SCIENCE OF HANDWRITING ANALYSIS

There is a wealth of material available on graphological research, theory, trait-stroke relationships, and effective real-world applications. Yet, skeptics can also be quite vocal about their opposition to the use of graphology because of their concerns about its lack of scientific validity. Some research psychologists have gone so far as to argue against the very existence of personality itself, the validity and usefulness of personality tests, and, by logical extension, the validity and usefulness of graphology. According to this line of reasoning, since there is no such thing as personality, then it makes no sense to use personality tests or analyze a person's handwriting to measure something that is non-existent. Consequently, any tools designed to measure personality, whether testing instruments or the Handwriting Analyst program, are inherently invalid.

Few psychologists agree, however, with this extreme position. Most of us recognize that we tend to behave in predictably patterned ways under a variety of conditions. Reliable knowledge of how a person acts, or might act, under certain circumstances can serve as a foundation for making sound, impartial decisions. Personality tests are questionnaires which are designed to measure these patterns, or traits. The results obtained by systematically gathering, scoring, and interpreting standardized personality tests have been obtained repeatedly by independent researchers. The appropriate use of scientifically validated personality tests is very useful - even essential - for making life-impacting decisions. Psychology departments in accredited universities along with thousands of practicing psychiatrists, psychologists and counselors, decades of scientific psychological research published in refereed professional journals, and courts of law across the country attest to their utility.

Personality tests are simply tools which can be used either appropriately or inappropriately, and the inappropriate use of a tool is not an argument against the appropriate use of that tool. If a hammer is used to hit someone, we don't argue that hammers are inappropriate tools for pounding nails. At the same time, we also recognize that a hammer in the hand of an experienced carpenter is a much more effective (i.e. appropriate) tool than a hammer in the hand of a novice. To be sure, some personality tests are more valid than others, just like some tools are better than others. Once again, a poorly-made tool is not an argument against a well-made tool. Though responsibility for an inferior tool falls squarely into the hands of the toolmaker, the choice and application of those tools rests firmly in the hands of the tool user.

Unfortunately, poorly designed graphological research studies abound, so evaluation of the body of findings as a whole can be equivocal. The designs employed in a number of graphological experiments were highly suspect and were neither replicable nor refutable. Other designs were based on flawed assumptions, codification procedures or conclusions, and served merely to confirm the subjective biases of skeptical investigators.

One example of a study with flawed assumptions, codification procedures and conclusions which confirmed the subjective bias of the investigator is provided by Bushnell (1996). In this study, 120 subjects were assessed by a standardized personality test (the Cattel 16PF) and by an earlier version of Handwriting Analyst. Two weeks after the assessment, each subject was presented with five Handwriting Analyst textual reports generated by the software program along with five Cattel 16PF personality reports produced by another software program, the "16PF Report Generator" (only one of five reports generated by each program was actually their own). The subjects were then asked to rank order each set of five reports on the basis of perceived accuracy (i.e., the probability that each report was their own, with 1 representing "most like me" and 5 representing "least like me").

The findings of this study supported the investigator's initial prediction that the 16PF personality reports would be reliably identified, while the handwriting analysis reports would not. Though acknowledging that "(this) direct comparison is of course between the two underlying assessment and interpretation packages adopted for this study rather than personality questionnaire and handwriting analysis methods themselves," he then concludes "The results therefore contribute to a literature that indicates a rather low possibility of handwriting analysis being useful and particularly indicates that on the basis of this research, the Handwriting Analyst system is not a sound choice for supporting selection and recruitment decisions...and once more calls the validity of handwriting analysis into question."

The use of perceived accuracy as a basis of validation is highly susceptible to contamination (even though researchers such as Crumbaugh and Stockholm found positive results using this methodology in 1977), and it is not an experimental procedure which is typically utilized for the validation of psychometric instruments. Standardized personality or psychological tests are not considered to be valid based upon the degree to which subjects are able to determine the accuracy of their test profiles. A rudimentary understanding of the function of defenses such as denial and repression - in addition to other sources of subjective perceptual distortion - would strongly suggest that some of the most significant features of personality assessment reports would be precisely those aspects which subjects would be LEAST likely to perceive accurately, simply because the information is so psychically uncomfortable and disruptive that they would be unwilling, or unable, to accept it.

Analytic psychology and projective testing are based upon this very understanding, and graphology is much more akin to a projective assessment technique than to an objective assessment methodology. In the mid 1920's in Switzerland, Max Pulver, graphologist and lecturer at the University of Zurich, demonstrated that both conscious and unconscious drives were projected in handwriting. In 1973, Felix Klein and his associates at the City University of New York found that graphology and the projective Thematic Apperception Test (T.A.T.) yielded statistically consistent personality profiles (though the sample size was quite small).

What was actually being measured by the methodology employed in the Bushnell study was the degree to which two software-generated reports could be accurately distinguished from "inaccurate" distractor reports produced by those same software programs. The only conclusion consistent with his findings was that the 16PF Report Generator produced reports which were accurately distinguishable from one another, while the the Handwriting Analyst did not. Only subjective experimenter bias could lead Bushnell to conclude on the basis of these findings that "The results therefore contribute to a literature that indicates a rather low possibility of handwriting analysis being useful and particularly indicates that on the basis of this research, the Handwriting Analyst system is not a sound choice for supporting selection and recruitment decisions...and once more calls the validity of handwriting analysis into question."

A much more reasonable explanation for these results is to be found in the software design differences between the two programs used in the study. Personality assessment software programs such as the 16PF Report Generator compose reports by producing pre-written paragraphs which appear when they are numerically triggered. Before a particular paragraph shows up in a report, all the items which contributed to that paragraph must be answered (i.e., an "all or nothing" approach). In contrast to these types of programs, Handwriting Analyst is designed to be open-ended and flexible in terms of the amount of information it requires for a report. Paragraphs are composed out of individual sentences, with each stroke or combination of strokes having a sentence or concept attached to it. Each sentence is carefully phrased to convey its associated concept accurately and to avoid implying a meaning which conflicts with another statement. Every output statement is then numbered, ordered, and grouped in associated conceptual categories (e.g. physical and material drives, emotional characteristics, social behavior, etc.), and the inter-relatedness of a particular trait with other traits becomes evident in the body of the report by simple proximity.

To accommodate the wide variety of handwriting samples potentially available for analysis, Handwriting Analyst© includes the response options of Mixed and Skip. The Mixed category accounts for combinations of handwriting characteristics while Skip covers inappropriate or unclear items. Under these conditions, since there is a high likelihood that writers will share a number of individual characteristics (especially mixed characteristics), there is an equally high likelihood that similar sentences will show up in the reports of different individuals, thereby making the reports less distinguishable from one another overall. In fact, we have found that the "Signature Only" reports are much more distinguishable from one another than are the "Text Only" reports, because the signatures themselves vary so much more from person-to-person. Interestingly enough, we have also found that people more readily accepted their "Signature Only" reports than their "Text Only" reports, most probably because signature represents public personality, which is the way the person wants to be seen by others.

While based on documented scientific findings in trait-stroke graphology, the Handwriting Analyst© software program is not a scientifically validated personality assessment instrument and has never been advertised as such. Rather, it is a closed, ready-to-run expert system which is insightful, practical and fun to use. People who were initially skeptical about graphology were frequently surprised by the accuracy of their reports. They were also impressed by the program's objectivity, since an accurate report could be produced by looking at the writing characteristics alone, without knowing the implications or desirabilities of the strokes. Consequently, they found that the analyses they received from the program were not based on writer conversations, observations, preferences or judgments, but were based on conclusions drawn from the writing itself, which were very amenable to agreement between independent raters (e.g., several people could readily agree that selecting small middle zone size clearly represented the size of the writer's sample).

A research version of Handwriting Analyst© could be developed, but such an instrument would have to undergo the same rigorous standardization, reliability, and validity studies as any other scientifically-validated personality test (i.e., by statistically establishing that "we are measuring what we think we are measuring, and if we did it again, would we get the same results"). It would then have to be distributed by reputable psychological test publishers and restricted to trained and licensed or certified professionals.

The most critical factor in the development of such an instrument would lie in the procedures for conceptualizing, measuring, and codifying handwriting characteristics. In 1985, Gilbert & Chardon used a computer to investigate the writings of 155 engineers and executives. These subjects, as part of an employment application, completed the Guildford-Zimmerman Temperament Survey, a standardized personality test. A worksheet was designed to record writing features on their job application letter, software was used to produce evidence of a connection between the two tests and factor analysis was used to establish the common factors which were present. These two authors have continued to use factor analysis in graphological research, with interesting results. Graphological research employing Klara Roman's Psychogram as a basis for systematic measurement of handwriting characteristics may also be a very promising approach to eventual psychometric validation.

Advances in computer and handwriting recognition technology are increasing the likelihood that a professional graphology software program may someday be designed where the user can scan the handwriting sample into the computer and instantly generate an accurate, customizable report. The primary obstacle which remains to be surmounted continues to be the difficulty inherent in recognizing handwriting in which each letter is not even the same size each time it is written. The software must recognize characters and be able to determine where one letter stops and another begins. Then, the program must be able to compare the relative sizes of the letters, determine how much they differ from each other (the variability of the writing), and compare the shapes and letter formations to the appropriate Standard of writing.

Once these measurement steps are implemented, the software design team must carefully define the rules for interpretation of the writing sample. While all graphologists use rules, most also reinterpret the meaning of strokes based upon the context in which they are found. They must be very clear about when a trait has the usual meaning and when it does not. The output statements must clearly express their meaning in ways that most clients will interpret correctly, they must be written separately from their context since the program will combine different statements for each report, and they must make sense regardless of the other traits with which they are combined. Only then could accurate, useful reports be produced.

Structurally, a professional graphology program should be flexible, with the content of the report determined by the needs of the client. It might contain an extensive list of personality traits to be analyzed, and the user could select those which were relevant to the specific client. Based on this user-defined set of traits, the program would select the appropriate items required to determine those traits, ask the necessary questions or apply the programmed expert-system decision rules, and compose a targeted report. A graph could then be printed, and additional information added by the user. The program might provide composite profiles for comparison (e.g., all successful employees, unsuccessful employees, a management team, a jury, etc.) and contrasting reports (e.g., comparing the traits of a manager to a job applicant or the traits of a jury to a defendant). One professional graphology program which implements some of these features was developed recently by Sheila Lowe.

Ongoing scientific research in graphology continues to be conducted by individual investigators and programs conducted under the auspices of professional graphological organizations such as the American Handwriting Analysis Foundation. Such projects include double-blind studies examining the statistical relationships between handwriting characteristics, observational ratings, performance evaluations, neurophysiological measures and standardized personality, interest and ability test scores (e.g., Hartford, 1973 and Wellingham-Jones, 1987, 1991). Additional information on recent studies and the current status of graphology can be found in the reference materials listed in the Bibliography. Rigorous certification programs have been instituted, and increasing numbers of individuals with advanced degrees in psychology or counseling are deciding to become Certified Graphologists (CG) and Certified Master Graphologists (CMG).