Giving up Cherished Ideas: The Rorschach Ink Blot Test¹

Robyn M. Dawes^*

One of the most dearly (expensively) held beliefs of many clinical psychologists is the belief in the validity of Rorschach inkblot interpretation.  While this belief may be common in the general American population, it is particularly strong among clinical psychologists, many of whom still give Rorschachs despite the consistent research findings — of literally thousands of published studies — that the Rorschach interpretation is unreliable and invalid.  The plausibility of Rorschach interpretation is so compelling that it is still accepted in court proceedings involving involuntary commitment and child custody, with psychologists who offer such interpretations in these hearings being duly recognized as "experts."¹

American Psychological Association rules of ethics prohibit my presenting an example of a Rorschach inkblot.  (Presumably, prior exposure to these blots would contaminate the validity, if there were any, of any subsequent use.)  Suffice it to say that there are ten blots on cards roughly the size of regular typing paper.  Six of these are black and various shades of gray; the remaining four have color.  The blots themselves cover roughly half the area of the cards on which they are reproduced, in a horizontal orientation — that is, the position of a sheet of typing paper turned on its side.  These blots are symmetric around a vertical axis in the middle of the card.  They were developed by the psychiatrist Herman Rorschach (1884-1922), for purposes totally unrelated to assessing character structure and personality problems.

The subject is asked to say what the cards look like to him or her.  The instructions are purposely vague, allowing subjects to make associations from the form, shading, color, or texture of the blots.  Moreover, the subject can respond to each blot in its entirety, to major portions of the blot ("large details"), or to small details in the blot's structure; subjects are also free to make use of the white spaces surrounding the blot or within it.  Finally, the subject is free to rotate the cards from the positions in which they are presented — and even to turn cards over and look at the back of them.

After the subject gives a response, the examiner asks him or her to explain it with such questions as "Why does it look like a bat?", "What in the blot makes it look like your grandfather drunkenly falling off his chair at his fifty-fifth wedding anniversary celebration?", or simply, "Tell me more."  Moreover, subjects are urged to see more than one percept per blot with queries such as "Anything else?"

The theory behind the test is simple.  The world contains ambiguity; people respond to the ambiguity in habitual ways, and the more ambiguous the situation in which they find themselves the more important these habitual response styles become.  An inkblot, being the ultimate of ambiguity, should be an ideal way to "tap into" such habitual responses.  Moreover, the content that people see may give valuable clues about the types of materials they "have on their mind" in that these are free to be projected into the stimulus situation, because it in fact has no structure of its own.  Hence, the Rorschach is termed a projective test.  Moreover, as Freud has suggested, the content of dreams and fantasies is particularly indicative of our unconscious needs and conflicts, because there is no external stimulus to which we are responding.  By virtue of being a stimulus with minimal structure, the Rorschach inkblots elicit projections of internally generated "percepts" — which can then be used to make inferences about unconscious needs and conflicts.

The theory is not only plausible but compelling.  For example, I recall testing a very depressed individual who immediately responded to a blot by saying, "It looks like a bat that has been squashed on the pavement under the heel of a giant's boot."  What response could possibly be more "one-down?"  The fact that the individual was obviously depressed led to my belief in the validity of the Rorschach.  (Note that my observation can be reframed to indicate that his response provided me with no information that I did not already have.)  Of course, if he had been obviously psychotic, I could have noted that his percept concerned material not present in the blot (e.g., the giant and the boot).  That would have also impressed me, because that response would indicate how he attended to stimuli not present in the environment — virtually the definition of psychosis.  Or, if he had suffered from aggressive outbursts, I would have noted the hostility in the response.

I also recall testing a homosexual male nurse (at the time, homosexuality was termed a "disease") who gave approximately forty "vista" responses — for example, vistas of Chinese junks on lagoons with mountains in the background.  At the time, the prevailing theory about the etiology of male homosexuality was that it was due to childhood withdrawal of feelings from an overpowering mother who aroused incest fantasies and identification with a weak, passive father — who had to be weak and passive or the mother would not have been that way.  How clearly these vista responses indicated the man's pathological tendency to distance himself from emotionally threatening material!

The compelling plausibility of Rorschach interpretation should now be apparent.  Clearly, for example, responding to the blots in their entirety would seem to indicate a tendency to search for the "big pictures" in life — even when they aren't there; motion responses must indicate an active imagination; the use of white spaces, a tendency toward oppositionality and perverseness, et cetera, et cetera.  Moreover, seeing something that the examiner cannot see must indicate very poor "reality testing" — most probably psychosis.  (At one staff meeting I attended, the head psychologist successfully lobbied to have someone labeled "schizophrenic" after waving a Rorschach blot in front of the group and demanding, "Does this look like a bear to you?")  Like the unstructured interview, the Rorschach inkblot test is a major technique used by many clinical psychologists.

In contrast to the compelling plausibility of the inkblot test, what does the research show?  Based on thousands of studies addressed to this question, the answer is simple: damn little support for the projective hypothesis.  For example, one consistent finding is that the number of responses the subject makes correlates with scores on intelligence tests.  But then again the amount a subject talks in any situation may have such a relationship, and intelligence tests are better measures of intelligence than is the Rorschach test.²

There are also certain intriguing findings, such as that concerning the "index of existential pathology."  This index refers to the tendency to see part-human, part-nonhuman things — for example, cartoon characters, elves, satyrs, and witches.  One study indicated that "neurotics" have a much higher tendency than do "normals" or "eminent physical scientists" to see such things.³  This finding appeared, however, at the end of a long paper on the "psychodynamics of eminent physical scientists" in which none of the other hypotheses tested was supported-for example, the hypothesis that eminent physical scientists should have a greater tendency than others to refer to "mother nature."  Moreover, the scores on this index were trichonomized in a post hoc manner into 0-3, 3-6, and 6 or more.  Perhaps this categorization was made to maximize the value of the statistic used to assess "significance."  Moreover, there is no mention in the literature of replicating this difference.  I mention this finding because it is typical of intriguing findings involving the Rorschach.  They appear and then disappear from our body of knowledge.

What about the basic dimensions of personality and psychopathology that the Rorschach purports to assess as a projective device?  Does it work?  The answer to this question — at least up until 1978 — may be found by reading the reviews of the Rorschach in the Mental Measurement Yearbook.  First published in 1938, the Yearbook was the work of Oscar K. Burros, who edited it until his death.  Beginning as a modest compilation of reviews of intelligence, aptitude, interest, and personality tests, it became the major source of information about all the tests in psychological literature — until its last publication in 1978, which consisted of two volumes of roughly 1,000 pages each.  It was not truly a "yearbook," since it was published only every five years or so.  The Rorschach, and other projective tests, were first reviewed in the third volume in 1949.  There were two reviews, one favorable and one unfavorable.  The unfavorable one was by J. R. Wittenborn, who wrote (p.133):

What passes for research in this field is usually naively conceived, inadequately controlled, and only rarely subjected to the usual standards of experimental rigor with respect to the statistical tests and freedom from ambiguity.  Despite these limitations, the test flourishes, its admirers multiply, and its claims proliferate.

The favorable review was by Morris Kruguman (p. 132):

The Rorschach withstood the clinical test well throughout the years and has come out stronger for it; on the other hand, attempts at atomistic validation have been unsuccessful and will probably continue to be so.

Note that neither reviewer cited any studies that demonstrated validity.  The favorable one views these as "attempts at atomistic validation" — which the author derogates.  (But what could be more "atomistic" than a psychological diagnosis presented in a custody dispute?  Such a diagnosis is a qualitative characterization, and if attempts to validate such characterizations on the basis of the Rorschach have been unsuccessful, how can they be made on the basis of this test?)  The unfavorable reviewer's characterization of the research as shoddy left open the possibility that good research might show the Rorschach to have some validity.

But if it does have validity, it is not reported in the next Yearbook. The favorable review by Helen Sargent asserts instead (p.218) that "the Rorschach test is a clinical technique, not a psychometric method."

By the time the fifth Yearbook was published in 1959, the world's leading expert on psychological testing, Lee Cronbach, is quoted in a review: "The test has repeatedly failed as a prediction of practical criteria.  There is nothing in the literature to encourage reliance on Rorschach interpretations."⁴  In addition, major reviewer Raymond J. McCall writes (p.154): "Though tens of thousands of Rorschach tests have been administered by hundreds of trained professionals since that time (of a previous review), and while many relationships to personality dynamics and behavior have been hypothesized, the vast majority of these relationships have never been validated empirically, despite the appearance of more than 2,000 publications about the test."  (Italics are in the original.)  The other major reviewer, Hans J. Eysenck, was even more negative.  After presenting the Cronbach quote, he reiterated again that there is absolutely no evidence for any of the claims of the people using the Rorschach test.

In the sixth Mental Measurement Yearbook published in 1965, Arthur R. Jensen wrote (p.509):

Many psychologists who have looked into the matter are agreed that the 40 years of massive effort which have been lavished on the Rorschach technique have proved unfruitful, at least so far as the development of a useful psychological test is concerned.

And later,

The rate of scientific progress in clinical psychology might well be measured by the speed and thoroughness with which it gets over the Rorschach.

In the seventh Yearbook, John F. Knudsen, a professor of clinical psychology and a practicing clinician, wrote (p. 440): "The Rorschach has continued to be characterized by numerous scoring systems and an overwhelming helming amount of negative research."

Finally, in the eighth Yearbook (1978), Richard H. Davis (p. 1045) concluded: "The general lack of predicted validity for the Rorschach raises serious questions about its continued use in clinical practice."

Are there not other reviewers in these same volumes who support the test?  Yes, but none of them refer to any research results.  Instead, they justify use of the Rorschach on the basis that it is a "very novel interview," a "behavior sample," or "source" or that it is a type of "structured interview" with which many clinical psychologists have become comfortable.  These claims somewhat vaguely reference "experience"; in addition there are a few suggestions that appear once and then disappear for using the Rorschach in a novel manner — such as having each member of a distressed couple take the test separately and then requiring the couple to reach a joint conclusion about what the blots look like.  The problem, of course, is that there is no evidence that this particular form of "structured interview" is more effective than any other, and, as pointed out elsewhere, only interviews structured to elicit certain specific information are valid.  Even if it were demonstrated that this type of interviewing does provide valid information, there is still the question of whether any of it has incremental validity — that is, whether it provides any information that cannot be obtained from simpler and more reliable sources, such as the history of past behavior.  Why then does the Rorschach continue to be used?

The answer may be found in the review of A. G. Bernstein in the seventh Mental Measurement Yearbook in 1972.  He wrote (p.434): "the view that recognition, the act of construing an unfamiliar stimulus, taps central components of personality functions is one that will remain crucial in any psychology committed to the understanding of human experience."  Despite his misuse of the term recognition (which means noting that a stimulus has appeared before in one's experience — the exact opposite of "construing an unfamiliar stimulus"), I agree with Bernstein.  He refers to a view, a plausible assumption.  If we adopt this assumption, the Rorschach should work.  The overwhelming evidence that it does not work is ignored.  Perhaps some other test works, but this particular one fails.  (One really interesting question is why the same ten blots have been used for over fifty years, given the failure of the technique and the simultaneous belief in the underlying "theory."  The best hypothesis I can provide is that of "institutional inertia.")  What I believe is "crucial in any psychology," however, is not a belief in the validity of the Rorschach, but an understanding of why people believe it.  The fascinating question is, who's projecting what and why?⁵  In contrast, a popular new scoring system ("the Exner system") has empirical validity.  The major variables in this system that correlate with behavior, however, are based on assessing the quality ("form level") of the responses.  Such assessment is based on the assumption that parts of the blots do resemble some shapes more than others, an assumption totally counter to the "projective" one that it is the lack of structure in the blots that leads to valid interpretations of subjects' responses.
  

Footnotes

¹  Now that I am no longer a member of the American Psychological Association Ethics Committee, I can express my personal opinion that the use of Rorschach interpretations in establishing an individual's legal status and child custody is the single most unethical practice of my colleagues.  It is done, widely.  Losing legal rights as a result of responding to what is presented as a "test of imagination ," often in a context of "helping," violates what I believe to be a basic ethical principle in this society — that people are judged on the basis of what they do, not on the basis of what they feel, think, or might have a propensity to do.  And being judged on an invalid assessment of such thoughts, feelings, and propensities amounts to losing one's civil rights on an essentially random basis.  [Back]

²  A very eminent psychologist once proposed that "intelligence is whatever it is that is measured by intelligence tests."  [Back]

³  McClelland, D. C. (1962). On the psychodynamics of creative physical scientists. In Gruber, H.; Terrell, G.; and Wertheimer, M. (Eds.). Contemporary Approaches to Creative Thinking. New York: Mather, 141-174.  [Back]

⁴  Cronbach, L. J. (1958). Assessment of individual differences, Annual Review of Psychology, 7, Stanford, CA.: Annual Reviews, Inc., 448.  [Back]

⁵  The materials in this section have been taken from the third through the eighth Mental Measurement Yearbooks, all edited by Oscar K. Burros.  The third was published in New Brunswick, New Jersey, by the Rutgers University Press; the fourth through the eighth were published in Highland Park, New Jersey, by the Gryphon Press; the years were 1953, 1959, 1965, 1972, and 1978.  [Back]