(I wrote this essay with Andrew Smith and Yuanbo Wang.)

JD & Cox

Oh, my God! I care so little, I almost passed out!

~ Perry Cox, MD

This year, we are turning our Big Five guns on the characters of Scrubs, the popular comedic hospital show. After starting with the European cult classic Dinner For One, we took a look at Friends and followed up with How I Met Your Mother. Whereas Dinner sported six octogenarians—four of whom dead—Friends and Mother opened a window into the U.S. American flirtation with prolonged, if not eternal, youth. Year after year, the same characters hover at the threshold of adulthood, a pose that becomes more precarious with each passing year. Mother is still hovering, which presumably will end when the mother of the poor kids who have been listening for a decade will finally come into motherhood.



operates on the same juvenile premise but takes itself less seriously. Scrubs is a parody of the genre, and it distinguishes itself by including a few fogies with non-trivial airtime. These characters, one might speculate, throw the young ones into sharper relief. The cast is familiar. There is JD, the (anti-)hero, who has to settle for initials. Will he get to use his full name when he passes into adulthood? Then there is Turk, whose name is ethnic and misleading. Turk and JD are inseparable, modeling male bonding in a way that challenges a host of encrusted notions of gender and race. Elliot’s name evokes notions of masculinity, in an intriguing contrast to her ditzy-blonde presentation. Carla, the multi-ethnic, diverse dominatrix is also stereotyped, as are the rest of them. Interestingly, she is the only character with an ordinary—even boring—name. In this coterie of four, Carla can actually pass as a grown-up. The


middle-aged Dr. Cox, with a name surely not randomly drawn from the big book of names, presents himself as a testosterone-poisoned he-man whose m.o. is to simultaneously lecture and humiliate the young crew while claiming to be ignoring them. Finally, Dr. Kelso is the (comparatively) ancient chief of medicine whose distinction is that he undercuts his status with random acts of buffoonery.

These characters are caricatures. Why should one bother to rate them on a set of personality traits, as if one were rating real people or characters meant to portray real people or characters than might as well be real? Perhaps brief narrative sketches of the type we attempted here are sufficient as a description.


There are two answers to this question. First, one idea popular among Big-Fivers is that as a parsimonious yet exhaustive taxonomy of dimensions of individual differences, the Big Five offer a comprehensive, though not detailed, view of the person. A person is summarized by five numerical scores. If we limit ourselves to three levels (high, middling, low) on each trait dimension, there are 3^5 or 243 different profiles, and with the Big 5 being independent of one another, as they are supposed to be, this number is not to be sneezed at. Evidently, use of this limited framework allows us to describe many distinctive types. In the Big-5 world, it is important that every rater consider each trait for every target person. The rater cannot, as with the narrative-sketch method, focus on one or two salient characteristics and ignore everything else. Using a common framework for all targets allows us to compare them with one another.


The second answer is that there are no rules or conventions that tell us what ought not be rated. When personologists started using the Big 5 to describe the personalities of dogs they opened a door. If dogs can be rated, so can cats, crocodiles, or cacti. In the old days, personality psychologists in the tradition of Gordon Allport regarded personality as the way in which a complex nervous system expresses itself. Once we allow that any mammalian nervous system is complex enough, no one can stop us from accepting any nervous system—or beings without nervous systems, or artifacts—as vessels of personality. We can describe a painting or a snow shovel using the Big 5 on the grounds that we cannot make a distinction between a real personality and a metaphorical one.


How is this possible? Many influential Big-5 psychologists no longer make assumptions about what it is within the person that gives rise to patterns of thinking, feeling, and behaving that we might call personality. Some still hold on to the notion that personality scores (e.g., Big 5 scores) ought to predict relevant behavior. But others have withdrawn even further to the idea that all that matters is agreement among observers about what a target person is like. Agreement is a matter of reliability, not validity, but when the quest for validity is over, and when personality is defined in terms of observer agreement, measurement becomes self-sufficient, that is, it no longer needs support from reality. 

When personality is bestowed by observer agreement, there is no distinction between reality-based personality and metaphorical personality. Personality will emerge from the ratings as long as raters are


willing to make ratings and as long as their responses have variability without being random. If personality were anchored in reality, we would know when there is no such thing. The measurement scale would have a true zero point. A two-dimensional plane has height and width but no depth. What does it take to have no ratable personality?

28 undergraduate students enrolled in the laboratory course watched season 1, episode 16 ("My Heavy Meddle") and then rated JD, Turk, Elliot, Carla, Dr. Cox, Dr. Kelso, and themselves on the five adjectives


used in previous exercises. “Is open to new experiences and complex” represented dimension O, openness. “Is dependable and self-disciplined” represented dimension C, conscientiousness. “Is extraverted and enthusiastic” represented dimension E, extroversion. “Is critical and quarrelsome” represented the low end of dimension A, agreeableness. “Is anxious and easily upset” represented dimension N, neuroticism. We standardized the ratings so that the grand mean rating of each trait was 0 with a standard deviation of 1.

The first set of six panels (above) shows the profiles of mean standard scores over the Big 5. For each target person, the variation over the five means indicates to which extent this person is seen as differentiated from the others. By this measure (i.e., the standard


deviation, SD, over the five means), JD (SD = .33) and Elliot (.43) come across as ordinary, whereas Turk (.85) and Dr. Cox (.72) show the greatest trait-to-trait variation. Dr. Kelso (.61), and Carla (.57) fall in between. The most outstanding person-trait pairs are Cox’s disagreeableness and Kelso’s introversion and lack of openness. This result may seem surprising on the assumption that older individuals are there to give context to the younger ones, to allow them to be individuated. Evidently, however, the elders steal the show.

Alternatively, we can examine the correlations among the target characters (excluding the self) over the five traits. The more positive these correlations are, the more similar two targets are to each other. The median of the 15 pairwise correlations is -.41, indicating


that the target characters are quite well differentiated from one another. Turk (-.66) is the least similar to the rest, and Dr. Cox (surprise) is the only one with a positive median correlation (.25).

The second set of five panels shows the average standard scores for the six targets and the self, separately for each trait. This way of displaying the results satisfies the conventional interest in individual differences, one trait at a time. We notice that the variation of the means over targets is about the same for each trait, but this is an artifact of the method of standardization.


Finally, we consider the correlations among the traits over targets. Here, we are reminded that with a small sample of targets and the small sample of respondents, we cannot expect to find all traits being unrelated to one another. In this sample, Extroversion, Agreeableness, and Openness form a tight cluster. Although this result may simply signal the poverty of the method, the data might be trying to tell us that the personalities of the Scrubs team are less complex than the personalities of us real people. The Scrubbers are, after all, caricatures.


The reduction of validity to reliability (noted above) has an interesting consequence. It makes the error or personal validation salonfähig (presentable in polite society). Personal validation occurs when the recipient of a piece of evidence judges its validity by the fit with her own preconceptions. If it seems valid, it must be valid. In our present exercise, we nod at the graphs inasmuch the depicted personality profile matches the image we ourselves have of the target character. To see that Dr. Cox’s scores lie on the undesirable side of the line for each trait may seem fitting to you. That’s what you thought anyhow. Therefore, the data must be valid. But it has to be so for most observers because the raters who contributed the numbers are people much like the rest of us. Therefore, it is likely that we agree with them. We could have been part of the sample ourselves. When we perceive our agreement as a signal of validity, that is, as an indication that the scores would be correct if there were true reality criteria (which told us what Dr. Cox is “really” like), we go beyond the data given. This is fine, except that we often don’t realize that it’s happening.

You are reading

One Among Many

Didgeridoo Dilemma

When the sale failed what kind of dilemma was it?

Friends as Allies

The demystification of friendship in the time of game theory .

Divide and Conquer

In math and in psychology, ratios are tricky.