Terence Hines - Pseudoscience and the Paranormal

According to the claims of graphology’s proponents, it is possible to determine various characteristics of an individual, especially those relating to personality, from his or her handwriting. In the early 1980s about three thousand U.S. firms use graphology in employee selection (Rafaeli and Klimoski 1983). Eighty-five percent of firms in Europe were said to use graphological analyses in making their hiring decisions (Levy 1979). Van Deventer (1983, p. 74), managing editor of the periodical United States Banker, states that “graphoanalysis reveals capabilities and aptitudes in an individual, many of which the applicant may not even be aware of.” Given the popularity of graphology and the importance of the decisions that depend, at least in part, on its use, it is important to discover if one’s handwriting really does reveal anything about one’s personality characteristics.

There is a growing body of empirical research literature on graphology. It is almost uniformly negative as regards graphologists’ claims. Most tested among these claims is that job success can be determined or predicted from handwriting. Ben-Shakhar et al. (1986), Keinan (1986b), Rafaeli and Klimoski (1983), and Zdep and Weaver (1967) have all found that graphological analysis did not reveal anything about job success. Drory (1986) did find significant positive correlations between job ratings and graphological analysis, but a serious problem with this study is that the handwriting samples used were autobiographical sketches. Thus, the graphologists may have based their judgments on the content of the writing rather than on the handwriting itself. That this is the explanation for Drory’s findings is strongly suggested by the fact that studies using such information-filled autobiographical sketches find that when graphologists perform better than chance and a control group of nongraphologists also makes judgments based on the written sketch, the nongraphologists do as well as or better than the graphologists (Ben-Shakhar, Bar-Hillel, and Flug 1986). Ben-Shakhar, Bar-Hillel, and Flug showed that nongraphological information (autobiographical information, presence or absence of spelling and grammatical errors, and so forth) in the sketches predicted job ratings as well “as the professional efforts of experienced graphologists” (p. 187).

Studies of graphology have been conducted on variables other than job success. Keinan (1986a) found that graphologists could not distinguish at a level above chance between writing samples provided by soldiers in a highly stressful situation (half an hour before their first night parachute jump) and a nonstressful, relaxed situation. Jansen (1973) found no relation between graphologists’ judgments and various personality ratings. Frederick (1965) found that graphologists could not distinguish between mental hospital patients and undergraduate college students.

Some studies have reported positive results in which graphological analysis does allow above-chance discrimination between groups. The differences in the handwriting between the groups studied, however, reflect nonpersonality variables and do not support the graphological claim that personality is reflected in handwriting and can be assessed from it. Sex can be determined from handwriting with about a 70 percent accuracy (Goldberg 1986). Professional graphologists and nongraphologists are equally accurate at making this judgment (Goldberg 1986). Goldberg also found that nongraphologists could distinguish the writing of Americans and Europeans at a level higher than chance. This is presumably due to the different writing styles used and taught in the United States and Europe. Wing and Baddeley (1978) found that drinking alcohol changed some characteristics of handwriting so it should be possible to distinguish the writing of sober from that of intoxicated individuals.

Ratzon (1986) found that handwriting could be used to distinguish Holocaust survivors from psychiatric patients and from a group of “Nazipersecuted” individuals who escaped the Holocaust by leaving Germany. A subgroup of the Holocaust survivors with organic brain damage could also be distinguished from the other groups by their handwriting. This last finding is not at all surprising, as brain damage can easily be expected to affect, for the worse, almost any type of motor behavior. The finding also gives an important clue to the explanation of the more unexpected finding that the non-brain-damaged survivors’ handwriting differed from those of the Nazi-persecuted group and the psychiatric group. The individuals who suffered through the Holocaust, even if not brain-damaged, were very likely in poorer health, due to their horrendous experience, than were individuals in the other two groups. Poorer health would be expected to result in changes in handwriting.

Nevo (1986) reanalyzed data from a paper by Hönel (1977) in which it was claimed that criminals and noncriminals could be distinguished by their handwriting. Nevo’s reanalysis shows that statistical problems contributed to the large effects that Hönel reported. The positive effects were greatly reduced when appropriate statistical analyses were performed, but there was still a small ability on the part of a group of graphologists, considered as a whole, to classify the criminal versus the noncriminal. One variable that contributed much to Hönel’s finding was socioeconomic class. Criminals tend to come from lower socioeconomic classes than noncriminals, and socioeconomic class does seem to be reflected in handwriting, perhaps as a function of better education and more emphasis on good handwriting in the upper as opposed to the lower ranges of the socioeconomic class structure.

One study exists that seems to demonstrate graphoanalytic ability for which no obvious alternative explanation comes to mind. Frederick (1968) found that graphologists could discriminate between suicide notes written by actual suicides and the same notes copied by normal writers, who copied from typed versions of the notes. Police detectives and secretaries could not make this discrimination. This is an intriguing study and should be replicated. To my knowledge it never has been.

With the exception of the Frederick (1968) study, the results of studies of graphological claims that handwriting reflects personality variables are entirely negative. Why, then, is graphological analysis so widely used and accepted? The fallacy of personal validation, along with the selective nature of memory, accounts for the wide popularity of graphology. Reid (1983) provides an excellent example of the fallacy of personal validation as it applies to graphology: “As a reliable first test, I would suggest that you commission a graphological analysis of your own hand-writing and show the report to someone who knows you well, a marriage or business partner. In my case, both I and my wife were satisfied that the assessment was remarkably accurate. It also added to self-knowledge and has since contributed to greater effectiveness in doing my job” (p. 71). This is just the sort of glowing testimonial one receives from a good cold reading.

Reid (1983), who at the time was managing director of an executive search company, also provided excellent examples of selective memory as he recounts instances where graphological analysis had apparently been successful. One case is that of a man recently released from prison after serving a term for embezzlement. The graphological analysis showed that “his declaration that he was reformed needed to be taken with some caution” (p. 71). Thus, he obtained a job where “there was no possibility of his being exposed to temptation” and he has “not trangressed since.” This is seen as evidence for graphology. In the situation, the graphological analysis could not be falsified since the man had no opportunity to embezzle again. In another case, an executive was hired by a firm and, although a physical exam showed him to be fit, he became seriously ill three months after starting his job and ended up not returning. A graphological analysis carried out after his illness was known on a sample of handwriting taken before he was hired “suggested he had an incipient, serious medical problem” (p. 71).