What a Safety Assessment Actually Measures

The nuclear industry has spent decades building one of the most rigorous safety cultures in any sector. The systems, processes, and training frameworks that govern physical safety on nuclear sites are among the most scrutinised anywhere. Which makes it an interesting question to ask why, in a sector with such an extensive safety architecture, the most consequential decisions - the ones that determine whether a critical concern gets raised early or managed quietly until it becomes a crisis - still depend so heavily on whether a specific individual feels, at a specific moment, that speaking will cost them less than staying silent.

The safety culture frameworks that nuclear has developed with great rigour are primarily designed around physical safety: process compliance, near-miss reporting, procedural adherence. They are extraordinarily effective at what they were built to measure - whether people are following the systems correctly. The gap they leave relates to something completely different (to paraphrase Monty Python!): whether the project controls team will name a diverging cost trajectory they have been privately tracking for two months, whether the alliance leadership team will surface a fundamental disagreement about risk allocation before it crystallises into a dispute the contract was not designed to resolve, or whether anyone at the table will proactively and positively challenge a senior leader's technical judgement at the moment when the decision actually matters.

What determines whether those three things happen - naming, challenging, surfacing - is a set of professional and psychosocial conditions that general engagement surveys were neither designed to reach nor equipped to measure. These conditions are separate from the overall sense that a workplace is supportive, which most professional organisations achieve at an adequate level. They sit specifically in the territory of what it costs, in practice rather than in policy, to speak a difficult truth, and whether the culture of the organisation makes that act feel like professional responsibility or like professional risk.

Amy Edmondson's work at Harvard Business School draws this distinction carefully, and it is the foundation on which a properly designed assessment is built. What her research identifies is that the conditions predictive of whether teams perform the specific acts that honest challenge, learning, and early problem-naming require are not the same conditions that produce general workplace wellbeing. They form through repeated experience of how the organisation actually responds when someone speaks. Whether naming a concern has historically landed as constructive or as disloyal - in practice, not in the policy document - whether challenging upward is actually experienced as professional responsibility or quietly understood as a career cost, and whether what people say in the governance meeting matches what they believe and then share with team colleagues: these questions have answers, and those answers are measurable in ways that aggregated sentiment scores cannot replicate.

An assessment built around these conditions tells you something an engagement survey cannot. It calibrates whether the specific professional safety conditions your programme actually requires are present, what the gaps are, and what those gaps mean for how the programme should be structured and led. For a complex infrastructure delivery programme where decisions made by leadership teams under commercial pressure carry consequences in the hundreds of millions, that calibration needs to reflect a considerably higher bar than the one an organisation operating in stable and well-precedented conditions would need.

Hinkley Point C is under construction, Sizewell C is in development, and the small modular reactor programme is forming a delivery pipeline that will bring organisations together in combinations that have no direct precedent in UK nuclear. Each requires leadership teams to make consequential decisions under sustained commercial and schedule pressure, in conditions where the cost of not raising the concern is potentially greater than the cost of raising it too early. Whether those teams have the professional safety conditions to make those decisions well is invisible to general engagement surveys and safety compliance metrics alike. It requires assessment of the specific conditions that predict whether honest challenge, early problem-naming, and genuine accountability for decisions will actually occur, as distinct from being stated as organisational values.

For any organisation entering a complex major programme, having a positive and behaviour-led safety culture is the starting assumption - all organisations of any credibility do. The more precise and more useful question is whether the specific professional safety conditions relevant to the decisions it is required to make have actually been calibrated, and whether that calibration has happened before the programme design is locked, while the evidence still has ability to change something, rather than after the first crisis has confirmed a gap that was already there.

A psychological safety assessment designed for this purpose creates a baseline that grounds subsequent decisions - about programme structure, leadership selection, and what support people will need to challenge well - in evidence rather than assumption. The starting point is the calibrated picture of where the specific conditions for professional honesty under pressure actually stand, and what that means for the programme before anything more costly or harder to reverse is committed to.

What does your organisation's psychological safety assessment actually measure - and would an honest answer to that question change any of the programme decisions you are about to make?

References
Edmondson, A. (1999) Psychological safety and learning behavior in work teams, Administrative Science Quarterly, 44(2), pp. 350-383.
Edmondson, A. (2018) The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley, New Jersey.
Great British Nuclear (2025) Small Modular Reactor programme framework. Available at: https://www.gov.uk/government/organisations/great-british-nuclear
Advance Consultancy (2026) Activate Foundations: Psychological and psychosocial safety assessment for complex delivery environments. Overview document available on request from Advance Consultancy.