David A. Kenny
April
25, 2000
Overview of the Study
What We Found
What 20/20 Did Not Say
Who Helped Me?
Go to a List of Related Sites
My research team and I conducted the following study in consultation with 20/20 staff. The show was aired on November 6, 1998. The study consisted of three phases.
Recruitment of Targets
A week before classes, we approached University of Connecticut undergraduate students to participate in a study of first impressions. Targets had to agree to be rated by strangers, have those strangers discuss on national television the strong and weak points of their personality, and then have their friends and family members rate and discuss them, again on national television. Understandably, not many people were willing to participate. We also took special efforts to make sure that the targets wanted to participate and understood the risks as well as the benefits. (This was not made clear in the show.) A select group of targets were selected and we also asked them to rate their own personalities.
Ratings by Strangers
The six targets were planted in my Social Psychology class which was meeting for the first time on morning of September 3, 1998. Each target was rated by about eight members of my class on four sets of personality traits: conscientiousness, extroversion, agreeableness, and culture or sophistication. These rating are called "zero-acquaintance ratings." That is, the strangers had not talked to the targets nor had they really interacted with them. After these ratings were completed, the targets left the classroom, and the strangers were interviewed by the ABC correspondent Michael Guillen. He asked the strangers to state their impressions of the targets that they had just rated.
Ratings by Friends and Family (Informants)
Later that afternoon, we assembled groups of about four who consisted of at least one parent, other family members, and friends of the targets. Very often the friends were roommates or boyfriends or girlfriends of the targets. The informants evaluated the target on a rating form and were interviewed on camera by Michael Guillen.
We computed the correlation between how the set of strangers viewed the target with how the informants viewed the target. The average correlation was .39. This somewhat corresponds to about a 70% rate of accuracy. So 7 of 10 ratings of the strangers were correct (assuming that informants really knew the targets' personality) where 5 of 10 would occur by chance. We obtained about the same rate of accuracy when we correlated stranger ratings with self-ratings. With such a small sample we were lucky to have gotten any accuracy.
Nonetheless, this result on a small scale replicates larger studies that show that zero-acquaintance ratings correspond to self ratings, ratings by knowledgeable informants (e.g., best friends), and behavioral observations. These studies have been here at Connecticut and by research groups elsewhere. As Nalini Ambady of Harvard University has observed, "thin slices" of behavior have surprising validity.
The show did not make clear a number of points. Of course, ABC is not a scientific journal and I did not expect it to make all of these points. But you might want to know the following.
First, research on this topic has been repeatedly conducted at Connecticut and at other universities. I regularly conduct a similar study in my social psychology class. Thus, the basis for the conclusion is extensive.
Second, we do not find agreement or accuracy in the judgments of others on all personality traits. We usually do find it for Extroversion and often for Conscientiousness (to learn about the Big Five), but we do not usually find agreement or accuracy for Agreeableness, Emotional Stability, or Culture. However, others have found both agreement and accuracy for these traits.
Third, it is a technical point but strangers and friends or family might agree but both might be wrong. For instance, in judging a male, strangers and friends or family may see him as aggressive, but they may be using the male stereotype which may be false.
Fourth, accuracy occurs in the average rating and is much lower in the individual's ratings. By and large, only when the person agrees with others is he or she right. So if your "gut" tells you one thing, but everyone else tells you something different, listen to them and not your "gut."
Fifth, not surprisingly there was no mention of the Social Relations Model. This analytic method has lead to real sharpening of the questions that can be asked. Thus, the advances came about by important statistical advances.
Sixth, accuracy at zero acquaintance is interesting, but we were interested in comparing accuracy in close relationships. We do find more accuracy there but not much more. Thus, we learn more about the person (at least in terms of personality) in the first few minutes, than we do in years. This is the most surprising results.
I am totally indebted to my research team, most of whom are students at the University of Connecticut. First, Cynthia Mohr helped in all phases of the research. She was also in charge of the project while I vacationed in California in late August. Second, Lynn Winquist provided key advice and assistance. She was especially helpful in the design of the informant phase of the project. Third, Jason Themanson conducted a videotape study and was also in charge of the data analysis phase of the study. Fourth, Jill Santopietro assisted in the recruitment of the targets and many other tasks. Fifth, Carl Schneider, Tony Lemieux, Christie Cathey, Madeline Fugere, Michael Whitcomb, Steven Rumery, and Jo Korchmaros helped in many phases of the study. Finally, Virginia Carrow and members of the Psychology Department (under the able leadership of Charles Lowe) provided extensive support.
I want to thank especially Linda Albright and Tom Malloy. They were my coauthors in the original zero acquaintance study (so with me they thought up the idea, they provided key advice about the study, they helped in the data collection, and finally they hosted the victory party. I have also published in this area with Debby Kashy, Caryl Horner, Jenny Chu, and Maurice Levesque.
I owe an unpayable debt to the research participants. Deserving special mention are the targets who allowed themselves to judged by a national television audience. They are the stars of the show. I also want to thank my Social Psychology class for letting us use the first day of class for the study.
Thanks to everyone! I apologize for not thanking the many others who helped.
My Recent Book on Interpersonal Perception
Read about the Social Relations Model (the statistical model)