A first impression is a verdict looking for evidence.
During the First World War, the US Army wanted a fair way to judge its officers. The psychologist Walter Dill Scott designed a rating form with four boxes: physical qualities, intelligence, leadership and character. Raters were told to judge each quality on its own and disregard the others.
Edward Thorndike of Teachers College, Columbia University, later studied the completed forms. One very conscientious flight commander had rated 137 aviation cadets. A cadet he rated as physically impressive tended also to be rated clever, a strong leader and a man of good character. Thorndike called this the halo effect.
He published the finding in 1920, and the name stuck.
The halo effect is the tendency to let one overall impression of a person, product or company colour our judgement of its separate qualities. If we like one thing about someone, we tend to rate their other traits higher too, even traits we have no evidence about. Edward Thorndike named it in 1920.
Where did the halo effect come from
Thorndike had seen the pattern earlier. In a 1915 study of employees at two large industrial companies, supervisors’ ratings of the same man for intelligence, industry, technical skill and reliability came out highly and evenly linked. The army forms showed the same thing.
He measured how closely the flight commander’s ratings moved together using a correlation, a number from 0 to 1 where 1 means two ratings rise and fall in perfect step. Intelligence correlated with physique at 0.51, with leadership at 0.58 and with character at 0.64. Thorndike thought the link between intelligence and physique should have been about a third as strong as the other two. In his words, the numbers were too high and too even.
The pattern was not just one rater’s. Eight officers who rated aviation officers for technical skill as a flyer, a specialised ability, produced correlations with general officer ability that averaged 0.67. Thorndike estimated the real figure could hardly be above 0.25. A general sense of merit had leaked into a judgement about flying.
Teachers fared no better. A colleague at Teachers College, Mr Knight, studied 129 teachers rated by their superior for salary and promotion decisions. The rating of each teacher’s voice correlated 0.63 with the rating of her intelligence. Nobody believes a pleasant voice makes a person clever. The raters had formed one general impression and spread it across every box on the form.
He was not the first to notice. In 1907, the psychologist F. L. Wells had found that people rating works of literature let their overall opinion bleed into every specific quality. Thorndike gave the pattern its name and its evidence.
His conclusion was sharp. Even a very capable foreman, employer or teacher, he wrote, could not treat a person as a set of separate qualities and judge each one independently. His fix was practical: observers should report the evidence rather than a score, and each quality should be rated separately without knowing the evidence about any other.
The popular version today leans heavily on looks. That strand came later, from studies such as Karen Dion, Ellen Berscheid and Elaine Walster’s 1972 paper “What is beautiful is good”. Thorndike’s original point was wider and more uncomfortable. His raters knew the people they were rating, often well, and still could not keep the boxes apart.
Halo effect examples at work and in business
The cleanest demonstration came in 1977 at the University of Michigan. Richard Nisbett and Timothy Wilson filmed two interviews with the same college instructor, a man who spoke English with a Belgian accent. In one, he was warm and friendly. In the other, he was cold and distant. The words about his teaching changed, but his face, his clothes and his accent did not.
In all, 118 students watched one video or the other and rated him. Those who saw the warm version found his appearance, mannerisms and accent appealing. Those who saw the cold version found the same appearance, mannerisms and accent irritating.
Then the researchers asked the students why. Most insisted that liking or disliking the man had not affected their ratings of his accent. Students who saw the cold instructor went further: they said his irritating accent was the reason they disliked him. They had the direction exactly backwards, and they could not see it.
The same effect works on companies. In 2008, Satyam Computer Services, then among India’s best-known IT firms, received the Golden Peacock Global Award for Excellence in Corporate Governance from the UK-based World Council for Corporate Governance. Months later, on 7 January 2009, its founder Ramalinga Raju wrote to his board confessing that the balance sheet carried Rs 5,040 crore of cash and bank balances that did not exist. The award was withdrawn.
Nobody can prove the award jury was dazzled. But governance is hard to see from outside, and growth, big clients and stock-market listings are easy to see. When a quality is invisible, a halo from the visible ones fills the gap.
You meet the halo effect in quieter forms every week. The candidate who speaks fluent, confident English is assumed to be better at the job itself. The colleague who delivered one big project gets the benefit of the doubt on everything that follows. The reverse also happens: one poor trait drags down the rest, sometimes called the horns effect.
Why does the mind work this way? Judging many separate qualities is slow and effortful. Forming one overall feeling is fast. Once the feeling exists, it shapes how we read everything else, and as Nisbett and Wilson showed, it does so without telling us.
Is the halo effect always a mistake
Not always, and the research has been honest about this.
In 1993, Kevin Murphy, Robert Jako and Rebecca Anhalt reviewed decades of work on halo in performance ratings. They argued that most of the standard assumptions about it were wrong or at least problematic. The core difficulty is that some qualities really do go together. A hardworking employee may genuinely also be reliable. A good teacher may genuinely keep better discipline. When ratings move together, you usually cannot tell how much reflects real overlap and how much reflects the rater’s general impression. Murphy and his colleagues also pointed to studies where high halo did not make ratings less accurate.
Nisbett and Wilson’s experiment avoids this problem because the instructor’s accent and looks were held constant. That is why it is persuasive. But it was one study, with one instructor, on one campus.
The fair reading is this. The halo effect is real, and it is strongest when a visible quality stands in for one we cannot see. It is less of a problem when the qualities being judged are genuinely linked, or when the rater has direct evidence for each one.
How to reduce the halo effect when you judge people
Thorndike’s 1920 remedy is still the best place to start. Judge each quality on its own evidence, and do not look at your other ratings while you do it.
In interviews, score one question across all candidates before moving to the next, rather than scoring each candidate on everything at once. That keeps a strong first answer from carrying the rest.
Write down the evidence before the score. “Delivered the Pune launch two weeks early” is evidence. “Strong performer” is a halo waiting to happen.
When a quality is hard to see, such as honesty or financial controls, ask what you would believe if you knew nothing else about the person or company. Then go and check that quality directly.
A halo tells you how you feel about someone. It does not tell you what they are like.
Daily Reflection
Think of one person whose work you rate highly. Pick one specific quality you have never actually checked, and ask what evidence you have for it.
