On the failures of graphology

Full text

On the failures of graphology
Michael Moore
Technion – Israel Institute of Technology
Abstract: Research evidence concerning the lack of validity of graphology for
personnel selection or placement is followed by several possible reasons for the
continued popular belief in the usefulness of graphology. The latter include content,
mistaking reliability for validity, intuitive appeal, limited predictiveness, and illusory
correlations .
Key words: graphology, validity, personnel selection, illusory correlations
Advertisements like the following ones often appear worldwide:
--Handwritten application should be sent to: THE ADVERTISER P O BOX 4719,
MARINA, GPO, LAGOS
--Qualified and experienced candidates can send handwritten application along with a
passport size photograph to the Principal, BGS National Public
School, Hulimavu, B'G-Road, Bangalore 560076.
-- Send handwritten application to: Taylor Industries / Attn: Maria Cornelius 6015 N.
Xanthus Ave. / Tulsa, OK 74130-1508 Or fax to: 918- 266- 4194
--Please send handwritten application with CV to: Denise Lavey, St Albans
Medical Centre, 26-28 St Albans Crescent, Bournemouth BH8 9EW.
--les candidatures manuscrites accompagnées d'un curriculum vitae détaillé sont à
adresser à: la guerche sur l'aubois18150 la guerche sur l'aubois
--Vergiss auch nicht den handgeschriebenen Lebenslauf. Besten Dank. Schicke die
Bewerbung an: Gemeindeverwaltung Hittnau Lehrlingswesen Postfach 8335 Hittnau

These employers will submit the handwritten applications they receive to a
graphologist in order to determine the candidates' suitability for the advertised post.
Here are some claims made by professional handwriting analysts:
--Graphology helps to "decide which person to use as your accountant, who you
should hire as your baby-sitter, who you should go into business with, who you
should date, who you should trust, and innumerable other applications."
--Graphology can be practical when redundancy is inevitable. It can detect new
directions in which the employee could channel his energies.
--Graphology can save time, money and effort in this area by seeing that tghe
essential characteristics required for the job are present.
--Graphology recognizes a sign that is "associated with bitterness, bad instincts and
guilt. The higher the claw, the worse the situation is, and the more conscious the
writer is of her guilt and criminality."
--Ludwig Klages (1872-1956), the father of German graphology, even claimed that
graphology was "effective in detection of 'non-Aryans."'
While employers may be tempted to use a relatively inexpensive selection (and
placement) method, those familiar with the difficulties of personnel selection, as well
as knowledgeable about the rigors of validation are likely to raise an eyebrow: Are
such claims based on acceptable research?
Among the multitude of replies to this question a much quoted one is a study by
Rafaeli and Klimoski (1983). The latter examined the relationship between
assessments of the handwriting of 104 real estate agents conducted by expert
graphologists and measures of the performance of those agents. Their research was
funded by the American Association of Handwriting Analysts. When these researchers
concluded that the graphologists' assessment bore no relationship to actual
performance, the funding organization threatened to sue them in order to prevent
publication of their findings. Neter & Ben-Shakhar's (1989) results went even further:

Their examination of 17 studies involving 63 graphologists, 51 nonprofessional
analysts, and more than 1,200 handwriting samples concluded that graphologists were
no better than non-graphologists at predicting job performance .
These studies dealt with personnel selection. A whole slate of other studies
investigated the ability of graphologists to diagnose personality traits. Their results
were similarly negative: A meta-analysis drawn from over 200 studies concluded that
graphologists were generally unable to predict any kind of personality trait on any
personality test (Jennings, Amabile, & Ross, 1982.(
What can then be the reasons for the continued use of graphology, in spite of hundreds
of findings that dispute its ability to provide valid advice to employers?
--Content. A study predating the above (Jansen, 1973) found that psychologists
analyzing typewritten transcripts of the handwriting samples seen by graphologists
made predictions of equal validity. This suggests either that graphologists can be
pretty good psychologists or that any fairly intelligent person can draw some
meaningful conclusions from a CV…
--Mistaking reliability for validity. Graphologists tend to agree with each other not
only about the coding of graphological signs (such as slant, size, rhythm etc;
correlation coefficients of 0.6 to 0.85 between two independent readers of the same
document), but also about the interpretation (0.42). But as it turns out, even non-
graphologists tend to agree with each other in invalid naïve interpretations (0.30), for
instance about depression, methodicalness, or originality (Dean,1992). King &
Koehler (2000) have referred to this as "shared but invalid beliefs about the
relationship between handwriting and personality variables ."
--Intuitive appeal. Handwriting appears to be such a good source of information: It
differs from person to person; it is rich in detail (400 features by one system of
analysis); there is a belief that everything we do expresses something about us
(Allport & Vernon, 1933). So size is thought to indicate degree of egoism, forward
slant might be related to outgoingness, and so on. Some of these associations have a
semantic character: Regularity of rhythm is thus purported to indicate reliability of
behavior .

--Limited predictiveness. Gender, SES, and degree of literacy have low but
significant correlations with handwriting. To the extent that any of these is correlated
with personality traits or with job performance measures, there will be some low
correlations between handwriting signs and some criterion variables. This is of no
practical use, since there are far stronger (i.e. valid) indicators of each of these status
variables. The same holds for the tremor found in the handwriting of alcoholics and
the relationship between poor handwriting and poor school performance. If
graphologists had no greater pretensions than these, there would be fewer objections
to their activity. But there is a huge conceptual leap between Miss Taylor failing
Johnny (whose handwriting she cannot decipher) and an employer's decision not to
hire someone because her writing slants downward.
-- Illusory correlations. Nearly half a century ago Chapman & Chapman (1969)
introduced this concept to designate the seeing of an expected relationship between
variables even when no such relationship exists. This is a far reaching concept,
touching on such phenomena as superstition, prejudice, stereotype, as well as the
topic at hand, and related to confirmation bias, selective attention and similar
phenomena. Graphology itself is but one of a long list of supposedly diagnostic
devices which fail when submitted to rigorous psychometric analysis. (The Chapmans
used the Draw-A-Person projective test, as well as the Wheeler signs of
homosexuality in the Rorschach to illustrate their point(.
A typical example of an illusory correlation (Kammann & Cambpell, 1982) is the
following: Objective data clearly show that there is no useful correlation (meaning
that the median r is 0.10, range 0.02 to 0.18) between happiness and several life-
circumstance factors, such as income, type of work, gender, religion, race, age, city
size, level of education. And yet when respondents estimated the percent of people
who were happy in various categories of such variables, they thought a far higher
percent were happy among those earning a lot than those with a low income, those
living in rural areas than those in large cities, etc. They were also asked about the
importance of these variables to happiness. Here, too, the illusion continued: for
instance, high education was thought as important for happiness by 86%.
King & Koehler (2000) applied the concept of illusory correlation to graphology.
They had participants judge the relatedness between certain handwriting features

(such as size) and personality dimensions (such as modesty/egotism). Some other
pairs, drawn from graphological manuals, were speed of writing and impulsiveness,
slope and pessimism or spacing and extraversion. They were given random pairings
of handwriting samples and brief personality profiles; that is to say, there was zero
correlation between handwriting and personality. Yet the participants found a
correlation of .65 !
In the following I shall describe my own informal investigation concerning the
validity of graphology. A few years ago I contacted seven graphological organizations
around the world, and requested information about "controlled studies in which the
results of handwriting analysis are shown to predict employee characteristics." I
received a reply from four. These responses contain important information about the
mindset of professional graphologists .
The president of the American Association of Handwriting Analysts (he who
threatened to sue Rafaeli) referred me to a published article, which, as it turns out,
says that the study "of the validity of /handwriting/ variables has yielded up to now no
convincing success" (Lockowandt, 1976, p. 5). He also suggested that I inquire at the
Israeli Graphological Institute. These, in their turn, referred me to Allport & Vernon
(1933) who had indeed been sympathetic to the cause in their famous book on
Expressive Movement. It is unfortunate that they had not read the book, for its authors
concluded that "...the [graphological] terms employed often seem to obscure rather
than reveal the personality." Allport & Vernon also reported the results of an
experiment, in which 10 handwriting samples were to be matched with 10 personality
descriptions. Graphologists averaged 2.4 hits, non-graphologists' average success rate
was 1.8. In other words, years of training and practice enabled the professionals to
miss 76%, while those who had no training missed 82% of the cases presented to
them. I was also visited by the institute's secretary, himself a practicing graphologist.
He told me that when asked by a kibbutz how to assign a new member to work, he
could advise them, on the basis of a handwriting sample, whether that person should
work in wood or in metal. He also told me that there was a good article on graphology
in Playboy.
The 3rd response came from the president of the Handwriting Analysts International.
He was extremely pessimistic about the existence of adequate research in this field,

and added that "it would be difficult if not impossible to assign any specific writing
characteristics... to a given syndrome". Yet he referred me to a European source
(whom I could not locate) who "has done a linear study of 70 plus years on a single
subject."
Lastly, Prof. Marchesan of the International Society of Handwriting Psychology sent
me a list of 307 scholarly works he had co-published in Italy, together with an English
language flier about the society. The latter contained no information about the
reliability, validity or utility of graphology. It did, however, claim that the theory
underlying graphopsychology can be used "for the reconstruction of the
characteristics of historical personages for whom there is no handwriting sample
available". The examples of such personages given were Dante, Jesus, and Mary. To
make absolutely sure that my query had not been misunderstood, I wrote Prof.
Marchesan again, and asked him for empirical data. He responded by an updated
bibliography, and by saying that his group "has applied, with considerable success,
Handwriting Psychology for self knowledge, for work aptitude related to scholastic
preparation, for choosing one's partner in life, for the selection and hiring of personnel
or, for change of position or promotion of personnel already on the staff."
The main conclusion I can draw from the evidence presented in this article
coincides with a statement made by Pirsig: "The real purpose of the scientific method
is to make sure Nature hasn’t misled you into thinking you know something you
actually don’t know" (1974, p. 94.(
References
Allport, G. W. & Vernon, P. E. (1933). Studies in expressive movement. New York:
Macmillan.
Chapman, L. J., & Chapman, J. P. (1969). Illusory correlation as an obstacle to the
use of valid psychodiagnostic signs. Journal of Abnormal Psychology, 74, 271-280.
Dean, G. A. (1992). The bottom line and effect size. In B. L. Beyerstein& D. F.
Beyerstein (Eds.), The write stuff: Evaluations of graphology, the study of
handwriting analysis (pp. 269-341). Buffalo, NY: Prometheus Books.

Jansen, A. (1973). Validation of graphological judgments: An experimental study. The
Hague: Mouten.
Jennings, D., Amabile, T. M., & Ross, L. (1982). Informal covariation assessment:
Data-based vs. theory-based judgements. In A. Tversky, D. Kahneman, & P. Slovic
(Eds.), Judgement under uncertainty: Heuristics and biases. Cambridge: Cambridge
Univ. Press.
Kammann, R., & Campbell, K. (1982). Illusory correlation in popular beliefs about
the causes of happiness. New Zealand Psychologist, 11, 52–62.
King, R. N. & Koehler, D. J. (2000). Illusory correlations in graphological inference.
Journal of Experimental Psychology: Applied, 6, 336-348.
Lockowandt, O. (1976). Present status of the investigation of handwriting psychology
as a diagnostic method. JSAS Catalog of Selected Documents in Psychology, 6, 4-5.
Neter, E., & Ben-Shakhar, G. (1989). The predictive validity of graphological
inferences: A meta-analytic approach. Personality and Individual Differences, 10,
737-745.
Pirsig, R. (1974). Zen and the art of motorcycle maintenance. New York: William
Morrow.
Rafaeli A, Klimoski R. (1983). Predicting sales success through handwriting
analysis: An evaluation of the effects of training and handwriting sample
content. Journal of Applied Psychology, 68, 212-217 .