Validity of Graphology

Cecil's Columns/Staff Reports
1

Well, Cecil threw this out with a wave of his hand saying

I’d like to know about those 200 objective scientific studies, and a web search brought up nothing. Cite, please.

I have nothing but my own experience to dispute this. But I’m a scientific guy, not into crystals or psychics or that sort of nonsense. Granted, there are graphologists out there who claim they can do highly unlikely things like medical diagnosis for example. However, the fact that there are quacks, and the fact that some people who espouse graphology also believe in numerology or pyramids or whatever does not in itself disprove the notion that people reflect various personality traits in their handwriting.

I’ve studied graphology for some twenty years, and collected and analyzed well over a thousand handwriting samples. For me, it’s just an amusing party trick and a way to know more about people; I don’t do it professionally or claim to be any sort of expert. But in my (some might argue very experienced) opinion, the core claim that the way a person writes reflects elements of their personality is true.

Some specific examples:
[ul]
[li]People that write quickly tend to think quicker then average. and vice-versa.[/li][li]People that write hard (push the pen into the paper and leave an impression) tend to be more passionate then average. And vice-versa.[/li][li]People that lean their letters further to the right then average tend to be more extraverted then average. And people that lean more to the left tend to be more introverted. (In this example, it’s important to compare them to the way they were taught as “normal”, as different countries have different “normals”)[/li][/ul]
There are many more examples.

Is this a 100% kind of thing? Well, no more then any other personality analysis technique. Can you diagnose someone’s ailment with it? Of course not. Can you even tell if a person is a good job candidate by the way they write? Well, even of this, I’m not so sure myself. But… can you tell a lot about what basic traits a person has by their writing? I say yes. And I have a fair amount of experience in the practice as well as some credibility as a scientific thinker and skeptic of the paranormal.

Anyway, back to my question. Cites please for those “200 objective scientific studies”. Oh, and I’m assuming in Cecil’s research through those 200 studies that he came across none supporting the validity of handwriting analysis, or certainly they would’ve been mentioned. But in the off chance I’m mistaken and it’s not unanimous, please post those studies as well.

Thanks.

2

Sorry, here’s the link: http://www.straightdope.com/columns/030418.html

3

http://www.bccla.org/positions/privacy/88graphology.html

http://www.reall.org/newsletter/v05/n08/graphology-write-and-wrong.html

4

I saw the first one on a web search before I posted. Frankly, I didn’t even read beyond the first few paragraphs because:

a) it’s from a civil liberties organization
b) the title is “The use of graphology as a tool for employee hiring and evaluation”

Sure, they don’t feel handwriting analysis should be used for employee hiring and evalution. They are also against Lie detectors (http://www.bccla.org/positions/privacy/77polygraph.html), Blood tests (http://www.bccla.org/positions/privacy/82blood.html), Drug tests (http://www.bccla.org/positions/privacy/89drugtest.html - link is down, but it’s on their site), DNA testing (http://www.bccla.org/positions/privacy/94dna.html), psychological testing (http://www.bccla.org/positions/privacy/87psytest.html), and many others.

Sorry, but there is such extreme bias in that cite that it just isn’t reasonable.
In the second article, the first study (which you cite), studies

This is also cited in your first article.

Sorry, but that’s nothing that I or any reasonable person would claim that handwriting analysis can answer (success of a salesman).

That’s the only test this article gives the parameters for. It cites others, but doesn’t talk about what’s actually tested.

Also, I should add that the second article is clearly discussing whether handwriting analysis is a good tool for employee qualification. And as I stated in the OP, I don’t think it is.

Cecil was clear:

He wasn’t referring to it’s validity to prove a murderer had the propensity on the witness stand, and he wasn’t referring to it’s use as a testing element for employee selection. He said it can’t predict personality. I say that’s not true.

Let me give you a specific scientific test that I can (and do) pass on demand, and that any other studier of handwriting analysis could:

Introduce me to 10 people. Let me get to know them for a day or less, so I have a quick understanding of their personalities. Then give me a handwriting sample from each of them. Each sample can be the exact same words, copied from anywhere. I can match the handwriting to the people based on their personality.

Have I done this in a laboratory under the supervision of trained professionals? No. Will I be right 100% of the time? No. But I’ll be right far more then average. I’ll put even money that I can do better then 90% over a series of these trials. And by the way, when I do this for new friends, I’m rarely not at 100%.

5

Oh, one thing that may be interesting:

When someone gives me some handwriting to analyze, there’s one question they ask above all others: “Is this person honest?” And certainly, that’s the biggest question that an employer would want to know as well.

My opinion is that there is no way to know. I’ve read many people’s thoughts on how it can be determined, but I haven’t seen any conclusive way of knowing.

There are other important traits I can tell with reasonable (although not perfect) certainty: Is this person tenacious? Are they a coward? Are they sensitive to criticism? Are they creative? Are they detail oriented?

Anyway, as a big reader and fan of Cecil, I’m sure that when he says he’s aware of (and I would hope has at least partially read) “more then 200 [relevant] objective scientific studies,” that this figure didn’t just come from the bibliography count of an ACLU site. Surely Cecil did his homework. Cites, please.

Thanks.

6

There’s certainly extreme bias somewhere here.

7

Bill H comments << [Cecil] said it can’t predict personality. I say that’s not true. >>

That means you say that handwriting analysis can predict personality. Interesting, because I quote the learned Bill H:

The bold emphasis is mine. Note the word “tend” – “tend” does not mean “predictor.”

  • People who commit street crimes tend to be blacks rather than whites.

Does that mean that race is a predictor of criminality? Certainly not. Does that mean that if I find a street crime, I can assume a black person was the culprit? Certainly not. Those nasty words “tend to be” get ignored, and so destroy the fine line between a broad generalization and an inevitable cause-effect.

Yet you blithely go on with the same sort of nonsense about handwriting analysis. Even if it is true that “people who write quickly tend to think quickly,” does that mean that if you find someone writing fast, you can conclude they are a fast thinker? Certainly not. Does it mean that if you find someone who is a fast thinker, that you can assert that they write fast? Again, certainly not. Hence, you don’t have a predictor.

In short, what you are doing is stereotyping – applying a broad (unproved) generalization to all individuals in that class.

So: what does “tend to” mean in your comments? 6 out of 10? 51 out of 100? 500,001 out of 1,000,000? Or does it just mean that you personally have noticed…

8

On the other hand, he does provide a quite testable claim, with regards to identifying people based on handwriting. Bill H., do you think you know regular posters here well enough to pull off that test? And would scanned samples of writing be sufficient material to check from? It might be instructive (for at least one side of this debate) to run such a test online.

Assuming that Bill H. agrees to such a test, do we have any other volunteers to be identified?

9

“Testable”? You mean << There are other important traits I can tell with reasonable (although not perfect) certainty: Is this person tenacious? Are they a coward? Are they sensitive to criticism? Are they creative? Are they detail oriented? >>

Most of that is typical horoscope nonsense.

First off, every human being has times when they are tenacious and times when they are not, times when they are cowardly and times when they are not. Find me a person who is not “sensitive to criticism” or is not “creative” and I’ll show you a corpse.

The “test” that you describe – meeting ten people (same gender) for a day each, without seeing their handwriting, and then asking them all to write out the same paragraph and telling which is which from the handwriting – would be an interesting test, but impractical.

10

Well, hey, I’ll volunteer, but somebody will have to help me with the tech stuff, how to upload it or e-mail it or whatever.

11

C K Dexter Haven wrote

Now now, C K. I’ve read plenty of your posts and have a good regard for you, and if you’ve read many of mine, I expect you’ll also find I’m a reasonable scientifically-minded person, and generally polite to boot. You can be a little nicer. Accusing me of stereotyping and racism isn’t very kind (and not far off from the inevitable Hitler comparison, which I’m sure you’re above). And if your stance has scientific validity, then you can back it better then using phrases like “Yet you blithely go on with the same sort of nonsense…”

I’m afraid, my friend, that science is filled with “tend to’s,” especially in trying to quantify human personality traits. Why is it that you can find two expert witnesses arguing over whether an individual is insane? Does the fact that it’s debatable make psychiatry a non-science?

Now, to your actual question:

Excellent question. And I’ll be the first to say I don’t know. In fact, I’ll be the first to say that I haven’t seen a scientific study backing what I claim. Which is really why I want to see these 200 studies that Cecil has in that briefcase. But I’ll put it this way: in my estimation, it’s a heck of a lot better then 500,001 out of 1,000,000. Is it better then 80%? I have no scientific study to back me, but I’d guess yes. Is it better then 90%? I’d guess probably.

And by the way, even if it were a measly 6 out of 10, that still makes it a predictor by definition. Not a good predictor of course, but a predictor none the less. And those magic 200 objective studies claim prediction is impossible.

12

C K Dexter Haven wrote

I don’t follow why it’s impractical. Here’s the definition for an objective study. I assume that each of the magic 200 are similar to this, and that 200 of 200 failed, which is (again) why I’m very curious to see them.

a) Bring in X subjects (10? 100? you’re the tester, you make the call.)
b) have each of the them write a few paragraphs, and give the writings to the handwriting analyst(s) under test.
c) have the handwriting analyst(s) meet and ask questions of each of the subjects for a period of Y (an hour? a week? your call)
d) have each analyst match writing to subject.
e) analyze their success rate

Why is this “impractical”?

13

Chronos wrote

I don’t, but maybe we can still do this. I could rely on past postings by the subjects and email questions and answers with them.

A couple of concerns:

  • I won’t be able to gauge pressure of writing, unless the samples are mailed to me.
  • Standard fax is pretty thick-grained; I’ll take a look at a fax tomorrow to see if it works. We may need to make scanning the sample with reasonable pixel density a requirement.
  • I’m concerned about personality traits that will be lost through email communications. I’m thinking specifically about how spontaneous people are, how organized their thoughts come out, and other traits that are editable via postings and email.

If we do this, I ask for a couple of conditions:

a) No fakes. Nobody having their sister write a sample, or copying Hitler’s handwriting or writing different from how they normally do. You want a scientific study, and so do I. Be fair.
b) I’d like some diversity of personalities. It’s a hell of a lot easier for me to tell the difference between a physics major and a cheerleader then between two physics majors.

Let me give this a little thought. But I think we can do this.

14

Okay, so everybody mail their handwriting samples to Chronos, then he can collect and mail the set to Bill H. That way all material comes to Bill from one source.

We need to pick a standard writing sample that everyone copies. It needs to be sufficient length for Bill H. to get an accurate read. How many words, do you think? Or pick an appropriate Cecil Column.

Statement of test: some selection of handwriting samples will be collected by Chronos and then mailed to Bill H. Bill will evaluate the samples, and compare to a list of posters on this board that are the designated participants. He will communicate via the board and email to get his assessment of their personality. Then Bill will compare the handwriting samples to his impression of each of the posters and match them up. His results will then be posted to the board.

Hmmm, I guess we need some number or letter designator applied to each handwriting sample by Chronos to identify them. Then Bill can post “sample x = Irishman” or whatever. Then Chronos can post the answer key.

Guess that makes Chronos the non-participating mediator.

Hmmm, I’m trying to think of how to keep it double blind. I guess best way is to keep all discussion between Bill and posters via email. He can read any posts he wants, but no questions and responses via the board.

If we work all this out, I’ll participate.

Bill H., do you agree to Chronos as the mediator? Or prefer someone else?
Chronos, that okay with you?

15

Very interesting! If this goes forward, I’ll gladly be one of the volunteers.

16

Hey, everybody! I have a really cool ability, too. I can tell people all about what “kind of person” they are, and all I have to do is hold a dowsing rod over them while they lay out Tarot cards on a Ouija board under a pyramid. I mean, when I say “Ah! I see you tend to be passionate sometimes, and that you are often creative!” how can anybody refute it? And what a fine example of clear, measurable, objective science it makes. But it works better if there’s full moon. . . .

I sympathize with ya, Dex. It’s tougher to fight ignorance when it wants so badly to fight back.

17

In defence of Bill, I wouldn’t characterize his position as one of “ignorance”, any more than I would characterize a Christian who was arguing for the validity of Jesus’ resurrection as being “ignorant”. The Christian has something he believes works; so does Bill. The big difference between them is that graphology is something we can test.

Which hopefully we will do, and without any “nyah nyahs”, either, preferably. :wink:

Because I’m curious, too. I understand Dex’s point about the dangers of generalizing, but I also understand the graphology point that it does sometimes seem like you can tell things about people by their handwriting. So I’d for one would like to see it tested.

18

I don’t trust the test setup as it is yet. How do we do a control?

19

Irishman wrote

Well, here’s something amusing… Of everybody who’s posted to this thread thus far, Chronos’s writing is the one that I have some certainties about already. (okay, they’re not “certainties”, just a strong hunch.) So therefore, I’d prefer that he be one of the subjects.

However, there’s no reason he can’t be a subject and the tester. I’m not sure double-blind is required in this instance. The results are pretty objective; either I match samples or I don’t. That is unless you suspect Chronos and I are in cahoots, but you’ll have to make your own call about that.

I’m still nervous about a couple things I mentioned earlier:
a) I’m not sure how good a read I can get on some key personality traits via email questions and answers.
b) I’m concerned that the subject pool may be very narrow. The average reader of CCC is very specialized in some important “personality trait” areas. Expanding the pool to include Cafe and MPSIMS people who don’t typically read CCC or GQ (and I’m sure there are many) would be a good start.
c) Honestly, I’m concerned that there are people who will submit false samples.

I suspect we can work out #a and #b, and really hope we can eliminate #c.

The best way I can see to reduce #a and #b is to allow me to make choices in the subject pool, both in the posters, and the samples. Choosing among posters (based on reading their past postings and having conversations with them via email) will let me choose those that I feel I can get a good read on how well I understand their personality. And choosing among samples will let me eliminate ones that are too similar.

I sent myself a sample fax today, and though it’s not totally unworkable, I would much prefer that we use physical paper.

So… I’ll stand up to the plate and risk the potential “nyah nyah”'s. Let’s come up with rules agreeable to all and see what happens.

And of course I must mention again: this will be an interesting experiment, and actually have some level of validity. But surely there are professionals who have tested the validity of handwriting analysis much more precisely and generally. I really am interested in these 200 reports. I mean if you had 200 tests on whether tea leaves predicted the sharpness of razor blades, I’d be shocked to hear that not even one came back positively. That’s just the nature of studies, especially ones about something as subjective and complex as personality. Cecil has some pretty strong indisputable evidence against handwriting analysis, and I sure would like to read it. Where’s those cites, Unca Cece?

20

Yeah; I wonder a little bit about that, too. The nice thing about Bill H.'s proposed experiment is that the outcome is completely unambiguous: either the sample is matched, or it isn’t. On the other hand, the claim is that handwriting analysis can predict personality, and that (it seems to me) is not being unambiguously tested. For example, I could imagine handwriting being matched to the person on the basis of physical traits: male/female, left-handed/right-handed, athletic/non-athletic, tall/short, etc. Admittedly, difficult to perceive via email, but perhaps a question nonetheless. All you’d need is four or five yes/no physical factors to successfully pigeonhole ten writing samples and get the attributions correct.

Let me throw this out as an idea: How about choosing four or five distinct personality traits for Bill H. to ID. Then have, say, 20 people volunteer to be ranked-ordered by fellow Dopers (from most passionate to least passionate, for example). From these 20, randomly choose 10 to send handwriting samples to Bill H., who will do his own rank ordering, for each trait, based solely on the handwriting. Should be pretty easy to calculate what the (actual rank - Bill rank) would be for random guessing; compare that with Bill’s score.

I realize there’s some issues with this procedure too; most notably the reliance on subjective voting to get “actual rank ordering”. Still, I wanted to throw it out as an idea.