Should employees be able to go over their manager's head?

Speaking up can produce the very hostility it reports.

This is the finding that makes the dilemma real rather than squeamish. A two-wave field study of 93 supervisors and 533 subordinates at a large Chinese company found that employee voice was positively related to leader identity threat, and had an indirect effect on abusive supervision through it. Supervisors experienced being told about a problem as a challenge to their authority and competence — evidence they had failed to spot it themselves.

Crucially, the effect was conditional. Where supervisor traditionality was high, "the relationship between employee voice and leader identity threat was significantly positive." Where traditionality was low, the connection was not significant. The risk of speaking up is not uniform: it depends on how much the individual manager's authority is bound up in deference.

Scope: one Chinese firm, one study. Directional, not settled, and not demonstrated in a Gulf setting.

The harm from bad supervision is heavily documented.

A systematic review covering 171 empirical manuscripts and 239 samples published between 2010 and July 2020 synthesises the consequences of abusive supervision for subordinates: increased turnover intention, emotional exhaustion and aggression, reduced performance, workplace deviance, destructive attitudes and daily counterproductive behaviours.

Whether organisations respond to it effectively is not covered by that evidence.

The same review does not examine whether organisational or people-function responses work. It documents consequences and analyses research methods, and contains no assessment of interventions, escalation protocols or reporting gaps. Across 171 manuscripts and a decade of accumulated research, the harm is measured and the remedy is not.

Where interventions have been tried and studied, the results are poor.

The silence and voice review states that "the accumulated evidence concerning speaking-up interventions in healthcare indicates that they achieve disappointing outcomes because of a professional and organizational culture which is not supportive."

Two cautions on that sentence. It is the review's characterisation of the field, drawn from meta-analytic work it cites rather than from its own data. And it is healthcare-specific. But it is the closest thing to direct evidence on our question that we located, and it points the same way as the gap above: the problem is not the absence of a mechanism for speaking up. It is the culture the mechanism sits inside.

Measures of employee silence are inconsistent.

An integrative systematic review of employee silence and voice covering 76 studies and 122,009 participants found the field's concepts and measures to be heterogeneous, with 45 distinct measures in use and no unifying theoretical background. It also records that silence and voice can occur at the same time in the same person, and that fear of retaliation appears as an explicitly measured dimension — one instrument asks how often respondents "remained silent because you are afraid of retaliation."

Scope: this review is healthcare-specific. Treat it as evidence about measurement quality, not as a general prevalence estimate.

The evidence leaves us with a genuine dilemma:

• Speaking up can expose the employee to professional risk, or to more hostile treatment.
• Silence can let abusive supervision continue, harming the employee, the team and the organisation.
• The existence of a complaint channel does not demonstrate that it can protect anyone or change an outcome.
• General survey scores do not reveal whether people will speak when the stakes are high.

So the decision cannot be reduced to "build an escalation channel." The sharper question is: what can that channel actually do once a complaint arrives?

Scope of the evidence overall. These findings come from China, from healthcare settings, and from a review of predominantly Western research. We found no Gulf evidence on escalation routes or on the consequences of raising concerns about a manager, and this page makes no claim about the region.

The strongest case against building an escalation route is that it can be worse than having none.

A channel that invites disclosure without being able to deliver protection transfers risk to the employee. They take the exposure; the organisation takes the reassurance of having a process. If the person receiving the complaint has no authority over the manager's decisions, an open channel is a trap with a friendly name.

Second, most conflict is not abuse. The bulk of what reaches a people function is ordinary disagreement — a hard conversation, an unwelcome piece of feedback, a decision someone dislikes. Routing that upward de-skills managers, teaches employees that disagreement is escalated rather than resolved, and makes managers defensive about exactly the conversations they should be having.

Third, escalation routes can be used against legitimate management. A manager addressing genuine underperformance is exposed to a complaint that reframes accountability as mistreatment. Any honest design has to survive that case, not just the sympathetic one.

Fourth, and most limiting: none of the evidence above tests escalation routes. The studies establish that voice can provoke hostility and that bad supervision causes harm. None of them shows that a skip-level channel, an ombuds function or a formal grievance process improves outcomes. A confident "build one" verdict would outrun the evidence.

Watch — and answer the authority question before building anything.

We are not able to tell you that escalation routes work, because nobody has shown that they do. What the evidence does support is narrower and still useful: speaking up carries real risk to the person who does it, that risk is higher under some managers than others, and the harm from leaving bad supervision in place is well documented. Those three things together mean the status quo — an open door with no authority behind it — is the weakest of the available options. Inaction is not the neutral choice it can feel like.

It also follows that any grievance or escalation process presented as evidence-based is overstating what exists. There is no body of research showing these processes work. That is not a reason to have none; it is a reason to stop claiming more for them than can be supported.

A related caution about measurement: an engagement score showing people feel able to speak up is thinner evidence than it appears. The instruments differ from one another, and the same person can be vocal in general and silent about the thing that matters.

Before designing anything, answer one question honestly: when a complaint concerns a manager, who has the authority to change the outcome? If the answer is the manager, or a peer of the manager who will not act against them, you do not have an escalation route. You have a listening service.

If you cannot create real authority, the honest move is to stop describing the channel as though it offers protection. Telling someone plainly what you can and cannot do respects them more than an open door that closes the moment it matters.

1. Trace the authority. Take the last complaint you received about a manager. Write down who ultimately decided what happened. If that person reports to, depends on, or is peer-protected by the manager concerned, your route is decorative.
2. Look at outcomes, not volumes. Count how many manager-related concerns you received in the last year, and how many resulted in any observable change. The ratio is the real measure of whether the channel functions.
3. Ask what happened to the people who raised them. Did they stay? Were they promoted? A pattern of quiet exits after complaints tells you what your process actually is.
4. Stop treating survey scores on "I can speak up" as settled. The measures are inconsistent, and the same person can be vocal generally and silent specifically.
5. Be honest in the moment. If you cannot protect someone, say so before they disclose, not after.

All three licences were checked by opening the article's own copyright statement on 9 August 2026.

1. Hu, Long, Wu and Xu — Frontiers in Psychology (2023) — An identity threat perspective on why and when employee voice brings abusive supervision. CC BY 4.0. Two-wave, multi-source field study; 93 supervisors and 533 subordinates at a Chinese company of 4,000+ employees. In our library · Original article
2. Gallegos, Guàrdia-Olmos and Berger — Frontiers in Communication (2022) — Abusive Supervision: A Systematic Review and New Research Approaches. CC BY 4.0. PRISMA review of 171 empirical manuscripts, 239 samples, 2010 to July 2020. In our library · Original article
3. Lainidi and colleagues — Frontiers in Psychiatry (2023) — An integrative systematic review of employee silence and voice in healthcare: what are we really measuring? CC BY 4.0. 76 studies, 122,009 participants, 45 distinct measures; database search January 2016 to January 2022. In our library · Original article

Further reading — verified primary sources, cited and linked, not summarised, and not counted toward the minimum.

• Previtali and Cerchiello (2021) — Organizational Determinants of Whistleblowing. A Study of Italian Municipalities, Public Organization Review 21, 903–918. CC BY 4.0 and genuinely usable, but not held in our library. Across 400 major Italian municipalities, actual reporting was positively correlated with having specific whistleblowing procedures and with training and education programmes, with anonymous reporting playing a relevant role. Note carefully what this measures — whether reports get made, not whether anything is resolved. Original article

• Richard, Boncoeur, Chen and Ford (2020) — Supervisor Abuse Effects on Subordinate Turnover Intentions and Subsequent Interpersonal Aggression: The Role of Power-Distance Orientation and Perceived Human Resource Support Climate, Journal of Business Ethics 164(3), 549–563. Copyright reserved and paywalled, so it cannot enter our library. It is listed because it is the most directly relevant study we located: across 324 supervisor–subordinate dyads in nine organisations, low power-distance orientation and high perceived HR support climate each independently weakened the effect of abusive supervision on turnover intentions. On that finding, being visibly supportive appears to matter even where the specific case cannot be resolved. We have read the abstract, not the full paper, and no open version was found. Original article