What is the halo effect?
The halo effect is one good impression colouring judgments nobody checked. It was named from army ratings in 1920, caught in a marking system, turned on company studies, and handled with procedure.
Answer to the question
What is the halo effect?
The halo effect is the tendency for one good or bad impression of a person to colour judgments of qualities nobody has checked. Thorndike named it in 1920 from army ratings. Some correlated ratings reflect real links between qualities, so researchers separate true halo from illusory halo, and organisations reduce it by scoring dimensions independently.
Key takeaways
- Thorndike (1920) named the halo after army superiors' ratings of officers on physique, intelligence, leadership and character agreed more closely than separate qualities should.
- Kahneman (2011) defines it as liking or disliking everything about a person, including things unobserved, and reports that grading one question across all students, not one booklet at a time, ended the consistency he had enjoyed.
- Murphy and colleagues (1993) argued that true and illusory halo are confounded in most rating data, so correlated ratings alone do not prove an error.
- Rosenzweig (2007) applied the term to companies: positive features are attributed to firms because their results are already known. A University of Wisconsin Oshkosh news release on Niendorf and Beck (2008) reported the eleven Good to Great firms averaged 202nd in the 2006 Fortune 500.
A century ago Edward Thorndike read the ratings that army superiors had given their officers for physique, intelligence, leadership and character. An officer who scored well on one tended to score well on all four, more consistently than Thorndike thought four separate qualities would. He called the cause a constant error and placed it in the rater, not in the man. The halo effect, as it is now defined, is the tendency for one favourable impression of a person to colour judgments of qualities nobody has checked, and for one unfavourable impression to do the reverse. Researchers have since complicated the claim, tested it in laboratories, and turned it on company studies and hiring practice.
A Name for Agreement
In the data Thorndike examined, raters scored each officer on several qualities, and the scores for one quality tracked the scores for the others. He wrote of a constant error he called the halo, and the remedy he implied was procedural: judge each quality on its own evidence. His finding concerned how judges behave once they hold a single overall impression, and it said little about the officers themselves.
The same mechanism turns up wherever people judge people. The earlier piece [How Far the Halo Effect Spreads](https://thewhyweproject.com/articles/how-far-the-halo-effect-spreads) follows it into beauty filters and organic labels, [Do Good-Looking People Earn More](https://thewhyweproject.com/articles/do-good-looking-people-earn-more) follows it into pay, and [Do We Judge Products on Too Little](https://thewhyweproject.com/articles/do-we-judge-products-on-too-little) follows it into shopping. This piece asks what the halo effect is when treated as something to measure and manage.
Daniel Kahneman defines it in Thinking, Fast and Slow as the tendency to like or dislike everything about a person, including traits never observed. His illustration is Solomon Asch's classic pair of descriptions. Alan is intelligent, industrious, impulsive, critical, stubborn and envious. Ben is the same six words in reverse order. Alan is judged much more favourably, because the early adjectives change what the later ones mean: stubbornness reads as a price worth paying in a capable man and as a flaw in an envious one.
Kahneman adds a variation. When people were given the first three adjectives as one person and the last three as another, and then asked whether all six could describe a single person, most thought it impossible. Once an impression has formed, he writes, the halo effect raises the weight of first impressions, sometimes until later information is largely wasted. The book gives no percentage for the variation and does not name who ran it, so it is best read as Kahneman's illustration of the pattern, not as a measured effect.
Kahneman Changes His Marking
Kahneman also reports catching the halo effect in his own marking. He had graded student essay examinations booklet by booklet, so a student's first answer coloured his reading of the rest. He changed to grading one question across all students, with the scores hidden on the inside back page so that earlier marks could not steer later ones.
The result was less comfortable. His grades for the same student now varied over a considerable range, and his confidence fell. He read that as a sign the new method was better, since the consistency he had enjoyed earlier was spurious. The old steadiness had come from his own impression of the student, not from the answers.
His earlier steady marks came from his own impression of the student, so the new scatter was the more honest picture.
The halo effect is therefore hard to see from inside. A coherent set of judgments feels like an accurate one, and the agreement among the scores is the very thing that makes them look reliable. This is one reason the remedy has to be a change of procedure, one that forces the evidence for each question to arrive before the overall impression does. Willpower does not reach a distortion that feels like clarity.
Good Qualities Often Correlate
Not every correlation among ratings is an error, and the strongest objection to Thorndike's reading came in 1993. Kevin Murphy and colleagues reviewed the research on halo error in the Journal of Applied Psychology and argued that the standard definition, which had dominated for most of the century, rested on assumptions that were wrong or problematic. In real people, qualities are often truly related, so high agreement across ratings can reflect true halo as well as the illusory kind, and the two are confounded in most data. They recommended a moratorium on halo indices as outcome measures in applied research.
That objection weakens the oldest evidence most. Thorndike's strong officers may have been strong all round. Experiments that hold the person fixed are harder to dismiss, and they show the halo effect without that confound. Richard Nisbett and Timothy Wilson showed 118 University of Michigan students the same instructor on film, once behaving warmly and once coldly. Ratings of his appearance, mannerisms and accent followed the warmth, and the students denied that their liking had influenced them. When the target stays still and the judgment moves, the movement cannot be merit.
When the target stays still and the judgment moves, the movement cannot be merit.
The two findings fit together. A halo effect that tracks real qualities cannot be separated from a true correlation by looking at ratings alone, which is the point of the 1993 review. A halo effect that moves with an irrelevant manipulation, like the filmed instructor's warmth, can be separated, and it is the second kind that organisations try to design out.
Companies Get a Halo
Phil Rosenzweig carried the idea out of the laboratory. In a 2007 article in California Management Review, and in a book of the same year, he described the halo effect in business as the tendency to attribute many positive features to successful companies, with the opposite attributions when they falter.
In his account the trouble lies in the data. If executives and journalists describe a firm's culture and strategy after its results are known, the descriptions are partly inferred from the results, and a large study correlating culture with performance is partly correlating performance with itself. Rosenzweig concluded that this contaminates many large-scale studies of company performance and encourages the belief that companies can follow formulas. He argued instead that performance is relative, and that high performers separate themselves from rivals by making choices under uncertainty. The halo effect heads the list in his book, and its subtitle counts eight other business delusions beside it.
An independent test exists for one such study. Bruce Niendorf and Kristine Beck looked again at the eleven companies singled out in the 2001 book Good to Great. A University of Wisconsin Oshkosh news release on their 2008 paper reported that the eleven ranked 202nd on average in the 2006 Fortune 500, and their long-run stock returns did not differ significantly from those of the average company in the S&P 500. One of the eleven, Circuit City, had filed for bankruptcy. The eleven had been chosen as examples of a leap to lasting excellence.
Judging Dimensions Before Discussing
Noise, the 2021 book by Daniel Kahneman and his co-authors, turns the halo effect into a design problem. In its account of corporate performance reviews, the authors describe a system that asked for forty-six ratings on eleven dimensions per person, and say the halo effect makes nonsense of it, since an early positive rating pulls the later ones along. In hiring, the authors describe Google's rule that interviewers rate a candidate separately, before they communicate with one another, and a structure that breaks the decision into separate assessments with information collected on each independently.
Google's four assessments were general cognitive ability, leadership, cultural fit and role-related knowledge, and the final decision came only after the separate scores were in. The authors call the third principle delayed holistic judgment, which keeps intuition in but postpones it.
The same structure screens out cues that invite a halo. The authors point out what is left off Google's list: a candidate's good looks, smooth talk and exciting hobbies. Candidates and interviewers dislike structured interviews, the authors add, and the method still earns its place. They report that structured interviews predict job performance better than unstructured ones, with a stated chance of 65 to 69 per cent that the candidate rated higher will perform better, against 56 to 61 per cent for unstructured interviews. The structure keeps judgment in the process and stops one early impression from setting every later score.
Common questions
What is the halo effect?
It is the tendency for one good impression of a person, such as looks, warmth or past success, to colour judgments of qualities that were never observed, and the reverse also occurs when one bad impression drags everything down. Edward Thorndike named it in 1920 after army ratings of physique, intelligence, leadership and character agreed more closely than he thought separate qualities would. Whether agreement in a given set of ratings is halo or a real link between the qualities is hard to settle from the ratings alone.
Is the halo effect always a mistake?
Not necessarily. Murphy and colleagues (1993) argued that true halo and illusory halo are confounded in most ratings data, because real qualities are often related. Experiments that hold the person constant, such as Nisbett and Wilson's warm and cold instructor, show the illusory kind more cleanly, which is why the halo effect survives the critique.
How can the halo effect be reduced at work?
Noise recommends judging dimensions separately and independently before anyone discusses them, as in Google's interview rule that interviewers score a candidate before they communicate. The book also advises leaving cues such as good looks and smooth talk out of the assessments altogether.
Sources
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. Asch's Alan and Ben and the grading episode are as Kahneman reports them.
Thorndike, E. L. (1920). 'A constant error in psychological ratings.' Journal of Applied Psychology, 4(1), 25-29. doi:10.1037/h0071663.
Asch, S. E. (1946). 'Forming impressions of personality.' Journal of Abnormal and Social Psychology, 41(3), 258-290. doi:10.1037/h0055756. Cited through Kahneman's account, not re-read.
Murphy, K. R., Jako, R. A., & Anhalt, R. L. (1993). 'Nature and consequences of halo error: a critical analysis.' Journal of Applied Psychology, 78(2), 218-225. doi:10.1037/0021-9010.78.2.218.
Nisbett, R. E., & Wilson, T. D. (1977). 'The halo effect: evidence for unconscious alteration of judgments.' Journal of Personality and Social Psychology, 35(4), 250-256. doi:10.1037/0022-3514.35.4.250.
Rosenzweig, P. (2007). 'Misunderstanding the nature of company performance: the halo effect and other business delusions.' California Management Review, 49(4), 6-20. doi:10.2307/41166403.
Rosenzweig, P. (2007). The Halo Effect... and the Eight Other Business Delusions That Deceive Managers. Free Press.
Niendorf, B., & Beck, K. (2008). 'Good to great, or just good?' Academy of Management Perspectives, 22(4), 13-20. doi:10.5465/amp.2008.35590350.
Kahneman, D., Sibony, O., & Sunstein, C. R. (2021). Noise: A Flaw in Human Judgment. Little, Brown Spark.