Horns Effect

​

What is Horns Effect?

Horns Effect, on this page, is the 1977 finding that a cold interview made students rate the same man’s appearance, mannerisms, and accent as irritating, while the authors’ name for that spread is the halo effect.

The halo effect is generally defined as the influence of a global evaluation on evaluations of individual attributes of a person, but this definition is imprecise with respect to the strength and character of the influence.

Nisbett and Wilson print that sentence as the ordinary definition, and then say it is too loose. A weak reading would only fill in a trait the viewer never saw. A strong reading changes a judgment of something the viewer did see. Their cold tape is the strong reading in the negative direction. The man, the face, and the accent were on the monitor. The answers changed. The ratings of the face, the mannerisms, and the accent changed with the answers. A useful picture is one coat in two rooms. The coat does not change. In one room the voice is friendly. In the other room the same voice refuses questions. People then call the coat appealing or irritating. Negativity Bias is a different claim, about a negative item weighing more than a positive item in a list. This file has no such list. It has one man and two scripts.

​

Horns Effect in 3 Depths

• Beginner: Same man, colder answers, more irritating ratings of looks, manner, and accent. Liking means were 5.48 and 3.18.
• Practitioner: Keep two questions apart. Thirty-four students were asked whether liking had moved the attributes. Another 56 were asked whether the attributes had moved liking. Only the second question, in the cold group, left zero.
• Advanced: A silent tape, with 34 other students from the same pool, left appearance and mannerisms only trivially different, both t < 1. That check does not cover the accent. The paper never prints the word horn.

​

Origin

Horns Effect is not the title of the 1977 paper. The title is the halo effect. The authors say the halo is one of the oldest named effects in the field, and that surprisingly little was known about how strong it is. They sort the older evidence into two piles, and they say neither pile proves the strong reading. Correlations among trait ratings can be positive because the traits really do travel together, or because two raters use different pictures of what “likable” and “good-looking” mean. Adjective studies, including Asch (1946), show that a word changes meaning inside a list. Photograph studies, they say, only show that people expect an attractive person to have an attractive personality, because the photo does not let anyone judge the personality on its own. Those papers were not opened. No cell from them is printed here. The design they wanted is narrower. Show people enough of an attribute to judge it, change the global evaluation, and see whether the attribute rating moves. They also wanted to know whether people can report the direction of that move. Confirmation Bias is a habit of seeking material that fits a belief already in view. This experiment does not ask the student to choose a file. It asks for a rating after a tape the experimenter chose. Nisbett and Wilson say Thorndike (1920) gave the halo its name and treated it as an inability to keep a global feeling off a specific rating. That article was not opened. No correlation from it is on this page. The 1977 PDF does not say that Thorndike used the word horn.

​

Key Points

Four splits keep a liking mean, an attribute rating, a backward report, and an unchecked accent from becoming one verdict.

​

Applications

These four moves follow splits the 1977 paper prints. They are not a claim that a class, an office, a household, or another year copies 5.48, 3.18, or any of these t values.

​

Case Study

The counted case is the videotape experiment, not a semester of teaching and not a trial. The subjects were 118 University of Michigan students in introductory psychology, 62 men and 56 women. The two sexes were entirely similar, so the paper combines them. They came in groups of 6 to 17, seated so that they would not talk about the tapes. The authors say the conclusion is unchanged if the group, rather than the person, is the unit. They do not print that group test. The room had chairs facing a 19-inch monitor, which they also give as .48 meter. The cover story said the department had been evaluating introductory-psychology teachers and wanted to know whether a short tape would resemble a whole term. Nobody in this file had that instructor for a term. Every student first saw a 7-minute filler interview with a bright, pleasant instructor. Half then saw the Belgian instructor in a warm guise, and half saw him in a cold guise. Each of those interviews lasted about 7 minutes. He was a native French-speaking Belgian. His English had a fairly pronounced accent. The paper does not print the filler’s ratings. In the warm guise he was likable, respectful of the students’ intelligence and motives, flexible, and enthusiastic. Asked about discussion, he said he encouraged it, that the give and take stimulated him, and that it raised interest in the material. Asked about evaluation, he described paper assignments students seemed to like and exams that mixed objective questions with questions that asked them to think. In the cold guise he was unlikable, cold, distrustful, rigid, and doctrinaire. He said he did not allow much discussion, because there is a time to be a student and a time to be a professor, and he would not be the professor if he did not know more than they did. He said he gave weekly multiple-choice quizzes, because otherwise you cannot trust them to do the reading.

In fact, the subjects who saw the cold instructor actually believed that the direction of influence was opposite to the true direction.

That sentence is in the abstract. The results section says the same thing with the tests above. Liking was the thing the script changed. Appearance, mannerisms, and accent were the ratings that moved. The 34 students mostly denied that liking had touched those ratings. The cold students among the 56 said the ratings had lowered their liking. After the forms, most students still denied the path the design had built. One young woman said that before he even began talking she had judged his appearance and then stuck with it. One student who had marked an effect of liking later said he had turned the question around: he disliked the accent a little, so that made him dislike the man more. Those are two conversations. They are not a share of the sample. The authors add a logical escape. If the face, the manner, and the accent were unpleasant on their own, the cold group could be right that those attributes lowered liking, and the warm group could be wrong to call them neutral. The silent follow-up is their answer for the face and the manner. It is not an answer for the accent, and it prints no means, only t < 1. The lesson inside this file is narrow. On these tapes, a cold script moved liking from a mean of 5.48 to a mean of 3.18 and moved ratings of attributes the students could see or hear. The students who saw the cold script described the arrow as pointing the other way. Naive Realism is the feeling that one’s own view is simply the facts. The paper does not score that feeling. It scores a denial of a path the experiment built.

​

Boundaries and Failure Modes

Horns Effect, as this PDF lets us count it, is one confederate, two scripts, and 118 introductory-psychology students at one university, plus a silent follow-up of 34 from the same pool. Two uses overshoot the paper. One exports 5.48, 3.18, or a t value onto a class, an office, a household, a shop, or another year. The other treats the cold group’s backward report as if the silent tape, the open accent alternative, and the warm means had not been printed. The name on this page is a later label for the cold direction. The experiment also moved ratings upward after the warm script. A reader who quotes only the cold majority, and calls that a separate mechanism, has dropped half the contrast the t tests use. The Halo Effect page is the neighbor for the positive direction. This PDF uses one name for both. The silent check is easy to over-read. “Trivially higher” and t < 1 are not a printed difference of zero, and they are not a photograph of two identical frames. No mean from those 34 students is in the article. Accent was never on that silent form. The authors had already said the cold role might have made the accent harsher. A claim that every attribute was held perfectly still is stronger than the follow-up. The awareness items do not cover a sentence the paper never gives. It describes 34 students on one question and another 56 on the reverse question. It does not say that each of the 118 received one of those two forms. The six non-significant tests are the “did liking change the attributes” question. They are not a second copy of the cold group’s t = 3.28, 2.80, and 2.74. The discussion recommends suspicion toward people who insist they can judge a manuscript, a conflict, or a relative independently. That is advice attached to these ratings. It is not a count of reviews, cases, or families. Implicit Bias is a different page about a speeded pairing of names and words. Nothing in this file is a keypress. Landy and Sigall (1974), which Nisbett and Wilson describe as one attribute changing a grade on an essay, was not opened. No essay mean is printed here. Stereotype Effect applies a belief about a category to a member. These students rated one man they had just watched, not a category they had not seen.

​

Common Misconceptions

Three readings turn a cold script, a backward answer, or a later nickname into a different study.

​

Related Concepts

These six pages sit beside Horns Effect. Each one asks a different question.

​

One-Line Takeaway