http://www.straightdope.com/columns/read/2447/is-handwriting-analysis-legit-scienceIs handwriting analysis legit science?
April 18, 2003
Dear Cecil:
What's the Straight Dope on handwriting analysis? I know that handwriting experts' testimony can be accepted in court, so there must be something to it. But I have a hard time believing that a smart criminal wouldn't be able to change his writing to avoid detection. On a related issue, can an "expert" really tell something about your personality from your handwriting (e.g., that loops in your g's and y's indicate a high sex drive)? If that were true, it would seem that one's handwriting would change from day to day, which it doesn't.
— Kristin in Sausalito, CaliforniaAt first this question might seem like a great opportunity to lay out the difference between science and pseudoscience. On the one hand we have forensic handwriting analysis, in which an expert decides whether two or more samples were written by the same person, e.g., whether a signature was forged. On the other we have graphology, in which some sage tries to divine a subject's personality traits from his or her handwriting. While graphology enjoys about the same prestige as palm reading, forensic handwriting analysis has helped send people to jail since the days of the Lindbergh kidnapping. But in the eyes of the law, the credibility of such analysis is on the wane. Thanks to a landmark Supreme Court ruling in the early 90s, more and more federal judges are deciding that while forensic handwriting analysis may not be quackery, it's not exactly science either.
A meta-analysis of 200 scientific studies of graphology by Geoffery A. Dean (published in The Write Stuff: Evaluations of Graphology--The Study of Handwriting Analysis, edited by Barry L. Beyerstein and Dale F. Beyerstein, Prometheus Books, 1992) found that it was worthless as a predictor of personality. That hasn't prevented people who ought to know better from relying on it. In France, an estimated 70 percent of companies use graphology when making hiring decisions. (Between 5 and 10 percent of U.S. and UK companies do so.) Law enforcement authorities sometimes turn to graphology and kindred techniques when profiling criminals, as in the case of the D.C. sniper last fall. But such methods are often the last resort of police desperate to appear to be doing something. There's only one well-documented case of a bad guy actually being caught by a profile--George Metesky, the "Mad Bomber" of New York City in the 1940s and '50s--and he was nabbed less because of his handwriting than because he'd revealed too many clues about his past in a letter to a newspaper.
For a long time forensic handwriting analysis seemed more respectable, but its status has been shaky since 1993, when the Supreme Court handed down its ruling in Daubert v. Merrell Dow Pharmaceuticals. Previously the chief criterion for the admissibility of expert testimony had been whether it was based on techniques "generally accepted" by scientists. Daubert gave federal judges much greater discretion in deciding admissibility. It suggested they consider (1) whether a theory or technique can be tested, (2) whether it's been subject to peer review, (3) whether standards exist for applying the technique, and (4) the technique's error rate.
Sounds reasonable, eh? But Daubert created an uproar, because the dirty little secret of much so-called expert testimony was this: though it was possible in principle to test and validate most forensic techniques, in many cases no one had ever done so. In 2002 one judge even restricted testimony based on fingerprint analysis, saying he was unconvinced the technique was a science rather than a mix of craft and guesswork.
No forensic technique has taken more hits than handwriting analysis. In one particularly devastating federal ruling, United States v. Saelee (2001), the court noted that forensic handwriting analysis techniques had seldom been tested, and that what testing had been done "raises serious questions about the reliability of methods currently in use." The experts were frequently wrong--in one test "the true positive accuracy rate of laypersons was the same as that of handwriting examiners; both groups were correct 52 percent of the time." The most basic principles of handwriting analysis--for example, that everyone's handwriting is unique--had never been demonstrated. "The technique of comparing known writings with questioned documents appears to be entirely subjective and entirely lacking in controlling standards," the court wrote. Testimony by the government's handwriting expert was ruled inadmissible.
Prosecutors scrambling to find scientific validation for handwriting analysis last year touted a study by Sargur Srihari, a professor of computer science at the State University of New York at Buffalo. Srihari subjected 1,500 writing samples to computer analysis. Conclusion: In 96 percent of cases, the writer of a sample could be positively identified based on quantitative features of his handwriting such as letter dimensions and pen pressure. Skeptics objected that lab results using a computer prove nothing about what a human can do in the real world, and who can argue? If expert testimony is going to send people up the river, it better be more than some mope's prejudices dressed up as science.
— Cecil Adams
