Graphology, Yet Again

[zut]

Just as a comment on the two recent columns on graphology (Is handwriting analysis legit science? [18-Apr-2003] and Handwriting Analyis Revisited: Are elements of personality revealed through handwriting? [24-Oct-2003]):

I picked up a copy of the Globe and Mail this morning (I’m vacationing in Toronto, actually), and it had an article on graphology in the business section. The new and interesting aspect of the article is that it was based on a recent meta-study of earlier research, which (I’m fairly certain) was published this year in the International Journal of Testing under the title “A Position Statement by the International Graphonomics Society on the Use of Graphology in Personnel Selection Testing” by Marvin L. Simner; Richard D. Goffin (the abstract can be accessed online here).

The core of the abstract is interesting; even though the report “contains not only an evaluation of the current published scientific reviews on the use of graphology in personnel selection, but also an evaluation of several additional studies graphologists provided that seemed to have been overlooked,” the conclusions are that

(a) the continued use of graphology in personnel selection could prove harmful to many individuals and firms, and (b) it fails to approach the level of criterion validity of other widely available and less expensive screening devices used for personnel selection.

I haven’t seen the entire article, of course. However, the examples given in the Globe and Mail article are intriguing. On the one hand, there’s this:

In five other investigations, psychologists with no background in graphology were asked to make judgments about people based on the autobiographical content of handwriting samples. The results were virtually identical to those obtained through graphology. Prof. Simner and his colleague, Richard Goffin, concluded that analysis of autobiographical material is likely to seriously prejudice graphologists’ assessments.

But on the other, there’s this:

Another approach involved 80 handwriting examples, 40 supplied from the chief executive officers of successful, top-ranked private companies and 40 selected at random from a database of households with telephones. The examples were sorted into pairs, with one from each group, and three graphologists were asked to determine which of the two belonged to the entrepreneur group. On average, they were correct 83 per cent of the time, 33 per cent more than the expected chance score of 50 per cent.

THis last is presented without speculation as to how the surprisingly accurate score was accomplished.