by Laura Weis with permission
Source: by Laura Weis with permission

A review of the literature over the past 50 years reveals a consistent but limited interest in the relationship between personality and graphology (Eysenck, 1945; Greasley, 2000; Guthke, Beckman & Schmidt, 2002; King & Koehler, 2000).  Some organisations still use graphology for use in selection. Indeed as Cook (1998) noted: ‘If handwriting accurately reflected personality, it would make a very cost-effective selection method, because candidates could be assessed from their application forms” (p. 15).

Beyerstein (2003) has argued that graphologists reproduce numerous, but essentially specious arguments for why handwriting analysis ought to work in the sense that it is a marker of personality.  Handwriting, they argue is brainwriting; writing is individualised, and personality is unique so each must reflect each other. Some say writing is a form of expressive movement so it should reflect our personality. Other argue that the police and courts use graphology so it must be valid. Some supposedly hard-nosed personnel managers swear by graphologists’ usefulness in selecting employees. Each of these suppositions is refuted by Beyerstein (2003), who has worked extensively in the area.  He notes: “Despite graphology’s poor showing in these well-controlled tests, both practitioners and a good portion of the public at large continue to believe it works” (p.13). 

Although the term ‘graphology’ goes back to 1871, when it was first used by the French cleric Michon, the belief that personality is somehow manifest in handwriting existed even before then.  Graphology books describe one both what factors to look at (e.g., size, slant, zone, pressure) and what traits (e.g., temperament, mental, social, work and moral) are ‘revealed’, though there appears no consistency between them, especially in the description of personality traits.  There is also rarely an explanation of the process or mechanism linking personality to graphology.  There are furthermore various schools of graphology, each with a slightly different history, approach and ‘theory’.

Eysenck and Gudjonsson (1986), thirty years ago suggested that there appear to be two different basic approaches to the assessment of both handwriting and personality, namely holistic vs. analytic.  This gives four basic types of analysis. Holistic analysis of handwriting: This is basically impressionistic, because the graphologist, using his or her experience and insight, offers a general description of the kind of personality he or she believes the handwriting discloses; Analytic analysis of handwriting:  This uses measurements of the constituents of the handwriting, such as slant, pressure, etc., which are then converted into personality assessment on the basis of a formula or code; Holistic analysis of personality: This is also impressionistic, and may be done after an interview, when a trained psychologist offers a personality description on the basis of her questions, observations and intuitions; Analytical analysis of personality:  This involves the application of psychometrically assessed, reliable and valid personality tests (questionnaires, physiological responses to a person, and the various grade-scores obtained).

This classification suggests very different approaches to the evaluation of the validity of graphological analysis in the prediction of personality. Holistic matching is the impressionistic interpretation of writing matched with an impressionistic account of personality.  Holistic correlation is the impressionistic interpretation of writing correlated with a quantitative assessment of personality, while analytic matching involves the measurement of the constituents of the handwriting matched with an impressionistic account of personality.  Analytic correlation is the measurement of the constituents of handwriting correlated with a quantitative assessment of personality. 

Most of the studies have been of the last type, but few have found any effects.  Furthermore, few graphologists seem ‘educated’ in personality language, processes or theory, preferring lay terminology and consequent tautology.  Greasley (2000) has pointed out that graphologist use analogy, symbolism and metaphor and have no clear, explicit, way of making personality assessments from handwriting script. 

 Furnham (1988) listed the conclusions drawn from six studies conducted in the 1970s and 1980s:

1) “It was concluded that the analyst could not accurately predict personality from handwriting.”  This was based on a study by Vestewig, Santee, and Moss (1976) from Wright State University, who asked six handwriting experts to rate 48 specimens of handwriting on 15 personal variables.

2) “No evidence was found for the validity of graphological signs.” This is from Lester, McLaughlin, and Nosal (1977), who used 16 graphological signs of Extraversion to try to predict from handwriting samples the Extraversion of 109 subjects whose personality test scores were known.

3) “Thus the results did not support the claim that the three handwriting measures were valid indicators of Extraversion.” This is based on the study by Rosenthal and Lines (1978), who attempted to correlate three graphological indices with the Extraversion scores of 58 students.

4) “There is thus little support here for the validity of graphological analysis.” This was based on a study by Eysenck and Gudjonsson (1986), who employed a professional graphologist to analyze handwriting from 99 subjects and then fill out personality questionnaires as she thought would have been done by the respondents.

5) “The graphologist did not perform significantly better than a chance model.” This was the conclusion of Ben-Shaktar and colleagues (1986) at the Hebrew University, who asked graphologists to judge the profession, out of eight possibilities, of 40 successful professionals.

6) “Although the literature on the topic suffers from significant methodological negligence, the general trend of findings is to suggest that graphology is not a viable assessment method.” This conclusion comes from Klimoski and Rafael (1983), based at Ohio State University, after a careful review of the literature.  Yet many of these studies could be criticized methodologically in terms of measurement of both personality and graphology.

Furnham and Gunter (1987) also found far fewer correlations than would be expected by chance between 13 different graphological measures and Eysenck’s EPQ.  They argued that graphological evidence was not related to personality, and noted that even if graphological analyses were valid, the theoretical basis of the method appears weak, non-explicit, and non-parsimonious.  “Perhaps one should be forced to conclude, rather uncharacteristically for researchers, that no further work needs to be done in this field” (Furnham & Gunter, 1987, p. 434).

Edwards and Armitage (1992) found that graphologists assigned 65 per cent of scripts of scripts to the correct categories (i.e., high-flyers vs. low-flyers), but that a control group of non-graphologists had a 59 per cent success rate.  They claimed their results showed that graphologist failed to substantiate claims made on their behalf.

Tett and Palmer (1997) found a high level of inter-rater agreement when coders were trained to measure specific bad handwriting elements supposedly linked to personality traits.  However, when correlating these measures with a test measuring 15 normal personality traits (the Jackson Personality Inventory), they found that only 5 per cent were significant in the expected direction, and 4 per cent in the opposite direction.  As before, the researchers concluded that handwriting analysis is limited value in predicting stable individual differences.

Ben-Shaktar, Bar-Hellel, Belen and Flug (1986) conducted a major and well-controlled study, and concluded that if a correspondence were to be empirically found between graphological features and such traits, it would be a major theoretical challenge to account for it. Further they argued that, unless the graphologist makes firm commitments to the nature of the correspondence between handwriting and personality, one can find ad hoc corroboration for any claim.  They also note that handwriting is paradoxically not a robust and stable form of expressive behaviour. It maybe actually extremely sensitive to extraneous influences that have nothing to do with personality (e.g., whether the script is copied or not, or the paper is lined or not, the condition under which the writing takes place, who reads the script).

In another review, Neter and Ben-Shaktar (1989) asked 63 graphologists and 51 non-graphologists to rate 1223 scripts.  They found that psychologists with no knowledge of graphology out-performed the graphologists on all dimensions, and they suggested that the limited validity of handwriting analysis is usually based on the script’s content rather than on its style.

King and Koehler (2000) demonstrated that illusory correlation phenomenon as a possible contribution of the persistence of graphology’s use to predict personality traits.  They found semantic association between the words used to describe hard writing features (e.g., bold) and personality traits was the source of the perceived correlation which, in part, ‘may partially account for continued use of graphology despite overwhelming evidence against its predictive validity’. (p. 336).  Guthke et al (2002) examined five graphologists’ ratings of social and cognitive inhibition, achievement motivation, Conscientiousness, frustration tolerance, and calmness of 60 undergraduates.  The judgements showed no statistical relationship to questionnaires measuring these traits.

Dean (1992) as well as Dean, Kelly, Saklofske and Furnham (1992) attempted to examine statistical effect sizes in this literature.  They also tried to explain why, if the empirical research literature is almost uniformly negative, it has not seem to have shaken graphologist’ or lay people’s faith in this type of analysis.

Dean (1992) found over 60 reliability and 140 effect size study results for his analysis.  The effect size is defined as the mean correlation (weighted by number of scripts) between personality as predicted from the handwriting (by graphologist or others), and personality determined by tests or ratings.  After looking at 1519 correlations, he concluded that effect sizes are too low to be useful, and that non-graphologists are generally as good at handwriting analysis as graphologists.  He admitted that there is an effect, but suggests that at least some of it is due to content, not actual writing, and that graphology is not valid or reliable enough to be useful.

Dean et al. (1992) attempted to explain why, if all the evidence suggests that graphology is barely related to any personality variable, do clients of graphologists attest to their accuracy? They list 26 reasons why clients are convinced that graphology works, none of which actually requires that it is true.  Interestingly, this may account for some graphologists’ unshakeable beliefs in their ‘art’.  For various (placebo-type) reasons clients believe that graphology works, which increases the graphologists’ belief in their own skill.  Hence each reinforces the other, despite the possibility that there remains no validity in graphological analysis. 

Thus people are convinced that handwriting is linked to personality, yet nearly all of the good evidence suggests this is not true (Beyerstein & Beyerstein, 1992).  As Driver, Buckley and Fruk (1996) have concluded: ‘While a few articles have proposed that graphology is a valid and useful selection technique, the overwhelming results of well-controlled empirical studies have been that the technique has not demonstrated acceptable validity.  A review of relevant literature regarding both theory and research indicates that, while the procedure may have an intuitive appeal, graphology should not be used in a selection context.’ (Driver et al., 1996. p. 78)

There is however few, if any, psychological studies on the association between graphology and intelligence (or cognitive ability).  However earlier studies showed among school children significant correlations between handwriting quality, intelligence and school performance (Oinonen, 1960).  Many books and web-sites on graphology suggest that it can be used to assess abilities though these are never clearly spelt out.  Certainly graphologists are strong advocates for its use in personnel selection (Klimoski, 1992).  Various studies have looked at the ability of graphology to predict job success but have showed poor predictive validity (Rafaeli & Klimoski, 1983).

But very few studies have linked graphological analysis to any ability tests.  One exception is the work of Lockowandt (1992), who tried to replicate earlier studies that showed graphologists could predict psychometric intelligence from handwriting samples.  He reported that he was unsuccessful but had colleagues who did find highly significant correlations between school qualifications and characteristics of handwriting such as distance between words and lines.  However in this study the relationship between exam handwriting and intelligence will be explored.  It has been established in meta-analysis that intelligence is probably the best psychometric predictor of success at work (Drasgow, 2003).  Thus if graphology can predict work success it should be systematically related to intelligence.

Furnham, Chamorro-Premuzic and Calahan (2003) got students to complete the Big Five Personality Inventory (NEO-PI-R; Costa & McCrae, 1992) and three different intelligence measures (Baddeley, 1968; Philips & Rawles, 1976; Wonderlic, 1992) soon after arriving at the beginning of their course.  These scores were then related to a reliable graphological analysis of their hand-writing in exam scripts few months later (study 1) and twenty-months in another (study 2).  Results showed that the 14 graphological variables factored into two interpretable factors called dimension (size, width, pressure, percentage used etc) and details (loops above, below, dotted i’s, crossed t’s).  Correlational and regression analysis in both studies showed fewer associations with the Big Five personality variables than maybe expected by chance.  Graphological variables did correlate with both participants’ gender and intelligence but the pattern was different in the two studies reinforcing the idea that chance factors are operating here.  Thus, once again, despite attempts to use both psychometrically valid personality measures and reliably measured handwriting factors collected under non self-conscious conditions there appears to be no robust relationship between graphology and personality.

More recent studies have also found essentially little evidence for the validity of graphological manifestations of personality. Thus in a comparison of the clinical findings of a patient to a blind handwriting analysis of the same patient done independently by a professional graphologist found graphological inferences made from the five unique composite profiles to demonstrate strong similarities with diagnoses on Axis I and Axis II of DSM-IV-TR (APA, 2000) of the same patients, supporting the value of graphology in psychological assessment (Cronje & Roets, 2013).

Wang et al. (2009) found that Chinese character spacing was related to two traits of Cattell’s 16PF: reasoning and sensitivity, but not enough of the others suggesting a possible chance finding.

Gawda (2014) concluded, in two studies looking at whether graphic characteristics relate to two personality assessments, NEO-FFI and EPQ-R, that there was no writing characteristics specific to personality traits, therefore no evidence for assessment of personality on the basis of handwriting.

Gawda (2016) more recently found that out of the 32 specific graphical characteristics of handwriting, 7 of them were significantly different between patients diagnosed with schizophrenic disorders and healthy controls. These include the calligraphic forms of letters, loops in ovals, lacking of dots, tremor, sinusoidal baseline, and irregularities size of lower zone. The findings however could be explained in terms of motor disturbances in schizophrenia.

Handwritings of 30 patients with major depressive disorder and 30 patients in the bipolar depressive phase were compared to samples from 30 patients in the bipolar manic phase and 60 healthy controls. The findings show 3 out of 32 handwriting parameters to be different between controls and patients with both major depression or depressed bipolar patients. However, no significant difference was found in the graphical aspects of handwritings by depressed patients and bipolar patients in the manic phase (Gawda, 2013).

The jury has returned: Almost fifty years of research has shown precious little evidence that graphology is a good index of personality or intelligence. Careful researchers and impartial reviewers of this scattered and patchy literature have returned their verdict. Hence a rather unusual conclusion: No further work needs to be done in this area.


Ben-Shaktar, G., Bar-Hillel, F., Bilen, Ben-Abba, E., & Flug, A. (1986). Can graphology predict occupational success? Two empirical studies and some methodological ruminations. Journal of Applied Psychology, 71, 645-653.

Beyerstein, B. (2003).  How graphology fools people. 01QuackeryRelatedTopics/Tests?grapho.html

Beyerstein, B., & Beyerstein, D. (Eds.) (1992). The write stuff: Evaluation of Graphology – The study of Hand writing Analysis.  New York: Prometheus Book.

Bushnell, I. (1996).  A comparison of the validity of hand writing analysis with that of the Cattell 16PF.  International Journal of Selection and Assessment, 4, 12-17.

Cook, M. (1998).  Personnel Selection.  Cluchester: Wiley.

Costa, P., & McCrae, R. (1992). The Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor inventory (NEO-FFI).  Professional Manual, Psychological Assessment Resources, Odessa, FL.

Cronje, P. E., & Roets, H. E. (2013). Graphology in Psychological Assessment: A Diagnosis in Writing. Universal Journal of Psychology, 1(4), 163-168.

Crumbough, J. (1988).  Comment on the graphological studies of Lester, McLaughan and Nosal.  Perceptual and Motor Skills, 45, 494.

Crumbough, J. (1988).  On the validity of graphoanalysis.  Perceptual and Motor Skills, 67, 461-462.

Crumbough, J., & Stockholm, E. (1999).  Validation of graphological analysis by global or holistic method.  Perceptual and Motor Skills, 44, 403-410.

Dean, E. (1992). The bottom line: effect size.  In B. Beyerstein & D. Beyerstein (Eds). The Write Stuff: Evaluation of Graphology. Buffalo, NY: Prometheus Books, pp. 269-340.

Dean, E., Kelly, I., Saklofske, D., & Furnham, A. (1992). Graphology and human judgement.  In B. Beyerstein and D. Beyerstein (Eds.) The Write Stuff: Evaluation of Graphology. Buffalo, NY: Prometheus Books, pp. 269-340.

Driver, R., Buckley, M., & Fruk, D. (1996).  Should we write off graphology? International Journal of Selection and Assessment, 6, 78-84.

Edwards, A & Armitage, P. (1992).  An experiment to test the discriminating ability of graphologist. Personality and Individual Differences, 13, 69-74.

Eysenck, H. (1945). Graphological analysis and personality: An experimental study.  British Journal of Psychology, 35, 70-81.

Eysenck, H. & Gudjonsson, G. (1986).  An empirical study of the validity of hand writing analysis.  Personality and Individual Differences, 7, 263-264..

Fluckinger, F., Tripp, C. & Weinberg, E. (1961).  A review of experimental research in graphology: 1933-1960.  Perceptual and Motor Skills, 12, 67-90.

Furnham, A. (1988).  Write and wrong: the validity of graphological analysis.  Skeptical Inquirer, 13, 64-69.

Furnham, A. & Gunter, B. (1987).  Graphology and personality: Another failure to validate graphological analysis.  Personality and Individual Differences, 8, 433-435.

Furnham, A., Gunter, B., & Peterson (1994). Television distraction and the performance of introverts and extraverts.  Applied Cognitive Psychology, 8, 705-711.

Galbraith, D. & Wilson, W. (1964). Reliability of the graphoanalytic approach to hand writing analysis.  Perceptual and Motor Skills, 19, 615-618.

Gawda, B. (2013). Little evidence for the graphical markers of depression1. Perceptual & Motor Skills, 117(1), 304-318.

Gawda, B. (2014). Lack of evidence for the assessment of personality traits using handwriting analysis. Polish Psychological Bulletin, 45(1), 73-79.

Gawda, B. (2016). Dysfluent handwriting in schizophrenic outpatients. Perceptual and Motor Skills, 0031512516637019.

Greasley, P. (2000). Handwriting analysis and personality assessment: The creative use of analogy symbolism and metaphor.  European Psychologist, 5, 44-51.

Guthke, J., Beckmann, J. & Schmidt, G. (2002).  Is graphology of diagnostic value at all? Zeitschrift fuir Personalpsychologie, 1, 171-176.

Hall, K. (1999).  Practical guide to handwriting analysis.

Kimmel, O. & Wertheimer, M. (1966).  Personality ratings based on handwriting analysis and clinical judgement.  Journal of Projective Techniques and Personality Assessment, 30. 177-178.

Klimoski, R. (1992).  Graphology and personnel selection.  In B. Beyerstein and D. Beyerstein (Eds.), The write stuff, 232-268.  New York: Prometheus Books.  

Klimoski, R., & Rafaeli, A. (1983).  Inferring personal qualities through hand writing analysis.  Journal of Occupational Psychology, 56, 191.

King, R & Koehler, D. (2000).  Illusory correlations in graphological analysis.  Journal of Experimental Psychology: Applied, 6, 336-348.

Lemke, E. & Kirchner, J. (1972). A multivariate study of handwriting, intelligence and personality correlates.  Journal of Personality Assessment, 25, 584-592. 

Lester, D., McLaughin & Nosal, G. (1977).  Graphological signs of Extraversion.  Perceptual and Motor Skills, 44, 137-138.

Lockowandt, O. (1992). The problem of the validation of graphological judgements.  In B. Beyerstein and D. Beyerstein (Eds.). The Write Stuff.  New York: Prometheus Books, pp. 86-104.

Neter, E. & Ben-Shakhar, G. (1989).  The predictive validity of graphological inferences: A meta-analytic approach.  Personality and Individual Differences, 10, 137-145.

Oinonen, P. (1961). Poor hand writing as a psychological problem.  Acta Academicae Paedagogicae. Jyvaskylamisis. Jyvaskyla, 21.

Oosthynzen, S. (1990).  Graphology as a predictor of academic achievement.  Perceptual and Motor Skills, 71, 715-721.

Rafaeli, A. & Klimoski, R. (1983).  Predicting sales success through handwriting analysis.  Journal of Applied Psychology, 68, 213-211.

Rosenthal, D. & Lines, R. (1978).  Handwriting as a correlate of extravrsion.  Journal of Personality Assessment, 42, 45-48.

Tett, R. & Palmer, C. (1997). The validity of handwriting elements in relation to self-report personality trait measure.  Personality and Individual Differences, 22, 11-18.

Vanderberg, S. & Kuse, A. (1978). Mental rotation, a group test of three-dimensional spatial visualization. Perceptual and Motor Skills, 47, 599-604.

Vestewig, R., Santu, S., & Moss, M. (1976).  On the validity of graphoanalysis.  Journal of Personality Assessment, 41, 589-600.

Vine, I. (1974).  Stereotypes in the judgement of personality from hand writing.  British Journal of Social and Clinical Psychology, 13, 61-64.

Wang, T., Chen, Z., Li, W., Huang, X., Chen, P., & Zhu, S. (2009, December). Relationship between personality and handwriting of Chinese characters using artificial neural network. In Information Engineering and Computer Science, 2009. ICIECS 2009. International Conference on (pp. 1-5). IEEE.

About the Author

Adrian Furnham, Ph.D.

Adrian Furnham, Ph.D., is a professor of psychology at University College London and the Norwegian Business School.

You are reading

A Sideways View

The Freudian Account of Leadership Failure and Derailment

What do Psychoanalysts have to say about how and why leaders fail?

What Men Desire in a Woman

What characteristics are men most attracted to in women and why?

The Psychology of Queuing

Why is waiting in line so horrid?