What Your Drawing Says, According to the Machover Method (and What the Real Evidence Says)

In 1949, clinical psychologist Karen Machover published Personality Projection in the Drawing of the Human Figure, a book that would become one of the most widely used projective techniques of the 20th century. Her premise: when you ask someone to draw “a person,” they project onto the drawing aspects of their own body image and personality, in a way similar to how the Rorschach uses inkblots or the TAT uses ambiguous scenes: none of the three asks for a “correct” answer, and it’s in that ambiguity that projective theory locates the involuntary reveal of personality.

A system of isolated signs

Machover interpreted isolated features of a drawing. Large eyes signaled suspicion or paranoid tendencies. Missing facial features pointed to depression. Heavy shading reflected aggressive impulses. Multiple erasures, anxiety. Emphasis on the ears was linked to paranoia; emphasis on hair, to sexual concerns; inanimate accessories like guns or knives, to delinquent tendencies. On the eyes specifically, Machover wrote that they aren’t just “the window of the soul” revealing someone’s inner life, but the basic organ of contact with the outside world, which is why she considered them the point where both the sense of self and its vulnerability concentrate. Machover also proposed her body image hypothesis: the drawn figure would embody the central psychological and physical attributes of the person drawing it.

Our human figure drawing test today uses fifteen of these real indicators: the five that can be measured directly from the canvas (size, placement on the page, presence of hands and feet, torso shape, symmetry), plus ten more that depend on recognizing content or a choice of tool (head size, eyes, ears, hair, shoulders, clothing, shading, erasing, line thickness, and use of color). The last two, thickness and color, are recorded with certainty because the test itself offers a palette of widths and tones to pick from; the rest of the stroke-based signs, like shading with a real pencil, Machover could only infer from pressure on the paper, something a mouse-drawn image doesn’t reproduce. Next to each of the fifteen, the report shows something Machover never could: decades of later research testing whether her claims hold up.

A quick look at some of the most-cited signs, before getting into the full review:

Sign	Machover claimed	Evidence
Large eyes	Suspicion, paranoia	No support
No facial features	Depression	No support
Heavy shading	Aggression, anxiety	No support
Multiple erasures	Anxiety	No support
Emphasis on ears	Paranoia	No support
Emphasis on hair	Sexual concern	No support
Rounded torso	”Feminine” traits	Real support
Drawing in color	Anxiety	Real support

A near-unanimous verdict: 2 out of 30

Starting with Swenson (1957), a string of reviewers over four decades reached a nearly unanimous conclusion: the vast majority of human figure drawing signs have negligible or no validity at all. The most cited review is Kahill (1984), which analyzed the literature published between 1967 and 1982 on 30 specific Machover indices. The result: only 2 of the 30 found consistent support: a rounded torso as an indicator of traits labeled “feminine,” and drawing in color as an indicator of anxiety. The other 28, including the most popularly cited ones (ears and paranoia, internal organs and schizophrenia, accessories like weapons and delinquency, emphasis on hair and sexual concerns), showed no consistent relationships.

Worse still for the original theory, several findings ran in the opposite direction from what was predicted:

• Dudley, Craig, Mason, and Hirsch (1976) found that depressed people were less likely to draw faces in profile, the exact opposite of what Machover predicted.
• Cvetkovic (1979) found that people with schizophrenia were less likely to draw heads with no body, again the opposite.
• Handler and Reyher (1965), reviewing research on signs and anxiety, found that 30 of 255 findings were significant in the direction opposite to what was predicted.

Why reliability doesn’t help either

Even setting validity aside (whether a sign predicts what it claims to predict), reliability (whether different raters score the same drawing similarly) is uneven. Kahill (1984) found these ranges of agreement between raters, too wide to trust any single sign without checking it case by case:

• Shading: between .45 and .96.
• Head size: between .54 and .99.
• Facial expression: between −.13 and .60 (a negative value means that, in at least one study, two raters landed on opposite scores for the same drawing).

Vass (1988) summed up the general state of projective drawing tests as having “serious problems of reliability and validity.”

The phenomenon that explains why the method survived anyway

If the evidence is this unfavorable, why did so many clinicians keep (and some still keep) trusting these signs? A classic study by Chapman and Chapman (1967) offers the most-cited explanation: illusory correlation. When a sign has a strong, intuitive semantic connection to a psychological trait (large eyes with the wariness typical of paranoia is the textbook example), clinicians tend to perceive that sign showing up alongside that trait more often than it actually does in the data, even when two observers are looking at the exact same material.

A study by Smith and Dumont (1995) shows the problem in real practice: they asked 36 clinical and counseling psychologists (58% with specific training in projective techniques) to interpret human figure drawings out loud. Of the 22 who drew clinical conclusions, 20 based at least part of their interpretation on specific isolated signs, with statements like “the eyes are strange and exaggerated, may have issues with men, some paranoid suspicion” drawn from a single sketch. Stricker (1967), for his part, found that clinicians with specific training in the test were actually worse at identifying psychopathology than graduate students with no such training, a result that directly undercuts the added value of “knowing how to read” these signs.

A broader review, Lilienfeld, Wood, and Garb (2000), on the scientific status of projective techniques as a whole (Rorschach, TAT, and human figure drawing), reached a similar conclusion: the vast majority of indices across these three tests have “negligible or no validity,” with only a handful of specific exceptions that do hold up against data, much like the 2 out of 30 Machover signs here.

The little that does seem to hold up

Not everything is negative. There’s evidence that global approaches (judging a drawing’s overall quality, not isolated signs) can reach modest validity: poor overall drawing quality has long been recognized as a rough, if crude, barometer of psychopathology (Swenson, 1957). That’s a much weaker and less specific kind of validity than the one Machover claimed for her individual signs, and it isn’t something that can be responsibly automated in an online entertainment test, which is why our test sticks to the fifteen honestly measurable indicators, not a global verdict on “how healthy” your figure looks.

If you want to see these fifteen indicators applied to your own drawing, with the same evidence cited here next to each one, you can take the Human Figure Drawing Test. And if you’d rather take a drawing test that makes no claim about content, only about rhythm and use of space, the Expressive Style Quiz is the other half of this category.

Sources

• Machover, K. (1949). Personality Projection in the Drawing of the Human Figure. Charles C. Thomas.
• Swenson, C. (1957). Empirical evaluations of human figure drawings. Psychological Bulletin, 54, 431-466.
• Kahill, S. (1984). Human figure drawing in adults: An update of the empirical evidence, 1967-1982. Canadian Psychology, 25, 269-292.
• Chapman, L. J., & Chapman, J. P. (1967). Genesis of popular but erroneous psychodiagnostic observations. Journal of Abnormal Psychology, 72(3), 193-204.
• Smith, D., & Dumont, F. (1995). A cautionary study: Unwarranted interpretations of the Draw-A-Person test. Professional Psychology: Research and Practice, 26(3), 298-303.
• Lilienfeld, S. O., Wood, J. M., & Garb, H. N. (2000). The scientific status of projective techniques. Psychological Science in the Public Interest, 1(2), 27-66.
• Stricker, G. (1967). Actuarial, naive clinical, and sophisticated clinical prediction of pathology from figure drawings. Journal of Consulting Psychology, 31, 492-494.
• Handler, L., & Reyher, J. (1965). Figure drawing anxiety indexes: A review of the literature. Journal of Projective Techniques and Personality Assessment, 29(3), 305-313.
• Vass, Z. (1988), cited in the original article for its overall assessment of the reliability and validity of projective drawing tests; we haven’t been able to independently verify its complete bibliographic record.
• Dudley, H. K., Craig, P., Mason, R., & Hirsch, S. (1976), finding on depression and profile drawing cited in Kahill’s (1984) review; we haven’t independently verified the exact title of the original study.
• Cvetkovic, M. (1979), finding on schizophrenia and headless-body drawing cited in Kahill’s (1984) review; we haven’t independently verified the exact title of the original study.
• You can also read the Spanish-language Wikipedia article on the Draw-A-Person Test for a general introduction to the method.