Research note
The model everyone teaches was published in 1965 as a conceptual proposal that its own author said still needed testing. Twelve years later he came back, found almost nobody had tested it, and added a stage anyway.
Tuckman synthesised fifty studies drawn from therapy groups, human-relations training groups, and natural or laboratory task groups.
What he found was a consistent pattern running on two tracks at once. Groups develop as a set of relationships and as a way of handling work, and the two don't always move in step.
How members act toward each other, relate, and build structure. Who defers to whom, who challenges, what becomes acceptable.
How members orient to the work, interpret what it asks, and go about solving it. Independent of how they feel about each other.
Tuckman presented this as a conceptual statement, not a finding. He said in the paper that it still had to be tested.
Twelve years on, together with Mary Ann C. Jensen, Tuckman returned to a simple question: had anyone actually tested the model? They reviewed twenty-two empirical studies published since 1965.
Exactly one of them, Runkel and colleagues in 1971, had set out to test the four-stage sequence directly. Many others aligned with it partially. Several newer models had independently introduced a final stage about separation, disengagement or termination.
That last pattern was the decisive one. Across therapy groups, training groups, student cohorts and community groups, the end of a group's life kept showing up not as a fading out but as a phase of its own, with its own emotional and structural dynamics.
This five-stage version is the one now taught almost everywhere in leadership development, project management, education and organisational behaviour.
The model has taken real fire, and some of it lands.
The 2007 DAU study of 321 US Department of Defense teams suggests storming may not be a phase at all. In 2010 Doc Norton wrote a piece titled "Tuckman was wrong."
The fair verdict is that Norton goes too far. The model is a useful lens with known distortions, and a professional names the distortions rather than either defending or discarding the whole thing.
Regression. Lingering in a stage. Stages overlapping. And genuine doubt about whether storming is separable at all. It's a simplified picture of reality and it should be presented as one.
Clarify goals and roles. Keep the group small where you have the choice. Build in self-disclosure early. Treat psychological safety as the foundation rather than a bonus.
Most teams that struggle have vague agreements, unspoken rules, and a shared goal too abstract to move anyone.
The finding that storming may be an artefact rather than a stage matters directly to any model built on Tuckman's sequence. It supports treating conflict as something that recurs across phases from different sources, rather than as a gate a team passes through once, and it puts the weight on preconditions during early planning rather than on managing a predicted storm in week six.
This note is an editorial addition, not a finding from the source material.