In your column about handwriting analysis you wrote (http://www.straightdope.com/columns/030418.html), "More than 200 objective scientific studies have demonstrated that graphology is worthless as a predictor of personality." After I whined a bit, you conceded that you had misstated matters. It wasn't that 200 studies had independently concluded graphology was worthless; rather, one researcher, Geoffrey Dean, made this judgment based on a "meta-analysis" of 200 previous studies. Even Dean's study is seriously flawed and doesn't support the conclusions drawn.
Let me define what I believe is in contention. I specifically do not claim that handwriting analysis is useful for a purpose such as personnel selection; that people who practice it are honest, scientific, or consistent in their analyses; or that all or most of the personality traits they claim to find reflected in handwriting actually are. I propose simply that elements of personality are reflected in handwriting. That's all. And that's exactly what the column claims is not true.
Illustration by Slug Signorino
You and the gang went around on this for quite a while on the message board, Bill. Unsurprisingly, no firm conclusions were reached. (Such is the nature of message boards.) But you seem like a good fellow — I’d like to provide some closure, so that you, and everyone else who piled on once the debate got going, can get on with your lives. So I’ll happily concede one point, which doesn’t have much to do with the point I was making back in April: Even if you’re totally unschooled in graphology, you can use handwriting to predict, with greater accuracy than a coin toss, certain inherent personal characteristics (not personality traits). Splitting hairs, am I? Not at all. I’ve just learned from long experience that you can appear to prove all manner of astonishing conjectures if you set the bar sufficiently low.
Let’s consider your objections to Geoffrey Dean’s argument, which appears in The Write Stuff: Evaluations of Graphology — The Study of Handwriting Analysis (Beyerstein and Beyerstein, editors, 1992). At the outset Dean makes what you feel is an important concession: Studies have shown that untrained people given handwriting samples can guess with modest accuracy the sex and approximate intelligence of the writer — on the first count, 60 to 70 percent accuracy, vs the roughly 50 percent you’d expect from blind chance. But then, you say, “he removes these factors from his study.” (I’m quoting from your message board postings.) You find this outrageous. Sex and intelligence obviously have a major impact on personality. If graphology’s predictions about these traits — the very traits it’s best at spotting — are declared inadmissible, that eliminates the strongest arguments in support of the practice, yes?
No. You’ve misunderstood what Dean is studying. The issue is not whether you, regular guy, can deduce something about Citizen X from his handwriting. No one doubts that you can — if nothing else, that Citizen X knows how to write. But is it fair to call that a personality trait, like honesty or dependability? I venture to say you’d also have little trouble distinguishing the writing of a five-year-old from that of an adult, or the writing of a sober person from that of a drunk. In adding sex and intelligence to this list, Dean is simply affirming that certain personal characteristics (as opposed to personality traits) are evident in handwriting.
What does this prove? In itself, nothing. What amateurs can do is of secondary importance — amateurs are the baseline, the control group. The question is whether a professional — someone versed in the supposed science of graphology — can outperform a layperson. Dean, based on his meta-analysis (which I did mischaracterize, having relied on a mistaken summary in a scientific periodical), says it ain’t so.
To clarify: in a meta-analysis one aggregates the results of many small studies in order to eliminate errors resulting from differing methods or inadequate sample size. Dean notes that studies of graphology often arrive at contradictory conclusions — one finds that aggressiveness correlates with heavy writing pressure, another with light; one links emotionality to large writing, another to small. Add them all up (the process is more rigorous than my casual description suggests) and the spurious results tend to cancel out. Dean reports that, as an indicator of personality and IQ, graphology does better than astrology, palmistry, and phrenology (the study of bumps on the skull) but considerably worse than conventional IQ tests and personality inventories. Conclusion: Graphology is worthless as a predictor of personality. Other common techniques are much more reliable and accurate. Granted, a look at handwriting tells you something — but do you really need it to determine the sex of someone named Mary Smith?
Send questions to Cecil via firstname.lastname@example.org.