ArticlesIn 1995, I read a scientific paper in a journal – a real, live, on-the-shelf-in-a-library journal – that was to become the article that I have cited the most, ever. Unfortunately, I have lost the actual reference, which disappoints me greatly.
The author started off by making two simple assumptions.
The first assumption was that zodiac sign under which you are born significantly affects your likely personality.
This is obviously controversial assumption, but please, just accept it for now.
Note, though, that this is a very broad assumption. It is not saying that stars predict your future. It is not saying that all Leos are argumentative or any other specific trait. It is not saying that all Aries are the same. It is not offering a mechanism by which the stars affect your personality. It is not ruling out that personalities might be influenced by societal pressure to conform to assumptions about stars-signs. It is just assuming that, in some general way, astrology has some basis as a method of categorising personalities.
The second assumption was that personalities significantly affect job choice.
This is a fairly innocuous assumption. I can look around at the people in several industries that I am familiar with, and see that there are clear biases in the personality populations.
With these two assumptions, it should be clear that if you were to compare the distribution of birthdays of the general population (which is a fairly even distribution, but there are biases towards certain times of the year) against the distribution of birthdays of, say, baseball players, you might expect to see a significant difference. It isn’t clear from the assumptions which direction the bias(es) might be, but there should be at least one. Of course, you would need to have a reasonably large sample to be sure.
At this point, the author revealed that he had access to a publicly-available database of the personal statistics of 10,000 US professional baseball players, including their dates of birth. So he compared them and discovered that there was (oddly enough) no statistical difference.
Maybe this was some unexpected anomaly related to baseball, but no matter, because he also had access to a database of medical doctors. Again, there was no statistical difference.
So, he tried again, with a database of lawyers… and again, and again, each time with different professions that, it seemed clear, would be inviting to different personality types.
Each time, he found no significant difference between these large samples and the general population.
One of the assumptions that he started with must be wrong. It seems pretty clear which one it must be.
I found this paper quite delightful, for two main reasons.
I had heard (and made) attacks against astrology in the past that had:
- questioned the size of the impact that distant stars could have, explaining how the effects of gravity, magnetism, etc. dissipate over distance
- explained how the stars in each zodiac symbol are nowhere near each other, and that the shapes of the Zodiac are human interpretations, based on our location in space.
- explained how the Sun now “rises” through the zodiac symbols during different times of the year, compared to dates determined by Ptolemy.
- described how confirmation bias, Forer’s effect and wishful thinking add up to explain the perceived effect of astrology.
- debunked how modern “astrologers” actually produce their predictions.
This paper, however, showed that there was no need to muck around with the intricacies of astronomy, gravitation or psychology. You don’t need to debunk the proposed mechanisms or use Occam’s Razor to find better explanations of the effect that stars apparently have on our personality if you can simply show that there is no effect to explain.
The second reason I found the paper delightful was the summary. The author explained that he had been very careful with all of the statistics, checking all of his figures and methods very carefully. He had considered both sides of the debate and provided a balanced and measured argument. He had been fair and without bias. His evidence of this?
“Because I am a Libran, and that’s how Librans are.”
Comment by Sunny Kalsi on March 9, 2007
Playing devil’s advocate, I could argue that the second assumption is the one that is incorrect. The evidence?
House!
If you’ve ever watched the show you’ll realise that there are many different personalities in the show. Unfortunately, if someone said to me “the job you choose is probably not personality selective.” I would not really be able to argue that. You may share a skill-set, but that doesn’t mean that you don’t apply it differently, or that the subset of skills you excel in within your job is due to your personality. The problem is that jobs are too varied and complex, and have a bunch of extra factors (like opportunities, wealth, etc.) that influence a person’s job choice (e.g: someone becoming an accountant but really wanting to be a musician).
I think something like “favourite fruit” might offer a better test, but not necessarily your job.
Comment by Julian on March 10, 2007
Way to pick a poor example, Sunny! 🙂
Firstly, Even if I ignore the fact that show is fiction (and good fiction writers ensure that they have a diverse set of characters in order to play them off against other), that still leaves the fact that many of the characters are so similar as to be interchangeable. The characters Foreman, Chase and Cameron are all the same: young, arrogant, competitive smart-alecs, who take turns to suffer the pains of self-doubt, before pulling themselves together and regaining their pride. Sure, there are subtle shades of difference between the characters, but you can bet that they are all Taurus! 🙂
So, let’s skip your example, and target your real point.
Hmmm… the term “personality selective” sounds too strong to me; I want to avoid being pinned behind a strawman. No-one is saying that all people in an industry have the same personality. I am asserting that it is clearly the case that there is significant personality clustering/bias/trends [You can choose the least-absolute word here] within certain jobs.
Can you dispute for a moment that there are a preponderance of nerds within software engineering, as just one example? (While I hate to quote MBTI, its practitioners somewhat supports this claim by suggesting the software industry has a higher percentage of INTP personalities than in the wild.) One example is all I need to prove this point, which is a shame because I could happily cite similar biases in the population of lawyers, doctors, the military, circus performers, teachers, engineers, musicians, marketeers, etc.
While there are many diverse factors that affect career choices, and many diverse people in each role, many jobs have a significant personality bias too.
Comment by Aristotle Pagaltzis on March 10, 2007
Why do you hate to quote the MBTI?
[Ed: Oops, I missed Aristotle’s subtle correction here. I have been misspelling MBTI. Fixed.]
Comment by Julian on March 11, 2007
There you go again, Aristotle. I’d need to write an entire article to answer that question and articulate my love/hate relationship (correction: tolerate/hate relationship) with MBTI.
In short, if the science behind the statistics that claimed the software industry has a high number of INTPs is as weak as the science behind the basic model, then I should be embarrassed to quote them.
Comment by Jonathon Duerig on March 18, 2007
Umm… You failed to cite it. After reading about the paper, I want to read the paper itself. What journal was it from? What year? What was the title? Any of these things might be enough to find it, but I haven’t been able to find it based just on your description.
-D
Comment by Jonathon Duerig on March 18, 2007
By the way, I realize that you have lost the actual reference. But if you have any other clues that might help me find it, I’d be much obliged.
Comment by Julian on March 22, 2007
Jonathon,
I spent about an hour both hunting and trying to remember the circumstances in which I read it.
I believe I read it in February 1995, so that obviously cuts out anything since then. I believe it was written in the mid-70s. So we only have 20-odd years of science articles to look through. That couldn’t be more than a few hundred million, could it?
I have been racking my brain trying to imagine what I read that might have referenced that article in the first place. What skeptic material was I reading 12 years ago?
Metamagical Themas has been an old favourite, but looking through its index, I see nothing related to it. I was interested in the Journal for Irreproducible Results and the Annals of Improbable Research around that time, but I am not convinced either would have sent me there.
Ahhh… but could have been a reference cited in a newsgroup, so…
It was found in one of the libraries of University of Queensland, so if you can hack in there and find my old borrowing records, problem solved!
Sorry I couldn’t be more help.
Comment by Jonathon Duerig on March 22, 2007
A possible hit:
“Statistical Test of Sun-Sign Astrology, A”
Skeptical Inquirer 1(2)49-54
SPRING/SUMMER 1977
McGervey
I’m not certain because this issue isn’t one of the ones at my library. I’ll make an ILL request for it.
Positive indicators: Title, Skeptical Inquirer, at University of Queensland, Skeptical Inquirer was mentioned in a column of Metamagical Themas.
I’ll let you know when I’m sure.
-D
Comment by Julian on March 24, 2007
Jonathon,
Wow, when I read this, I got excited. Two days later, I am still excited.
Your guess sounds like a good one. Your logic is very plausible; I probably did read Metamagical Themas (again), and think “Hey, let’s go to the library and check out the Skeptical Inquirer.” I’m not sure why I would select 1977 as a promising year to read, but I am glad I did!
I am on tenterhooks to find how accurately I remember the article.
I was thinking I should have a scoring system, rating me out of ten. I lose two points if the first group wasn’t baseball players. I lose two points if the author wasn’t Libra. I lose one point if the other groups didn’t include doctors and lawyers. I lose one point if the sample size was not around 10,000. I lose another if it is an order of magnitude out. I lose ten points if this is the right paper, but it draws a completely different conclusion about astrology – e.g. that it works.
Then I realised that, given I had forgotten the title, author and journal, I had already failed this memory test. 🙁
Comment by Jonathon Duerig on March 30, 2007
Ok. So I’ve just got a copy of the paper and read it. The most likely scenario is that you read this paper and seriously misremember it. However, it is possible that you read a more comprehensive paper by the same author.
If you did read this paper, it was probably because it appeared in the second issue ever of the magazine (and the last issue that it was called The Zetetic). So you could have read the Metamagical Themas article, went to the library, got the first book and flipped through the contents of the first few issues.
The overall idea of the paper was correct. It was about testing sun-sign astrology by linking it to occupation. However, The Skeptical Inquirer deosn’t appear to be a peer-reviewed journal. The author doesn’t mention his sign. There are only two groups mentioned: Scientists (as listed in American Men of Science) and Politicians (as listed in Who’s Who in American Politics). The sample sizes were 16,634 and 6,475 respectively. It compares these two with each other, and then shows that a trend that they both display (extra births in July, August, and September), is also present in the population at large.
The author’s full name is John D. McGervey. I did a quick google search and couldn’t find a bibliography for him, and he appears to be deceased. I will keep looking to see whether this was maybe a preliminary paper which led into a peer-reviewed paper that you may have read.
Whether or not you remembered this correctly, or whether you were remembering a similar paper, you should now be able to reference this source definitively.
-D
Comment by Julian on April 1, 2007
Jonathon,
Firstly, you are a hero. Thank you for hunting down the paper.
I am quite gutted at the result. So close, and yet so far. I was prepared to misremember bits and pieces, but for the author to not even mention his star sign makes my story so very, very wrong. At the same time, it sounds like the exact same analysis that I was describing.
I’ve been searching for McGervey via Google – his 1977 paper seems to be cited frequently enough, but I haven’t seen anything else related that he might have written later. (He did write a book called “Probabilities in Everyday Life”, but this seems an unlikely inclusion.)
No baseball players mentioned at all? That’s a surprising detail for me to conjure from nowhere. I am starting to wonder if I am conflating two different articles.
I haven’t found his date of birth to confirm whether he actually was a Libran or not!
I think I need to spend a few hours in the library; I’ve got to read this paper and see if it is at all familiar.
(By the way, I recall now that I bought my father a Scientific American subscription around 1994, and was catching up on a few recent back issues in December/January 1995. Perhaps I stumbled across a reference to an astrology article in there? The Metamagical Themas link makes sense, but this opens up another avenue.)