You are on page 1of 3

Person. indiuid. D@ Vol. 8, No. 3, pp. 433435, 1987 0191-8869/87 $3.00 + 0.

00
Printed in Great Britain Pergamon Journals Ltd

Graphology and personality: Another failure


to validate graphological analysis

ADRIAN FURNHAM and BARRIE GUNTER


Deparrmenr of Psychology, University College London, London WCIH OAP, England

(Received 24 February 1986)

Sunnnary-A group of 64 adults completed the EPQ and copied out a set text in their own handwriting.
Independent coders (reliability 0.89) rated each sample of handwriting on 13 specific features which were
correlated with the EPQ. Results showed few significant differences, once again questioning the validity
of handwriting analysis.

INTRODUCTION
Despite a fairly considerable amount of research dating back over 40 years (Eysenck, 1945) there is substantial and
justifiable doubt as to the validity of graphology. Although some studies have obtained significant findings, more than may
be expected by chance (Linton, Epstein and Hartford, 1962), many have not (Lemke and Kirchner, 1971). Eysenck and
Gudjonsson (1986) have argued that this seems to indicate that “there are some slight relationships between certain aspects
of handwriting and certain personality traits, but that much depends on the skill of the graphologist and perhaps the method
he or she uses on the type of script furnished and on the atomist or holistic approach by the graphological analysis” (p.
263). Eysenck and Gudjonsson (1986) distinguish between four types of graphological analysis: holisfic matching
(impressionist interpretation of the writing matched with an impressionistic account of personality); holistic correlation
(impressionistic interpretation of the writing correlated with a quantitative assessment of personality); analytic matching
(the measurement of the constituents of the handwriting matched with an impressionist account of personality); and analytic
correlation (the measurement of the constituents of the handwriting correlated with an quantitative assessment of
personality).
Clearly of the four methods used the analytic correlational method is most empirically based. Studies that have used
this method have revealed some statistically significant and predicted results. Thus, Stabholz (1981) found extraverts had
higher middle-zone letters but did not find neurotics had a greater irregularity of writing. Similarly Williams, Berg-Cross
and Berg-Cross (1977) found extraversion linked to right-word slant, middle zone width of letters and general size on 46
female subjects.
This study was a replication and extension of the study of Stabholz (1981). The personality measure chosen was the EPQ
(Eysenck and Eysenck, 1975) because of its reliability and validity as well as the fact that so many graphological studies
have tried to link handwriting to extraversion and neuroticism (Eysenck, 1948; Eysenck and Gudjonsson, 1986; Lester,
McLaughlin and Nosal, 1977; Rosenthal and Lines, 1972; Williams et al., 1977). Furthermore, there appear to be theoretical
reasons as to why EPQ derived scores are related to handwriting. This study extended Stabholz’s (1981) study in that 13
characteristics of handwriting were considered, not just the five. They included slant (Rosenthal and Lines, 1978), t crossing
and i dotting (Lemke and Kirchner, 1971) pressure on the page (Lester et al., 1977), size (Williams er al., 1977) etc. Given
the previous positive results in this field it was predicted that extraversion would be positively associated with size, slant
and width.

METHODS
Subjects
Sixty-four subjects took part in this experiment. Approximately half were male and half female and ranged in age from
18 to 70 (the mode being in their early thirties). They were part of a heterogeneous psychology department subject panel
who volunteered to take part in another experiment.

Measure
Personality measure. Subjects were required to complete the Eysenck Personality Questionnaire (Eysenck and Eysenck,
1975). This is a 90-item yes-no personality questionnaire which provides 4 scores per subject: neuroticism, extraversion,
psychoticism and lie. It is one of the most extensively used and thoroughly assessed personality inventory in social and
clinical psychology.
Graphology measure. Subjects were each given an identical piece of unlined A4 white paper and a photocopy of half a
page of text (on tea, p. 26 in Mikes, 1978) which they were requested to copy in their own handwriting in their own time.
These scripts were later given to two independent raters who scored each one on 13 dimensions derived from popular
texts on graphology. They were: size of the writing; percentage of the page used; slant of the letters; width of the words;
connectedness of letters with words; pressure on the page; spacing of words; regularity of crossed t’s; regularity of dotted
i’s; where the t’s are crossed; where the i’s are dotted; whether the subject loops letters below the line and finally whether
the subject loops letters above the line.
The correlation between the two raters was 0.89. A third rater then examined all the scripts and resolved the dispute
between the two raters. This enabled a single set of scores (13 dependent measures) to be used for each subject.

Procedure
Subjects completed both measures in their own time in a laboratory and in the presence of an experimenter. They took
between 15 and 30min to complete the two tasks. Subjects were debriefed after the study.

433
434 NOTES AND SHORTER COMMUNICATIONS

Table I. Correlates between the four personality scores derived from the EPQ and the 13 handwriting measures (N = 64)

Graphological determinants E N P L

I. Size of writing (I large-3 small) -0.03 -0.191 -0.09 0.10


2. Percentage of page used (l-75%; 3-50%) 0.10 0.00 -0.03 -0.11
3. Slant of writing (I left, 2 upright, 3 right, 4 varying) 0.10 -0.20’ 0.00 0.00
4. Width of words (I narrow~3 broad) 0.15 0.14 0.12 -0.19’
5. Connectedness of letters (I disconnected~3 connected) -0.13 0.08 -0.01 0.06
6. Pressure on the page (I light-3 heavy) 0.00 0.04 0.05 0.05
I. Spacing of words (I wide-3 narrow) -0.07 0.06 0.08 -0.1 I
8. Crossed t’s (I alwaysv3 rarely) -0.11 -0.04 0.10 0.14
9. Dotted i’s (1 always-3 rarely) -0.15 0.24’ 0.20’ -0.01
IO. Where t’s are crossed (I left-3 right) 0.10 0.06 - 0.03 0.11
II. Where i’s are dotted (I low-3 top) PO.16 -0.01 -0.09 0.00
12. Loops below (1 always~3 rarely) 0.13 -0.15 -0.05 -0.12
13. Loops above (I always~3 rarely) 0.1 I 0.13 0.01 -0.02

lP < 0.05.
M S.D.
Extraversion 12.93 (4.90)
Neuroticism 12.17 (5.76)
Psychoticism 6.17 (3.35)
Lie 5.70 (3.41)

RESULTS
Various analyses were performed on the data including l-, 2- and 3-way Anova’s (with the EPQ as independent variable
and each of the 13 hand writing factors the dependent variables); multiple regression discriminant analysis etc. but the results
quite clearly demonstrated fewer significant effects than may be expected by chance. The results are perhaps most clearly
seen in the simple correlational analysis in Table 1.
Table I shows the correlation between the four personality measures and the thirteen graphological measures. Less than
6% of the correlations are significant and most of them marginal. Extraversion correlated with none of the 13 variables,
neuroticism with three (small size, left slant. always dotting i’s) and psychoticism and lie with only one.

DISCUSSION
These results do not confirm those minority studies which indicate some significant findings. Indeed the conclusion of
this research concurs with many other studies in this area:
1. “It was concluded that the analysts could not accurately predict personality from handwriting.” (Vestewig, Santee and
Moss, 1976, p. 592).
2. “No evidence was found for the validity of the graphological signs.” (Lester et a/., 1977, p. 137).
3. “Thus the results did not support the claim that the three handwriting measures were valid indices of extraversion.”
(Rosenthal and Lines, 1978, p. 45).
4. “There is thus little support here for the validity of graphological analysis.” (Eysenck and Gudjonsson, 1986, p. 264).

A number of conclusions may be drawn from this and other studies, It could be argued, for instance, that the above
method missed out crucially important handwriting variables or that handwriting does not relate to the four EPQ factors,
but to other aspects of personality. More radically, one may argue that it is the EPQ that is at fault, not being a ‘good’
measure of personality. These arguments are easily refuted: variables used in this study are highlighted by many
graphologists as being the most useful to look for; the four EPQ factors relate in systematic and predictable ways to nearly
all other important measures of personality (Eysenck and Eysenck, 1985) and the EPQ is one of the most robust and
validated of all self report personality measures (Furnham, 1986).
Even if graphological analyses were valid, the theoretical basis of the method appears weak, non-explicit and
non-parsimonious. Furthermore, it is unclear why it should be used if clearly valid and reliable measures exist to measure
the same thing (i.e. personality) more cheaply, accurately and efficiently. Perhaps one should be forced to conclude rather
uncharacteristically for researchers that ‘no further work needs to be done in the field’?

REFERENCES
Eysenck H. J. (1945) Graphological analysis and psychiatry: an experimental study. Br. J. Psychol. 35, 7G-81.
Eysenck H. J. (1948) Neuroticism and handwriting. J. abnorm. sot. Psychol., 43, 94-96.
Eysenck H. J. and Eysenck M. W. (1985) Personalily and Individual Differences: a Natural Science Approach. Plenum Press,
New York.
Eysenck H. J. and Eysenck S. B. G. (1975) Manual q/the Eysenck Personality Questionnaire. Hodder & Stoughton/EdITS,
London/San Diego, Calif.
Eysenck H. J. and Gudjonsson G. (1986) An empirical study of the validity of handwriting analyses. Person. individ. 018
7, 263-264.
Furnham A. (1986) Response bias, social desirability and dissimulation. Person. individ. 013 7, 385400.
Lemke E. A. and Kirchner J. H. (1971) A multivariate study of handwriting, intelligence, and personality correlates. J.
Person. Assess. 35. 584592.
NOTES AND SHORTER COMMUNICATIONS 435

Lester D., McLaughlin S. and Nosal G. (1977) Graphological signs for extraversion. Percept. Mot. Skills 44, 137-138.
Linton H., Epstein L. and Hartford H. (1962) Personality and perceptual correlates of primary beginning strokes in
handwriting. Percept. Mot. Skills 15, 159-170.
Mikes G. (1978) How to be an Alien. Penguin, Harmondsworth.
Rosenthal D. and Lines R. (1978) Handwriting as a correlate of extraversion. J. Person. Assess. 42, 4548.
Sonnemann U. and Kerman J. P. (1962) Handwriting analysis-a valid selection tool? Personnel 39, 8-14.
Stabholz M. S. (1981) Individual differences in the handwriting of monozygotic and dizygotic twins in relation to personality
and genetic factors. M. Phil. Thesis, University of London.
Vestewig R. E., Santee A. A. and Moss M. K. (1976) Validity and student acceptance of a graphoanalytic approach to
personality. J. Person. Assess. 40, 592-597.
Williams M., Berg-Cross G. and Berg-Cross L. (1977) Handwriting characteristics and their relationship to Eysenck’s
Extraversion-Introversion and Kagan’s Impulsivity-Reflectivity dimension. J. Person. Assess. 41, 291-298.
Zdep S. M. and Weaver H. B. (1967) The graphoanalytic approach to selecting life insurance salesmen. J. appl. Psychol.
51, 295-299.

You might also like