The Projective Hypothesis • The projective hypothesis (Lawrence Frank, 1939): When people try to understand vague or ambiguous unstructured stimuli, the interpretation they produce reflects their needs, feelings, experience, prior conditioning, thought processes • Shakespeare, Hamlet, II.ii: "Nothing is either good or bad, but thinking makes it so."
The Problem • The difficulty is answering: Which 'needs, feelings, experience, prior conditioning, thought processes' are reflected? • Projective tests can (by their own claim) draw equally upon the imagined and real, the conscious and unconscious, the recent and old, the important and the trivial, the revealing and the obvious • Much room for interpretation is left given to the tester, making test validation almost impossible • These tests flourished more in the psychoanalytical era, 1940-1960
The Rorschach Inkblot Test • The Rorschach Inkblot Test is the most commonly used projective test • In a 1971 survey of test usage, it was used in 91% of 251 clinical settings survey • It is one of the most widely used tests that exists: According to recent surveys by the American Psychological Association, 82 percent of its members ''occasionally'' and 43 percent ''frequently'' use the test
The Rorschach Inkblot Test • It is widely cited in research, third only to the MMPI and the NEO Personality Inventory (a five-factor personality measure) • It is estimated to be administered to 6 million people per year • “No general discussion of psychological tests is complete without reference the Rorschach, despite its scientific inadequacies.” Kaplan & Saccuzzo / Psychological Testing
History • The earliest use of inkblots as projective surfaces was J. Kerner's (1857) • He was the first to claim that some people make idiosyncratic or revealing interpretations • In 1896, Alfred Binet suggested that inkblots might be used to assess personality (not psychopathology) • Some work was done on this suggestion • The first response set was published by G. M. Whipple (1910)
History • Herman Rorschach, a Swiss psychiatrist, was the first to suggest (1911) the use of inkblot responses as a diagnostic instrument • In 1921 he published his book on the test, Psychodiagnostik (and soon thereafter died, age 38)
History • Rorschach's test was not well-received, attracting little notice in Switzerland or elsewhere • David Levy brought it to the States • His student, Samuel Beck, popularized its use here, writing several papers and books on it starting with Configurational Tendencies in Rorschach Responses (1933) • Several other early users also published work on he Rorschach • Several offered their own system of administration, scoring, and interpretation, each finding his or her own followers, leading to problems in standardization that continue to this day
What is the Rorschach? • The stimuli were generated by dropping ink onto a card and folding it • They are not, however, random: the ten cards in the current test were hand-selected out of thousands that Rorschach generated • He selected 5 black/white, 2 red/gray (II & III) and 3 color (VIII – X)
Administering the Rorschach • The test is usually administered with as little instruction and information as possible • The tester asks 'What might this be?' and gives no clues or restrictions on what is expected as a response • Anxious subjects often do ask questions, and vague answers are offered • Some advocate sitting beside the subject to avoid giving clues by facial expression • If only one response is given, some hint to find more may be offered: "Some people see more than one thing." • The orientation of the card and subject RT is recorded
Administering the Rorschach • The cards are shown twice • The first time responses are obtained (the inquiry); the second time they are elaborated (the elaboration) • The test administrator asks about: • i.) Location: Where did the subject see each item? • A location chart is used to mark location • W = whole; D = Common detail; Dd = Unusual detail; DW = Confabulatory response • ii.) Determinant: What determined the response? • Form (F)? • Perceived movement? Human (M); Animal (FM); Inanimate (m) • Color (C); shading (T = texture)
Administering the Rorschach • The test administrator asks about: • iii.) Form quality: How well-matched is the response to the blot? • F+ = good match; F = match; F- = poor match • iv.) Content: What was seen? • Human (H); animal (A); nature (N)? • The test administrator also scores popularity/originality: How frequently is the percept seen? • Norm books are available (i.e. Exner, 1974) [but are not always well-received in clinical settings]
Scoring the Rorschach • Some quantitative information is obtained: i.e. percent of W, D, Dd, and DW responses • Deviation from norms can mean an invalid protocol, or brain damage, or emotional problems, or a low mental age (or just an original person) • These quantitative measures can be validated • i.e. # of W responses has been linked to general intelligence (r = 0.4); Movement responses are said to suggest strong impulses or high motor activity; DW (confabulatory) responses are taken as signs of a disordered state; low response rate is associated with mental retardation, depression, and defensiveness • Alas, many attempts to validate signs are unclear • Often there is fail to replicate, or the findings contradict expert claims
Scoring the Rorschach • Most scoring is qualitative: i.e. analyzing content • There are no hard and fast rules • All but the most ardent proponents suggest that the protocol be analyzed in the context of other tests results and clinical information
Exner’s Comprehensive System • Exner (1974, 1995, 2003): scores location, developmental quality, determinants: form (quality), movement, color (chromatic/achromatic), texture, dimensionality, diffuse shading, popular responses + more • He normed his system on five groups: nonpatient, outpatient nonpsychotic, inpatient character problem, inpatient depressive, inpatient schizophrenic • His original norms were flawed: 221/700 were entered twice (this was fixed in the 1995 revision). • It is claimed they over-diagnose pathology
Exner’s Comprehensive System • “In the first edition of this book, published in 1982, we predicted that the 21st century would see the Rorschach elevated to a position of scientific respectability because of Exner’s Comprehensive System. Over the years we backed away from this position. Now, more than 20 years later, we must acknowledge that we were mistaken.” Kaplan & Saccuzzo / Psychological Testing, 6th Ed.
Psychometric Properties of the Rorschach • Obviously, it is almost impossible to measure any of the usual psychometric properties in the usual way • Validity and reliability are both rendered meaningless by the open-ended multiplicity of possibility that is allowed and by the lack of universally-accepted standardized instructions, administration protocol, and scoring procedure (but see Exner, 1974….if allowed)
Validity & reliability of the Rorschach • One approach to validity: Assess blind diagnosis made from a protocol alone • In one study, 85% of protocols were matched to case descriptions, in batches of 5 • Why might this be dubious? • Studies have suggested that changing instructions can change what you get (Blais et al., 1995)- not good for validity! • Average correlation of Rorschach with validity measures hovers low, around 0.30 (Garb, Florio, and Grove, 1998; Hiller et al., 1999) • Reliability studies that have been done find r-values varying from 0.1 to 0.9 • One was done on cases after electroshock, because it "wipes out memory for the first test but does not change personality" • Protocols were reported to be very similar
Validity & reliability of the Rorschach • The test does not consistently measure depression, anxiety disorders, psychopathy, impulsiveness, criminality, or sexual abuse • In a 1999 study of random people (at a California blood bank) one in six came up with Rorschach protocols supposedly indicative of schizophrenia: a hugely improbable finding! • Other evidence suggests that there may be cultural biases in the Rorschach: African-Americans, Native Americans, American Hispanics, and Central and South Americans all produce Rorschach scores that differ from the Exner norms
Validity & reliability of the Rorschach • In forensic settings, Rorschach interpretations can still have huge consequences, changing prison or even death sentences, or child custody decisions • Studies that support the validity of the Rorschach have been denounced for being methodologically flawed and inconsistent, and are almost always unreplicated • It is alleged by some that rare ‘master interpreters’ exist; that interpretation is an art that cannot be systematized
Projective tests as non-psychometric tools • “The test can give you objective data, like a blood work. But we all know that some physicians, for whatever reason, are skillful diagnosticians beyond the tests, in seeing the way the patient looks or moves. The Rorschach is a stimulus that generates a lot of information. You may generate hypotheses that aren't in the hard data yet, but that doesn't mean this is the same thing as reading tea leaves.” • Irving Weiner, clinical professor of psychiatry and behavioral medicine at the University of South Florida
Projective tests as non-psychometric tools • “The test can give you objective data, like a blood work. But we all know that some physicians, for whatever reason, are skillful diagnosticians beyond the tests, in seeing the way the patient looks or moves. The Rorschach is a stimulus that generates a lot of information. You may generate hypotheses that aren't in the hard data yet, but that doesn't mean this is the same thing as reading tea leaves.” • Irving Weiner, clinical professor of psychiatry and behavioral medicine at the University of South Florida However, I have know self-proclaimed ‘master interpreters’ who failed to convince me of the mastery of which they were convinced
A few final points • A MC version of the Rorschach was tried but was useless • Note this is a very time-consuming and difficult test • A lot of effort is required for the return on data, both in terms of test-administration and of training (some say at least 1 -2 years is required just to learn, and much more to master, the Rorschach)
The Thematic Apperception Test (TAT): • Introduced in 1930s by Christina Morgan and Henry Murray • It has been called “a clinician’s delight and a statistician’s nightmare”. • 30 grayscale pictures + one blank for elicitation of stories • Not all are (though all may be) seen by everyone: some are suggested for men, some for women, some for youth, some for elderly • Most subjects see 10-12 cards, over two sessions • Based on Murray's (1938) theory of 28 social needs (sex, affiliation, dominance, achievement etc.) • Thema = Interaction between needs and environmental determinants
The Thematic Apperception Test • Scoring takes into account many things: • Roles, press (environmental forces), needs (forces from the hero), themes (interact with press/needs), outcomes • Congruence with picture stimuli • Conformity with directions • Conflict • Literal Story Content
The TAT: Validity and reliability • Standardization of administration and scoring is minimal (Only 3% of psychologists use standard scoring systems) • As with the Rorschach, studies suggest that directions make a difference: loaded words produced more distress in several studies. • Internal consistency is low • 2 month test-retest r = .80; 10 month test-retest r = .50 • Inter-rater reliability varies with studies between 0.3 to 0.9 • A meta-analysis by Spangler (1992) found average correlations between TAT and other criteria around 0.20
The Thematic Apperception Test • As with the Rorschach, studies suggest that directions make a difference: loaded words produced more distress in several studies. • Many variations on this 'story-telling' test exist
Rotter Incomplete Sentences Blank (RISB) • The Incomplete sentences test (Julian Rotter, 1950) is designed as a screening tool for one construct: over-all emotional adjustment • This is a well-standardized projective that requires subjects to complete 40 short sentence stems in a way that “expresses your real feelings”
Rotter Incomplete Sentences Blank (RISB) Example sentence stems: I like …….. My greatest fear ……. I am ……. Men ……. Dancing ……. Sports …….
Rotter Incomplete Sentences Blank • Three forms: high school, college, adult • 40 incomplete sentences/stems, usually only 1-2 words long • Takes 20-25 minutes • The use of directed stems allows one to probe the subject explicitly for locus of control (internal/external), interests, likes, hopes, fears etc. • Each item scored on 7-point scale where higher numbers indicate more severe maladjustment • The scoring system is well defined, and as a result…
Rotter Incomplete Sentences Blank • There is good inter-rater reliability (about 0.90) • Cronbach’s alpha = 0.69 • Test-retest reliability close to 0.80 after 1-2 weeks; 0.50 after months; 0.38 after 3 years • Validity studies have tended to support the idea that the RISB measures adjustment
Other common projective tests • House-Tree-Person Test (Buck, 1948) & Draw-A-Person (Machover, 1949): Subject is asked to draw • Scoring is based on many aspects: • Absolute & relative size of elements • Sequence of elements • Omissions and detailing/emphasis/erasures (especially of body parts) • Verbalizations while drawing • Size & placement of figures on the page • Amount of action depicted • Systematization in doing the task • Shading
"If there is a tendency to over-interpret projective test data without sufficient empirical grounds, then projective drawing tests are among the worst offenders." Kaplan & SaccuzoPsychological Testing, 1993
The problem with drawing tests - Among the plausible but empirically invalid relations that have been claimed: - Large size = Emotional expansiveness or acting out - Small size = emotional constriction; withdrawal, or timidity - Erasures around male buttocks; long eyelashes on males = homoeroticism - Overworked lines = tension, aggression - Distorted or omitted features = Conflicts related to that feature - Large or elaborate eyes = Paranoia
Chapman & Chapman- Test Results Are What You Think They Are • Main point: People tend to over-estimate the frequency of correlations they believe in (i.e. of associations) -or, equivalently, people tend to confuse correlation with semantic association • This confusion is very resistant to change • It remains even when the actual correlation is negative, or when there are cash rewards for accurate estimations of correlation
Chapman & Chapman- Test Results Are What You Think They Are • Holzberg & Wexler tested whether paranoid subjects draw elaborate or exaggerated eyes • 18 paranoid patients were compared to 76 student nurses • No effects were found • Chapman & Chapman conducted a forced-choice survey, which found that that 91% (!) of clinicians believe that a suspicious person would draw large or otherwise atypical eyes • 82% believed that a person worried about his intelligence would draw a large or otherwise emphasized head, although there is no evidence.
Illusory correlation • This is an example of an illusory correlation, in which subjects believe two things that are semantically associated must also be correlated • Chapman and Chapman showed subjects word pairs that were either semantically associated or not • They controlled the pairs so that all pairs occurred equally often (25% of the time) • However, subjects reported that the associated pairs came up on average 47% of the time.
Illusory correlation • The same results were found with Draw-A-Person results in naïve subjects: Subjects who did not know the test insisted that semantic associations were more common than chance, even when they were controlled in a set of drawings • A $20 reward for accuracy did not reduce the illusory correlation • Allowing for direct comparisons of drawings, and measuring, and scratch paper, and no time limit, plus the $20 reward for accuracy did reduce error, but not to nothing: instead of 76% reporting an (nonexistent) illusory correlation, 45% reported it.
Illusory correlation • On a similar study of the Rorschach, clinician subjects found (non-existent) correlations between homosexuality and semantically-associated objects: buttock, male genitals, female clothing • Naïve non-clinicians then reported non-existent correlations of these with patient descriptions of homosexuality and precisely these same signs • Even when a negative correlation was built in (homosexuality never appeared with these signs), subjects reported a positive correlation
Illusory correlation & life: Forer, 1949 • Forer (1949) gave each student a personality profile based on a questionnaire he had administered to each of them • Students rated their profile on a scale of 1 (inaccurate) to 5 (accurate) • The average rating was 4.2
Illusory correlation & life: Forer, 1949 • Gave each student a personality profile based on a questionnaire he had administered to each of them • Students rated their profile on a scale of 1 (inaccurate) to 5 (accurate) • The average rating was 4.2 • The catch: Everyone had exactly the same profile, compiled from a book of horoscopes = The Barnum Effect
Illusory correlation is an illusion! • We do not hold it against ourselves if we see an optical illusion: • Illusory correlation is a cognitive illusion: we should not moralize or fret about it, but take sensible precautions to guard against it "senses are fallible…clinical judgments must be checked continually against objective measures"
So what? • Illusory correlations are rampant in real life in ways that matter • We easily over-weight events that are (especially, negatively) emotionally-marked: criticism from our parents or partners; personal failures; errors; superstition correlations • It helps to realize that, just like the lines are straight but we see them straight, our self-views are systematically incorrect: we should all probably distrust any views that says we are extremely bad in any way • Cognitive-behavioral Psychotherapy can use this: box scores for real life
The hermeneutic circle • ‘Hermeneutics’ is the art of interpretation, especially texts (and the study of that interpretation) • Hermes was a Greek God of boundaries and therefore of messengers and poetry, and was himself a messenger of the gods to humans. • There are several kinds of ‘hermeneutic circles’ but one is that those who do the interpreting are themselves open to interpretation • What we read into anything is as much about the reader as it is about what is read • Critics of projective tests have argued that this hermeneutic circle applies directly to projective tests- those who interpret say as much about themselves in their act of interpreting as they do about the one they interpret
The hermeneutic circle • "Objectivity in human relationships is impossible. Therapists affect the behaviour and feelings of patients, and patients affect therapists. When a chart notes that a patient is 'hostile', it should also note, in the interests of balance, that the therapist is 'paranoid'. If a therapist calls a patient 'defensive', chances are that the patient would call the therapist 'aggressive'. Both should be noted in a chart, if either is, since both are equally probable. " • Shelagh Lynne Supeene • As For The Sky, Falling
Projective tests as non-psychometric tools • Projective tests are clearly not so great as psychometric tools: they violate almost everything psychometric principle we have learned in this class! • However, they may nevertheless serve an important and useful clinical purpose as interview aids • They allow people to ‘say what they cannot say’ • They open up issues that might otherwise never be raised. • They provide a broad overview of main foci of attention