The halo effect, and how one trait colours the rest

An everyday example

A colleague who is warm and easy to talk to is also assumed, with no particular evidence, to be competent, honest and a good judge of other people. None of those extra qualities has actually been observed; they arrive as a kind of package deal with the warmth, because one strong overall impression tends to bleed into every other rating a person makes of someone, whether or not that other rating has anything to do with the first.

The reverse happens just as easily: one off-putting quality, an odd manner or an awkward first meeting, can drag down judgements of someone's competence or honesty that have no logical connection to it at all.

The classic studies

Edward Thorndike's 1920 paper is built from ratings that commanding officers gave of soldiers under them, on a set of distinct qualities: physical qualities, intelligence, leadership, and personal character, among others. If raters were judging each quality on its own evidence, the ratings across genuinely separate qualities should not move together very much: a soldier's build tells a rater little about his honesty. Thorndike found the opposite. Ratings across these unrelated traits rose and fell together far more than independent judgements should, as though officers had formed one general impression of a soldier, good or poor, and let that single impression colour every specific rating that followed. Thorndike called this a constant error, and the name that stuck to it later, the halo effect, points at the same thing: one bright or dim halo spreading over every trait in view.

Richard Nisbett and Timothy Wilson's 1977 study gave the effect a sharper, and stranger, demonstration. College students watched one of two videotaped interviews with the same instructor, a man with a noticeable accent, who behaved warmly and pleasantly toward his students in one version and coldly and irritably in the other. Students who saw only the warm version rated the instructor's appearance, his mannerisms and even his accent as considerably more appealing than students who saw only the cold version, even though the man, his face, his gestures and the way he spoke were identical on both tapes. When Nisbett and Wilson then asked participants directly whether liking the instructor's warmth had affected how they rated his appearance, most participants said no, and a number insisted the causal arrow ran the other way, that they liked him because his appearance and mannerisms were appealing. The halo had done its work invisibly: people could not see it operating in themselves even when asked to look for it directly.

Does it replicate?

Replication grade: Strong: reproduced across many domains; the size varies with how much direct evidence a rater has

The basic pattern, that a global impression bleeds into specific, logically separate trait judgements, has been documented across a wide range of settings in the decades since Thorndike's original paper: performance appraisals, teacher evaluations, product and restaurant reviews, and judgements of physical attractiveness spreading into assumptions about competence or honesty. The general shape of the effect is one of the more robustly observed patterns in the study of social judgement.

How large the effect runs varies a good deal by setting: it tends to be stronger when the rater has little direct information about the specific trait being judged, and weaker when the rater has solid, first-hand evidence to go on for that trait. Nisbett and Wilson's own further point, that people are often unaware of and will deny the influence when asked, is itself part of a larger and separately studied literature on how limited people's insight into the real causes of their own judgements can be, which the original 1977 paper connects to explicitly.

One lit tile, tinting the rest

The diagram on this page is a schematic, not measured ratings. It draws a grid of trait tiles, with one tile lit, standing for the trait actually observed, and the surrounding tiles tinted more strongly the closer they sit to it, standing for traits that were never actually seen but were judged anyway, coloured by the one that was.

Try it: three scenarios

Three short situations built around one trait colouring a judgement of another, unrelated one.

How to catch it

Because the halo effect is invisible from the inside, the most reliable fixes change the situation rather than relying on noticing the bias as it happens.

• Rate specific traits before forming or stating an overall impression, not after.
• Where possible, judge one quality at a time, on its own evidence, without knowing how the person scored on unrelated qualities.
• Be suspicious of any rating you cannot point to specific evidence for; if you cannot say why you rated a trait as you did, it may be borrowed from a different trait.
• Ask what you would rate this specific quality if you had never met the person and only saw this one piece of evidence.
• When you like or dislike someone strongly, treat your ratings of their unrelated qualities as more likely to be inflated or deflated, and check them against something concrete.