The MUM Effect: Why Students Withhold the Criticism You Most Need

Answer first: There is a well-documented human tendency to withhold or soften unfavourable information when we have to deliver it to another person. Psychologists call it the MUM effect — "keeping Mum about Undesirable Messages" — and it was demonstrated by Sidney Rosen and Abraham Tesser in 1970. In course evaluation, the MUM effect means the feedback most useful to an instructor — the specific, critical, actionable kind — is precisely the feedback students are most reluctant to record. Reducing the interpersonal cost of delivering it is a design problem, and "just make the survey anonymous" is a cruder solution than the evidence supports.

What the MUM effect is

Rosen and Tesser found that people are systematically reluctant to communicate bad news, even when they are not responsible for it and even when the recipient needs to hear it. In a much-cited field study, Tesser, Rosen and Tesser tracked how long it took to convey an agency''s decision to applicants for disability aid: participants took longer to communicate a denial than an approval (Tesser, Rosen & Tesser, 1971). The reluctance is driven by self-presentation (not wanting to be the bearer of bad news), sensitivity to the recipient''s likely emotional reaction, and politeness norms that make delivering criticism feel like a social transgression (Rosen & Tesser, The Reluctance to Transmit Bad News).

Now put a student in that position at the end of term. They liked the lecturer personally. They know a real human will read the comment. The course was genuinely disorganised, but writing "the module was poorly structured" feels like an accusation aimed at someone who was kind to them. The MUM effect predicts what happens next: they leave the box blank, or they write "it was fine," or they bury one soft caveat inside three compliments. The evaluation captures their politeness, not their experience.

Why this is worse than ordinary non-response

This is not simply non-response bias, where the missing feedback is missing at random-ish. The MUM effect is selective on valence: it suppresses negative information specifically. That has three consequences a quality office should worry about:

1. Upward bias in the mean. If criticism is systematically softened while praise flows freely, average scores drift higher than the truth — compounding the ceiling effect that already makes almost every course score four out of five.
2. Loss of the actionable part. The comments that would actually tell an instructor what to change are the ones most likely to be withheld or blunted. You are left with feedback that is pleasant and useless.
3. Distorted comparisons. Instructors who are personally warmer may attract more MUM-effect softening — students protect them harder — so warmth can inflate scores independently of teaching quality, echoing the warmth-versus-competence bias and the Dr. Fox effect.

Anonymity helps — the evidence

The obvious lever is distance: remove the sense that a specific person will be hurt by a specific comment. The survey-methodology literature is consistent that reducing the felt presence of a human recipient increases candour. Self-administered and computer-mediated modes elicit more honest reporting of sensitive or unflattering information than interviewer-administered ones, because respondents feel less social pressure and perceive the computer as an impartial, non-judging channel (Springer, disclosure across self-administered modes, 2014). In one widely-cited illustration, employees asked whether they had witnessed discrimination reported it far more often in an anonymous online survey than in face-to-face interviews. Distance lowers the interpersonal cost of the bad news, and the MUM effect weakens.

"So just make it anonymous" — the counterargument

Here is where the easy conclusion breaks. Maximal anonymity is not the same as maximal truth. Research by Lelkes, Krosnick and colleagues found that granting complete anonymity can, paradoxically, reduce the accuracy of self-reports and increase survey satisficing — respondents feel so unaccountable that they stop investing effort, straightlining and giving careless answers (Lelkes et al., 2012, Journal of Experimental Social Psychology). Anonymity removes the interpersonal cost of honesty, but it can also remove the sense of responsibility that makes people think before they answer — the same acquiescence and straightlining problems that plague careless responders. And it can swing the other way entirely: freed of all accountability, a minority of students weaponise the box, feeding the negativity bias that lets a few cruel comments dominate.

So the design target is not "as anonymous as possible." It is: remove the felt cost of delivering criticism to a person, while preserving the sense of engaged, accountable-to-the-process responding. A blank anonymous comment box does the first and undoes the second. That is a genuinely hard interaction to design — which is the point.

Where Koji fits

Koji for Education is built for exactly that middle ground. Students speak to an AI-moderated conversational interviewer, not to a form that a named lecturer will read line by line — which strips out the interpersonal cost the MUM effect feeds on. There is no human on the other end to disappoint, so the specific criticism that would never survive a politeness filter can actually be voiced. But because the interview is conversational and structured rather than a blank anonymous void, it also does what raw anonymity cannot: it keeps the student engaged, gently probes vague or softened answers ("you said the pace was fine — was there a point where it felt too fast?"), and applies quality scoring to flag low-effort, satisficing responses instead of treating them as data. The moderation is standardised and bias-aware, so every student gets the same careful follow-up with none of a human interviewer''s inconsistency — and confidentiality is handled in a GDPR/AVG-appropriate way so students can trust the channel.

The result is feedback that is candid and considered — the combination neither a face-to-face conversation (too much MUM effect) nor a blank anonymous box (too little accountability) reliably produces. The same conversational engine underpins general customer and user research on the main Koji platform, wherever honest feedback about a person or product is hard to extract.

The criticism you most need is the criticism your students are most reluctant to write down. That is not apathy; it is a deep-seated reluctance to deliver bad news to someone they do not want to hurt. Design the interaction to carry that cost for them, and the useful feedback appears.

The international-student wrinkle

The MUM effect does not fall evenly across a cohort, and that is where it stops being a curiosity and becomes an equity problem. Politeness norms around delivering criticism to an authority figure vary substantially across cultures. Students from backgrounds with stronger deference norms — a large share of any European university''s international intake — may experience the reluctance to voice criticism more acutely, softening or withholding negative feedback that domestic students would state plainly. If your instrument relies on students volunteering unprompted criticism, you are not measuring teaching quality evenly; you are partly measuring how comfortable different groups feel breaking a politeness norm. That interacts with the language barriers international students already face: a student writing critical feedback in a second language must overcome both the interpersonal cost and the effort of phrasing a delicate point precisely. The likely result is a quieter, blander comment — read by a committee as satisfaction.

The practical implication is that lowering the interpersonal cost of criticism is not just about getting more feedback; it is about getting feedback that is not systematically skewed toward the students who find candour easiest. A gentle, standardised prompt that invites elaboration — "many students find one part of a module harder to follow; was there a part like that for you?" — normalises criticism as an expected part of the conversation rather than an act of rudeness. That reframing, applied identically to every student, does more for equity than simply promising anonymity and hoping the confident students are representative.

Collect the honest, specific feedback a form suppresses — explore Koji for Education.