APPROVED FOR RELEASE
CIA HISTORICAL REVIEW PROGRAM
22 SEPT 93
A professional assessor supports the amateur graphologist's appeal for validity tests, although not sharing his enchantment with the art.
THE ASSESSMENT OF GRAPHOLOGY
E. A. Rundquist
Two threads of argument run through the foregoing article on handwriting analysis. The first asserts the great need for research studies because "a proper test run has never been devised and carried out, at least not in the United States, to determine whether any graphologist can consistently deliver accurate results in the area of character delineation." The second asserts the value of graphology here and now as an assessment technique, making sweeping claims of what it can do. The arguments are essentially incompatible. If the claims are correct, the research is unnecessary; if there is no research evidence, the claims are unsupported. With the need for research to establish the value of graphology as an assessment technique I am in full agreement. I disagree with the claims for its current effectiveness.
The article makes a number of cogent points. It distinguishes between the well-established branch of graphology devoted to problems of personal identification and the branch devoted to character analysis; it stresses the need for research studies; it recognizes many of the pitfalls that need be avoided in carrying out such studies; it acknowledges that traditional psychological assessment is preferable to handwriting analysis when direct access to the individual is possible. With these points I am in general agreement. A little elaboration of all but the first, which is too well established to require comment, may be helpful.
Scope of Research
In evaluating graphology--or any other assessment technique--not just one, but many studies are required. Studies of agreement among graphologists, the development of objective techniques for measuring characteristics of handwriting, refinements in the methods used, hypotheses such as "small handwriting with closed and knotted o's and a's indicates secretiveness"--all these are useful and interesting, but they do not answer the main question: How well does it in fact predict behavior? Or in the terms psychologists like to use: What is its validity? Studies should therefore be concentrated in this area, a point I stress not in disagreement with Mr. Laycock, but because of its importance.
Validation studies in the area of personality assessment are not easy to do. There are many complicating factors--getting a representative sample of persons to participate, getting the same kind of information about each, getting information in sufficiently specific terms on the behavior one is trying to predict. This last problem is recognized by Mr. Laycock as a semantic one. "What is a brave man?" he asks. If there is no agreement on what a brave man is, there is obviously no means of checking on anyone's assertion that a person is brave.
This semantic problem has another aspect which is often overlooked. It is not hard to write a personality description that applies to the vast majority of people. This "Barnum effect," as it has been called, is one of the charlatan's best friends. Psychologists prepare such descriptions to show their students that a person's agreement with the correctness of a personality description is not proper evidence of the value of any assessment technique. I once capitalized on the Barnum effect when instructing a group of twelve European intelligence officers, most of whom were favorably inclined toward graphology. I asked them for handwriting specimens, and after a suitable interval produced personality descriptions for each of them, which ten of the twelve agreed fit very well. Then they were allowed to discover that I had given them all the same identical description, one I found in a German periodical before I left the States.
To demonstrate further the dangers of accepting agreement with a personality description as evidence in favor of any assessment technique, I asked the twelve to describe themselves by answering true or false to a number of personality statements. All answered true to two of the statements--"You have a tendency to be critical of yourself" and "You prefer a certain amount of change and variety and become dissatisfied when hemmed in by restrictions and limitations." And ten answered true to a third statement--"While you have some personality weaknesses, you are generally able to compensate for them." Experimenting thus with a few more questions, one would soon have enough "true" statements to write a full description which every member of a group would agree applied to himself. This kind of demonstration underscores the passages in Mr. Laycock's article which call for specific, objective, and understandable items of behavior as the criterion or yardstick by which the validity of any assessment technique must be judged.
Capabilities of Psychological Assessment
"Where full access is possible, a battery of tests, particularly of the real-situation type used in OSS, and a careful study of the subject's past performance and reputation will give as reliable a result as we can expect at this stage of our knowledge of man and yield something like a scientific picture of his inner workings." I take this to mean that direct assessment of the kind done by my staff in the CIA Office of Training is to be preferred over the graphological technique when access to the individual is possible. With this view, of course, I should like to agree wholeheartedly. But this brings me back to the article's claim that "In most cases, competent graphologists can supply reliable estimates on ... disposition to talk too much . . . emotional stability under stress . . . aggressiveness, resistance, and tenacity . . . attitude toward money . . . disposition to deceive . . . inclination toward opportunism . . . desire for power . . . willingness to follow the lead of others . . . rebelliousness . . . rashness" and "reasonably good estimates on ... capacity for abstract thinking and logic . . . ability to deal with people, powers of observation, imagination" as well as "a good educated guess" about "sex difficulties . . . disposition to engage in criminal activities ... disposition to engage in violence against persons."
Even for the extremely thorough assessment process conducted by my staff I would not claim so much. Either our own methods have greater capabilities than we credit them with, or the article errs in conceding the superiority of "direct-access" assessment over handwriting analysis. If evidence can be produced to establish that graphology can do all this, I shall hasten to incorporate it into our assessment process and eliminate much of the interviewing and testing we do.
"There are certain things a graphologist can not tell," writes Mr. Laycock. Certainly my list here would be much longer than his. But I am genuinely puzzled by some of the things included in this list, and by the statement that "the graphologist is entitled to know the writer's age, sex, national origin, and profession, since he cannot tell these facts from the specimens, and they are invaluable interpretive aids." I am confused by the inclusion of sex, because there are studies indicating quite clearly that differences in handwriting do exist1 which permit determination of sex at a better than chance level. I haven't seen any studies on the other characteristics, but except for exact profession they are the kind of thing I would think might be inferred from handwriting at a little better than chance level.
Psychologists are impressed by the difficulty of making predictions about a changing individual in a changing environment. They are very much aware that such predictions can refer only to probabilities. Psychologists desire, therefore, as the core of their assessment process, means and techniques which have been validated by methodical research. Tests of general intellectual ability, of some aptitudes, and of interests, along with information about past behavior, are among these means. New means can be developed only by testing claims for special techniques in the same methodical way.
Prospects for Graphology
Up until recently the evidence concerning graphology as an assessment technique has been so negative that psychologists generally have preferred to concentrate on techniques that showed more promise. The negative evidence came from studies of graphological tenets equating specific handwriting characteristics, such as upward sloping lines, with specific traits, such as ambition. On the basis of such studies, graphology as a means of assessment has been lumped with astrology, phrenology, and other systems for reading character from physical characteristics such as length of fingers or color of hair. Handwriting is, however, the product of a person. There is therefore some reason to expect it might tell something about him. This reasoning, fostered by graphology itself as it became concerned with the movements underlying handwriting rather than the handwriting itself, has led to the devising of different kinds of studies. These studies, while not yet convincing, do make it clear that the value of graphology is not yet a closed question. One of the better ones, for example, found that a graphologist trying to infer from handwriting how 50 neurotics would answer 27 questions (1,350 items in all) achieved an accuracy of 62 percent as against the 50 percent to be expected by chance.2 The graphologist may have been helped by knowing that all were neurotics, and so the 62 percent may be a bit high. Even taking the data at face value, these predictions turned out not much better than chance results; but the study suggests that research in this area might be more worthwhile than many had thought. It also points to the need for more research to pin down just what kinds of things can be predicted and what kinds of things cannot.
In Mr. Laycock's list of things a graphologist can determine is included "important changes in character (by comparison of present with past calligraphies)." Research on change in handwriting over time and under various conditions appears to offer some promise. At least common observation suggests that changes are caused by illness, either physical or mental.
At the present time I do not consider the evidence for graphology as an assessment technique sufficiently impressive to include it in assessments for which we have direct access to the individual. I don't sponsor research on it for this purpose as a matter of economics. I have only so many dollars, and I think I will get a better return from other assessment techniques. And even if we did not have access to the individual, I'd still place my bets on investigation of his past behavior, his education, his jobs, social status, income, and so on.
It is interesting that graphologists, according to the article, require some of this investigative information (sex, age, national origin, and profession) as a prerequisite for their analysis. They also get the informational content of the handwriting specimens themselves. From these data a number of inferences are already possible. Consider, for example, the differences in characteristics one might assume with confidence between an age 50 female English secretary and a 21-year-old male German lawyer. I'd be inclined to rely on the implications of this information, and would be extremely cautious in accepting inferences, whatever their source, that were inconsistent with it. The article claims, for example, that "most accountants and bookkeepers can tell you, without even thinking, how a man feels about money by the way he writes a check." I'd rather have the evidence on how he uses his money that can be obtained by looking at his cancelled checks. So, I guess, would the banker, who lends money on investigation of background and permanence of job, not on handwriting analysis. It is dangerous to allow inferences from less well validated information to influence those obtained from valid sources.
For the clandestine services, however, graphology as a validated assessment technique might have application in a sufficient number of instances, those where background investigation is impossible, to warrant considerable research to determine its effectiveness. I would like to see these studies start on whatever simple verifiable characteristics graphologists are willing to try. Should these prove successful, studies of more complex traits can be undertaken.
I can agree with Mr. Laycock that the study should cover the abilities of a number of particular graphologists-that graphology may be an art, but certainly is not a science. In my mind there is even the nagging question whether it is a practical art. A problem with an art is that a particular person's skill in applying it may change over time. There is no way of knowing whether a practitioner's predictions a year later will have the same value as those he made when he was tested. It is for this reason that psychologists, as scientists, keep trying to find ways to convert the art of judging people to a science. They try to tease out, objectify, and measure the basis for their predictions, so that the assessment skill can be communicated to others and used reliably with a variety of persons in a variety of situations.
Mr. Laycock is greatly concerned with getting some research started. So am I; for until we get more information on the validity of graphology for specific purposes, the differences between his views and mine on graphology as an assessment technique, and my concern over the danger of unwarranted credence in graphological findings, will persist. Psychologists charged with personnel assessment are ready to cooperate in such studies. Their only requirement is that the research be so conducted that a group of scientists will agree on the kinds of conclusions that can be drawn from it.
1 A. Anastasi & J. P. Foley, Jr., Differential Psychology (Revised edition; New York: Macmillan, 1949) p. 663; C. L. Hull, Aptitude Testing (Yonkers-on-Hudson: World Book Company, 1928) p. 147.
2 H. J. Eysenck, "Graphological Analysis and Psychiatry: An Experimental Study," British Journal of Psychology, 1945, 35, 70-81.
Top of page