something for math nerds only, post it in Math and Science. If you can translate the Chi Factor divided by square root of your sister in law's bra size, then maybe some of us lesser intellects could comment. I am not going to spend hours trying to define the terms which you, in your infinite and obvious wisdom, could summarize for us within about thirty seconds.

Of the 1000 subjects, how many had disabling accidents on those four "cardinal" points she mentions? How wide an arc (in days) is this 5/10/15% nonsense she's talking about (yes, I actually tried to read parts of her paper, earlier), etc....

Maybe the paragraph you cited makes sense to statisticians or physics majors. To me, a mere working-class stiff, it's all gobbledygook. And I suspect that may be why her dissertation passed with flying colors. Professer Dumbledore couldn't figure out the math so it was sufficiently wooish that he gave her an A.

Hegel: There is nothing to be more abhorred than the genius who makes us strain to understand his every utterance.

[Malerin]

Good for you. That obliges me and my time how?



I can't speak for everyone. Look, if this had anything going for it -- the thesis coming from a reputable school, a study published in a peer-reviewed journal, someone else repeating the results, mention anywhere other than some dozen pro-astrology sites -- then I might be interested. But it doesn't seem to have anything going for it, so I'm not interested in spending my time trying to convince myself that some random piece of apparent woo has merit. If someone else is willing to do it, however, someone objective and qualified, then I'll be happy to consider what that person has to say.

In 1905, a Swiss patent clerk with no PHD revolutionized physics. Not that this woman is Einstein, but the point is that groundbreaking research doesn't always carry an Ivy League stamp. But since you can't be bothered to spend 10 minutes analyzing it, I trust you'll not clutter up the thread any more?

[AntiTelharsic]

groundbreaking research doesn't always carry an Ivy League stamp.

That's why I listed several ways her research could have gained some credibility.

I trust you'll not clutter up the thread any more?

Hey, I just pointed out that it came from a degree mill, and that was a perfectly reasonable contribution. You're the one who turned it into an argument about how you think I'm supposed to spend my time. Feel free to let it go whenever you want.

[Coveredinbeeees]

In 1905, a Swiss patent clerk with no PHD revolutionized physics. Not that this woman is Einstein, but the point is that groundbreaking research doesn't always carry an Ivy League stamp. But since you can't be bothered to spend 10 minutes analyzing it, I trust you'll not clutter up the thread any more?

That Swiss patent clerk published his groundbreaking papers in Annalen der Physik. If he had instead published them in Gardener's Monthly or used them as a doctoral thesis to graduate from Professor Trotter's Mail in Degree Emporium then the prominent scientists of the time could have been forgiven for not paying much attention. As it was he published in in a peer reviewed journal of considerable renown, thus generating interest in his astounding conclusions.

If you were linking to a thesis from a university of repute or to an article in Nature I would take the time to read.

[Mojo]

Does that really have any bearing?


Since you've argued that the fact that she earned a PhD for the work means that the maths is correct, it is valid to question the quality of the awarding institution. It was your argument from authority.

[Malerin]

Since you've argued that the fact that she earned a PhD for the work means that the maths is correct, it is valid to question the quality of the awarding institution. It was your argument from authority.

True, unless all the math is right there, in the paper (which it is). I assumed that members in a forum dedicated to skepticism would have at least a passing knowledge of elementary statistics. You don't need a degree in math to understand a Chi-square test.

http:// en.wikipedia.org/ wiki/ Chi-square_test

[erlando]

:popcorn2

[Mojo]

True, unless all the math is right there, in the paper (which it is). I assumed that members in a forum dedicated to skepticism would have at least a passing knowledge of elementary statistics. You don't need a degree in math to understand a Chi-square test.

http:// en.wikipedia.org/ wiki/ Chi-square_test


Here's what you posted about the validity of her maths:1. Her math is wrong. Very unlikely, if you read the paper. She earned her doctorate from this study and a basic statisitical error would have been spotted immediately.

[SezMe]

http:// en.wikipedia.org/ wiki/ Chi-square_test
Malerin's url (http://en.wikipedia.org/wiki/Chi-square_test)

[CFLarsen]

The problem is first and foremost with the data.

This sample included birth dates and dates of injury for each subject.
...
This sample excluded all cuspal birth dates and cuspal dates of injury before processing. (Cuspal dates are dates when the Sun is changing Sun sign, and because no exact birth times were available, cuspal dates were eliminated in all samples)

Oops. Already, we are facing a major problem: Astrological charts have to have a birth time, too - otherwise, the horoscope will not be accurate.

So, from the start, we are not looking at actual astrological data, but some imaginary "birth date".

It should be noted that this researcher uses cuspal dates for both birth and injury: That means that the researcher thinks we can cast a horoscope for an injury.

It should also be noted that, while this researcher uses cuspal dates because the actual time of birth is unknown, many other astrologers use 12:00 (noon). That means that it will be this particular researcher's idea of what astrology is that is tested.

Now, the question that immediatedly comes to mind is: Why would the researcher choose this particular method? Is it because she was so advised (and on what grounds?) - or was it because the data fit the hypothesis?

Sample B consisted of 609 subjects obtained directly from a Los Angeles clinic that evaluates and treats Hispanic (mainly Mexican and some El Salvadorean) injured workers. (All evaluations were done with the help of a Spanish speaking interpreter). A clinic staff member copied the required information from the clinic's files according to the same instructions that the medical-legal transcribers were asked to follow, with one addition: A separation into male and female subjects was recorded. Injury years for this sample ranged from 1981 to 1992. At no time did the researcher meet or see any of the subjects for either sample. Sample B was divided into Sample B/Females (N=126), and Sample B/Males (N=483). Six cuspal birth dates were eliminated from the female population of Sample B for the analysis by Sun sign, thus using only 120 females for that particular analysis. All other computations were done on the entire sample. From Sample B/males, 13 cuspal birth dates were eliminated for the same analysis, using only 470 subjects for the Sun sign distribution.

That's a lot of data massaging. Not only are we looking at several datasets over different timeframes, we are also suddenly having a major difference in gender distribution. We are given no reason as to why 6 cuspal birth dates were eliminated from the female group and 13 from the male group.

A replication sample of 55 subjects (Sample C) was obtained during the process of reviewing the literature, as C.E.O. Carter (1932) provides birth dates and dates of injury for 55 of his reported 168 cases. These were all birth dates with exact birth times, so no cuspal deletions were needed. All injury dates were prior to the year 1929.

Wait a moment. Where did she get those 55 subjects from? How come she suddenly knows their exact birth times, if she retrieved that particular dataset from clinics treating injured workers?

A control group of completed suicide subjects (Sample D) was provided by David Lester (1987). These were recorded suicide subjects who committed suicide in Philadelphia during the year of 1982. A total of 206 subjects had full birth dates and dates of suicide and were included in the sample. However, 45 of these were cuspal birth dates (21) or cuspal suicide dates (24),and were eliminated from the Sun sign computations. They were, however, included in all other computations. No separation into male/female sub-groups was conducted for this sample, as it was used as a control for the entire research sample of 1023 subjects (Sample A + B).

This is a classic example of how data can be fuzzied. Why suicide subjects? That has nothing to do with work accidents. And why eliminate those with cuspal birth dates or cuspal suicide dates, if the group investigated in the sample were all cuspal birth dates and cuspal accident dates? Yet, all of those eliminated were included in all other computations. Why? Why not separate in this case? That's not how you have a control group: The only thing that should distinguish the control group from the group with the purported effect is that the effect must be missing from the control group.

And so on.

[Foolmewunz]

I don't understand statistical analysis. Besides, they promised me there would be no math on this and that they'd grade on a curve.

What I do understand, though, is that Google Is Our Friend.
Someone in 2001, responding to a laudatory plug for the exact same study, and purporting to be Sarah Klein Ridgley, PhD.
(I have vetted the contacts... Dobyns and Pottenger are fellow woos who work with Ridgley on occasion.)

http://www.astrodatabank.com/ASWorkRelatedInjuries_FB.htm

So, apparently Doc Sarah might have some scruples. Seems the other randomly acquired set of data, with a much better defined sampling, came up with completely different results. A) It doesn't confirm the 1,4,7,10 dominance as the first test did. B) It actually seems to refute that data, entirely.

[Since writing my thesis in 1992,] we have acquired a set of data from the Swedish Government and did a replication and did not find anything like the results of my original study. Zip Dobyns, Mark Pottenger and I have been working on the data and found very little significance at all to the solar transits. The results have not been written up in an article yet, but Mark Pottenger gave two lectures on this effort at the International Astrology Conference in Plymouth, UK in August 1999.
We obtained a large database of injured workers from the Swedish Government's equivalent to our Workers' Compensation Board. We had close to 3000 individuals, of which over 2400 were TIMED accidents (i.e., we had the exact time of accident). We used this subsample in order to be able to use the Moon. As all of these people were Swedish, they all lived in the same time zone, so we could do that.

What was found was a SMALL significant effect to hard aspects between Mars and Saturn (P>.03)

Mars Saturn aspects were found also on the original study (the California Sample) when we re-analyzed it.

The Swedish sample showed no effects from the transiting Sun to the natal Sun either at the opposition or the square. However, a reversed effect was significant statistically on the conjunction: i.e., the LEAST amount of accidents occurred around the birthdate!. We found out that in Sweden, people get their birthdays off!! So they actually were not at work on their dates of birth.

Generally we feel that the difference in results stems from very different cultural values and psychological attitudes toward work, people in places of authority and the individual's self worth.

Obviously, more replications are needed, and we are constantly looking for additional samples that will fit our criteria to do such replications.

Sara Klein Ridgley



Malerin,

Buh-bye! Thanks for playing. Sorry that we have no lovely parting gifts for you, but being summarily PWND in your first few hours here disqualifies you from even receiving the home version of Stop That Woo.

[Ysidro]

In 1905, a Swiss patent clerk with no PHD revolutionized physics. Not that this woman is Einstein, but the point is that groundbreaking research doesn't always carry an Ivy League stamp. But since you can't be bothered to spend 10 minutes analyzing it, I trust you'll not clutter up the thread any more?

Oooh, all I need is "Quantum" and I can call "WOO BINGO!"

[Jaggy Bunnet]

It should be noted that this researcher uses cuspal dates for both birth and injury: That means that the researcher thinks we can cast a horoscope for an injury.

I don't think so. It seems to be that she has excluded "cuspal dates" for either birth or injury on the grounds that as the sun is changing sign on those dates, you need to know the exact time of birth to know whether it is the sign it is leaving that is relevant or the one it is moving into.

That's a lot of data massaging. Not only are we looking at several datasets over different timeframes, we are also suddenly having a major difference in gender distribution. We are given no reason as to why 6 cuspal birth dates were eliminated from the female group and 13 from the male group.

If my interpretation is correct, it would be because there were 6 cuspal dates in the female group and 13 in the male - i.e. ALL cuspal dates are eliminated, the number is dependent only on how many were in the original population.

Wait a moment. Where did she get those 55 subjects from? How come she suddenly knows their exact birth times, if she retrieved that particular dataset from clinics treating injured workers?

In the bit you quoted, it states "(Sample C) was obtained during the process of reviewing the literature, as C.E.O. Carter (1932) provides birth dates and dates of injury for 55 of his reported 168 cases. These were all birth dates with exact birth times, so no cuspal deletions were needed."

I think this answers both your questions, but for me it also raises a further one - was the time of injury known? If not, then I would have expected there to be cuspal deletions on the basis that you need to know both the time of birth AND the time of injury to be certain that you know which sun sign is relevant for cuspal dates. As it only states that the birth times are known, I assume that injury times are not. It is therefore incorrect to state that no cuspal deletions were needed purely because birth times were known.

This is a classic example of how data can be fuzzied. Why suicide subjects? That has nothing to do with work accidents. And why eliminate those with cuspal birth dates or cuspal suicide dates, if the group investigated in the sample were all cuspal birth dates and cuspal accident dates? Yet, all of those eliminated were included in all other computations. Why? Why not separate in this case? That's not how you have a control group: The only thing that should distinguish the control group from the group with the purported effect is that the effect must be missing from the control group.

And so on.

Completely agree.

[Malerin]

Buh-bye! Thanks for playing. Sorry that we have no lovely parting gifts for you, but being summarily PWND in your first few hours here disqualifies you from even receiving the home version of Stop That Woo.\

I admitted in my first post that there was another study done with inconclusive results. Which was admitted by the resarcher herself. So what we have is one study that produced an extremely odd result, and another one in a different country (where people didn't even work on birthdays) that produced a normal result.

As the researcher of the first study put it:

"Generally we feel that the difference in results stems from very different cultural values and psychological attitudes toward work, people in places of authority and the individual's self worth.

Obviously, more replications are needed, and we are constantly looking for additional samples that will fit our criteria to do such replications."

The odds of her getting the results she did in the first study were over a million to 1. If she didn't play fast and loose with the data, that deserves a follow-up study with a larger sample size.

[Civilized Worm]

By all means, go ahead and replicate it. Until then I remain sceptical.

[Foolmewunz]

\

I admitted in my first post that there was another study done with inconclusive results. Which was admitted by the resarcher herself. So what we have is one study that produced an extremely odd result, and another one in a different country (where people didn't even work on birthdays) that produced a normal result.

As the researcher of the first study put it:

"Generally we feel that the difference in results stems from very different cultural values and psychological attitudes toward work, people in places of authority and the individual's self worth.

Obviously, more replications are needed, and we are constantly looking for additional samples that will fit our criteria to do such replications."

The odds of her getting the results she did in the first study were over a million to 1. If she didn't play fast and loose with the data, that deserves a follow-up study with a larger sample size.

And she did the larger study with threefold the sampling, and with exact times of accidents and came to the conclusion, stated above that the sun sign had no correlation with the accidents. That was in 2000 or 2001, and she hasn't followed up on it.

Why would that be? Does it occur to you that it is just quite possible that the odds on her achieving the first results mean that it's just a fluke, a coincidence.

I truly doubt 1 mio : 1 odds - with 1000 samples and only hoping for results to be placed in four out of 12 signs, they seem considerably smaller, but as I said, I don't do math... But as we've all noted the odds of, say, a small gasket in the space shuttle causing a complete failure are about 22,000,000 to 1 if measured a certain way (prior to said event actually happening), yet it happened. Coincidences and anomalies do occur. That's why follow up controlled tests are necessary.

I'd also ask why there were 18000+ accidents in California in the year of her study and she only sampled the 1000. That table's in her appendix, and she makes no mention of why those other 17000 were not included. Is it possible that this is just an anomaly? Where and how did she get the sampling? Had someone stumbled across this lovely anomaly and fed her the results as something worth looking into.

If you knew of the further contradictory study, why did you trot out the one with all the woo woo sprinkles on top? Confirmation bias or confirmation desperation?

My compliments to "Dr." Ridgley. I don't often say that about homeopaths, and I limit it. She could at least take down the site with the crap results and hasn't done that. But admitting that their later study sort of quashes the first one is somewhat reputable.

What had my spidey sense tingling was that such a study would be slathered all over the place in the world of woo if it held up. I think it's possible that Ridgley has told others that the results don't stand up. Her letter was in about the second or third site that I found in Google. She may have sent similar corrections and disclaimers to others flaunting the study. Can't say.

[Malerin]

And she did the larger study with threefold the sampling, and with exact times of accidents and came to the conclusion, stated above that the sun sign had no correlation with the accidents. That was in 2000 or 2001, and she hasn't followed up on it.

Why would that be? Does it occur to you that it is just quite possible that the odds on her achieving the first results mean that it's just a fluke, a coincidence.

I truly doubt 1 mio : 1 odds - with 1000 samples and only hoping for results to be placed in four out of 12 signs, they seem considerably smaller, but as I said, I don't do math... But as we've all noted the odds of, say, a small gasket in the space shuttle causing a complete failure are about 22,000,000 to 1 if measured a certain way (prior to said event actually happening), yet it happened. Coincidences and anomalies do occur. That's why follow up controlled tests are necessary.

I'd also ask why there were 18000+ accidents in California in the year of her study and she only sampled the 1000. That table's in her appendix, and she makes no mention of why those other 17000 were not included. Is it possible that this is just an anomaly? Where and how did she get the sampling? Had someone stumbled across this lovely anomaly and fed her the results as something worth looking into.

If you knew of the further contradictory study, why did you trot out the one with all the woo woo sprinkles on top? Confirmation bias or confirmation desperation?

My compliments to "Dr." Ridgley. I don't often say that about homeopaths, and I limit it. She could at least take down the site with the crap results and hasn't done that. But admitting that their later study sort of quashes the first one is somewhat reputable.

What had my spidey sense tingling was that such a study would be slathered all over the place in the world of woo if it held up. I think it's possible that Ridgley has told others that the results don't stand up. Her letter was in about the second or third site that I found in Google. She may have sent similar corrections and disclaimers to others flaunting the study. Can't say.

I don't think she did the 2nd study. Wasn't that another group in Europe? I believe she was just commenting on the results.

[Piggy]

In 1905, a Swiss patent clerk with no PHD revolutionized physics. Not that this woman is Einstein, but the point is that groundbreaking research doesn't always carry an Ivy League stamp. But since you can't be bothered to spend 10 minutes analyzing it, I trust you'll not clutter up the thread any more?

You need to read up on Einstein.

You make it sound like he was an uneducated day jobber at the time of that publication.

He wasn't.

As for this:

There are obviously several conclusions that can explain this:

1. Her math is wrong. Very unlikely, if you read the paper. She earned her doctorate from this study and a basic statisitical error would have been spotted immediately.
2. She "massaged" the data to get the results she wanted. She was biased going in and deliberately excluded cases that didn't fit her hypothesis (or chose cases that did).
3. She got lucky with the sample size. Not very likely with a sample size over 1,000 and such an extreme probability result.
4. Astrology has predictive value regarding work place accidents.

I'm betting a lot of you are going to with (2). That's skeptical, right?

No, that's not skeptical.

No skeptics worth their salt are going to jump at option 2.

What we will do is to consider the evidence.

And yes, it makes a difference which school the work was done for, because standards vary greatly. There are plenty of creationist papers, for example, that come out of non-rigorous institutions which would never have passed muster at, say, an accredited state university.

[Piggy]

True, unless all the math is right there, in the paper (which it is). I assumed that members in a forum dedicated to skepticism would have at least a passing knowledge of elementary statistics. You don't need a degree in math to understand a Chi-square test.

http:// en.wikipedia.org/ wiki/ Chi-square_test

Not all of us are mathematicians.

My doctorate is in English, and my master's degrees are in Spanish lit/lang and Internet project management.

I use stats in my job, but only for split-run marketing tests.

If you understand what it is you're citing, it would help me greatly if you could summarize and provide a little defense.

[Piggy]

The problem is first and foremost with the data.

<snip>

And so on.

Nice. Very nice.

So much for the "they would have spotted it" hypothesis!

[Brendy]

You forgot a huge problem with the data.

Cause and Effect.

Just because there is a correlation does not mean there is a cause and effect relationship.

Maybe people are more careless on their birthdays because they are hungover from the celebration the night before. ect..

So many people make the cause and effect mistake and I think this study is one of them.

[CFLarsen]

I don't think so. It seems to be that she has excluded "cuspal dates" for either birth or injury on the grounds that as the sun is changing sign on those dates, you need to know the exact time of birth to know whether it is the sign it is leaving that is relevant or the one it is moving into.

If my interpretation is correct, it would be because there were 6 cuspal dates in the female group and 13 in the male - i.e. ALL cuspal dates are eliminated, the number is dependent only on how many were in the original population.

The problem is that she includes cuspal dates in one set (because she doesn't have the correct "birth" time), but leaves them out in another.

In the bit you quoted, it states "(Sample C) was obtained during the process of reviewing the literature, as C.E.O. Carter (1932) provides birth dates and dates of injury for 55 of his reported 168 cases. These were all birth dates with exact birth times, so no cuspal deletions were needed."

Could be. It's not particularly clear.

I think this answers both your questions, but for me it also raises a further one - was the time of injury known? If not, then I would have expected there to be cuspal deletions on the basis that you need to know both the time of birth AND the time of injury to be certain that you know which sun sign is relevant for cuspal dates. As it only states that the birth times are known, I assume that injury times are not. It is therefore incorrect to state that no cuspal deletions were needed purely because birth times were known.

The same problem again: She fiddles around with cuspal dates, without any single method.

[Foolmewunz]

I don't think she did the 2nd study. Wasn't that another group in Europe? I believe she was just commenting on the results.

So when she wrote (bolding mine)

[Since writing my thesis in 1992,] we have acquired a set of data from the Swedish Government and did a replication and did not find anything like the results of my original study. Zip Dobyns, Mark Pottenger and I have been working on the data and found very little significance at all to the solar transits.

you think she isn't saying that she/we did the study? Sounds pretty well worded to me.

[Malerin]

You forgot a huge problem with the data.

Cause and Effect.

Just because there is a correlation does not mean there is a cause and effect relationship.

Maybe people are more careless on their birthdays because they are hungover from the celebration the night before. ect..

So many people make the cause and effect mistake and I think this study is one of them.

She covered that possibility. It's a stretch to think that people are hungover from celebrating hard aspects of their sun sign (birthday, 3 months before and after, and 6 months after). Very few people even know what a hard aspect is.

[Malerin]

So when she wrote (bolding mine)



you think she isn't saying that she/we did the study? Sounds pretty well worded to me.

Yeah, you're right. I was thinking of the email she sent me. Made it sound like it was another group. I had asked her if anyone had tried to replicate her results.

[TheDaver]

She covered that possibility.
No, she didn’t. She reduced its effect, but it’s still definitely there.

[politas]

So, if I'm understanding this right, a study was done which showed statistical clumping of over 50% of results within a range including 33% of the possible results, and this is considered significant?

Was there any reason to predict the particular ranges discovered?

Or is this just a case of finding some odd data and pointing it out without any attempt at creating a hypothesis?

In any case, since further studies did not show similar statistical clumpng, it's all pretty irrelevant, isn't it?

Move along, nothing to see.