Why you need to know this: if you have ever sat in a meeting while somebody was treated badly and said nothing, or run a team in which that happens, this briefing explains what is most likely holding people back, what it costs, and what it takes for one person to change the room.
Most people who have worked for long enough have been in the room. A senior colleague takes apart somebody junior in front of everybody, the room goes quiet and the meeting moves on. Afterwards two or three people send the person on the receiving end a private message saying that was out of order. The usual explanation is that people are cowards, or that they do not care, and neither survives the evidence.
Observers do care. A 2025 pooling of the research on witnessed mistreatment at work found that people who see it disapprove of the person doing it and still tend not to intervene (Lyubykh, Zhong, Vuong, Robinson & Hershcovis, 2025). In one interview study, 34 of 40 professionals had stayed silent about a concern at work (Milliken, Morrison & Hewlin, 2003).
Five things hold them back, and fear is only one. Each person waits for somebody else and assumes the others are less bothered than they are (Westphal & Bednar, 2005). Many have learned that speaking up changes nothing: in two experiments with 654 people, those shown repeatedly that their voice made no difference stopped using new chances to speak (Andrieu, Milhabet, Denis-Noël & Steiner, 2024).
Speaking up can cost the speaker (Burris, 2012). Observers read the situation, and if they decide the target deserved it they tend to pull away from the target rather than help (Mitchell, Vogel & Folger, 2015). And the organisation sets the rules: where staff see their organisation as having switched off its own moral concern, they stay silent more, across 3,050 workers in 113 organisations (Fida et al., 2025).
The surprise is how much one or two people change. When two peers refused to go on in a classic obedience experiment, full obedience fell from 26 of 40 people to 4 of 40 (Milgram, 1965). And in CCTV footage of real public conflicts in three countries, somebody intervened in nine of ten, and more onlookers made intervention more likely rather than less (Philpot, Liebst, Levine, Bernasco & Lindegaard, 2020).
What to do. If you are in the room, decide before the meeting what you will do, prefer small moves that interrupt rather than heroic ones, and find the second person in advance, because the second voice may matter as much as the first. If you have power in the room, use it, since observers with more power are more likely to step in (Hershcovis et al., 2017).
If you run the room, watch what happens to the first person who speaks, make intolerance concrete rather than declared, and check what your reward system teaches. Be sceptical of training on its own: a randomised test of a speaking-up workshop for 71 anaesthetists found no difference in whether they spoke up (Raemer, Kolbe, Minehart, Rudolph & Pian-Smith, 2016).