People avoid giving compliments because they expect an awkward exchange, but experiments found recipients felt better and less uncomfortable than the giver predicted — and the person offering the compliment also walked away in a better mood, meaning both sides misread a kindness before it happens

A compliment can be sincere and still feel difficult to say. The hesitation often arrives before the words: perhaps the other person will find the comment strange, perhaps the delivery will sound clumsy, or perhaps a small kindness will create a social obligation neither person wanted.

Erica Boothby and Vanessa Bohns tested that forecast across five studies. People expected recipients to feel less good and more uncomfortable than recipients reported feeling. In the final study, the people giving compliments also reported a better mood after the exchange than before it.

This is one research programme, not settled consensus. The final clause of the headline also needs a factual boundary. Recipients were not asked to predict the exchange before it happened, and givers were not asked to forecast their own later mood. The studies show givers misreading recipients, alongside positive reports from both parties, rather than two forecasting errors measured independently on each side.

That distinction does not weaken the central result. It tells us where the result actually sits.

The first field experiment used a deliberately plain compliment

In Study 1a of the Personality and Social Psychology Bulletin paper, participants left a laboratory and went to assigned locations on a university campus. To limit cherry-picking, they approached the fourth same-gender stranger they saw and said they liked the person’s shirt, jacket or dress.

Before leaving, givers predicted how much the recipient would enjoy the interaction and feel good, pleased and flattered. The stranger then completed a confidential sealed survey containing the same seven-point questions.

The researchers recruited 190 people, forming 95 giver-recipient pairs. Four pairs were excluded because at least one person did not complete the survey. Recipients’ average positivity score was 4.66. Givers had predicted 3.93, a statistically significant gap.

Givers did not expect disaster. They expected a moderately positive response. Their error was subtler: the exchange landed better than they anticipated.

A genuine compliment produced the same miscalibration

A scripted remark about clothing might sound artificial. Study 2 therefore let 50 givers compliment a stranger on anything they genuinely liked, producing a 100-person sample. Recipients averaged 5.08 on the positivity index, while givers predicted 4.18.

The givers estimated the recipient’s reaction again after the interaction. Their average rose only to 4.43, a change that was not statistically significant, and it remained below the recipient’s report. Even after seeing the response, they did not fully update.

Most remarks concerned visible features or choices: 42 per cent were about a top, 24 per cent an accessory, 20 per cent a physical feature such as hair, 8 per cent shoes and 6 per cent an outfit. The authors argue that seemingly light observations can still affirm taste, identity or a choice someone made.

The study materials and data were made available through the researchers’ OSF project. All four numbered studies were preregistered.

The discomfort forecast was further from reality

Study 3 added questions about annoyance, discomfort and feeling bothered. Again, the design involved 50 givers and 50 recipients. Recipients rated their negative reaction at 2.42 on the seven-point index. Givers had predicted 4.17.

The positive side replicated too: recipients averaged 4.87, compared with the giver forecast of 4.16.

After the exchange, givers revised their negative estimate down to 3.39. They recognised that the interaction had gone better than feared, but still believed the recipient had felt more discomfort than the recipient reported. That lingering error may reflect an ordinary problem in reading politeness. A smile can be interpreted as courtesy rather than evidence that a comment was genuinely welcome.

This resembles the forecasting gap in Silicon Canals’ earlier report on people leaving conversations believing they were liked less than they really were. In both cases, the person speaking has unusually good access to their own uncertainty and unusually poor access to the other person’s private reaction.

Anxiety distorted the view from inside the exchange

Study 4 added third-party forecasters who read the same scenario but did not have to approach anyone. The full design involved 50 givers, 50 recipients and 50 outside forecasters, although four giver-recipient pairs were excluded for incomplete surveys or instructions.

The outsiders were more accurate. Their average forecast for a recipient’s positive response was 5.26, close to the recipients’ 5.16 and above the givers’ 4.45. For negative reactions, outsiders predicted 2.33, recipients reported 1.96, and givers expected 3.88.

Givers also felt more anxious than outside observers. Anxiety predicted how negatively they expected the interaction to land. Their belief in their own competence mattered too: people who doubted their ability to deliver the compliment expected recipients to feel less positive and more uncomfortable.

The mechanism was not manipulated experimentally, so the correlations do not prove that anxiety caused the mistaken forecast. They do fit a clear perspective difference. The person speaking monitors wording, timing and visible nervousness. The person listening may respond mainly to the warmth behind the attempt.

I have noticed the same structure in other social forecasting research, including the studies behind my piece on why strangers often expect deeper conversations to be more awkward than they become. We judge the exchange from inside the performance. The other person experiences the approach.

Givers returned in a better mood, but the design matters

Before approaching a stranger in Study 4, compliment givers rated how good they felt at 4.13 on a seven-point scale. Afterward, the average was 5.26. Their stated willingness to compliment another stranger also rose from 3.00 to 4.59.

Those are substantial within-person changes, but this part of the study had no no-compliment control group. It shows that givers returned in a better mood. It cannot isolate the compliment as the only possible cause. Participants had also completed an assigned task, moved through an anxious moment and received an observable response.

The title’s claim that the giver walked away feeling better is therefore descriptive and accurate. A stronger claim that compliments necessarily caused the mood improvement would go beyond the design.

A separate nine-experiment programme by Xuan Zhao and Nicholas Epley reached a similar conclusion about recipients. Across compliments written for people participants knew, expressers underestimated recipients’ positive feelings and overestimated awkwardness. That work traced part of the gap to competence: writers worried about finding the right words, while recipients registered more warmth than writers expected.

The result has limits that matter in ordinary life

Boothby and Bohns studied brief, genuine compliments, largely between same-gender strangers in university settings. They explicitly warn that the pattern should not be extended to sexualised remarks, catcalling or inappropriate comments at work. In those cases, a giver may do the reverse and underestimate discomfort.

Hierarchy, relationship history, culture and context can change the meaning of praise. A manager commenting on an employee’s body is not equivalent to a stranger admiring a jacket. A compliment is not a demand for attention, and good intent does not cancel a recipient’s boundaries.

The samples were also young and located in the United States. The researchers note that stronger modesty norms, less anxiety-provoking settings, anonymous online interaction or established relationships could alter the forecast. Their field method improved realism, but recipients knew an experiment was involved when they completed the survey. Third-party forecasts matching recipient reports makes a pure demand-effect explanation less likely, not impossible.

The useful finding is narrower. In the ordinary exchanges tested here, people gave too much weight to the risk of imperfect delivery and too little to the warmth the recipient might hear.

The words did not need to arrive without awkwardness to be received as kind.