In December 1994, observers at the University of Geneva Hospitals began the first of seven hospital-wide surveys of routine care. Every time a nurse or doctor reached a moment when hands should be cleaned, the observer noted whether it happened. It happened 48% of the time.
The misses had a pattern. In a study of 2,834 such moments, Didier Pittet and his colleagues found that the more often care called for clean hands, the less often staff managed it. Intensive care, with about 20 opportunities per patient-hour, had the lowest rate, 36%.
The usual remedy was more persuasion. A 2002 US guideline that Pittet wrote with John Boyce summed up the record of lectures, leaflets, workshops, feedback and even automated dispensers: they “have been associated with transient improvement.” Geneva’s campaign kept posters and feedback. But a sink wash takes time, and one intensive care study found nurses needed an average of 62 seconds to leave the bedside, wash and get back. So the campaign’s special emphasis went elsewhere. Custom-made holders for bottles of alcohol hand rub were mounted on every bed.
The handbrake, not the pedal
Behaviour settles where pushes and brakes balance, and easing a brake often beats more push.
Kurt Lewin set this out in 1947, in “Frontiers in Group Dynamics,” in the first issue of the journal Human Relations. In his account, a level such as a factory’s output stays steady because the forces pushing it up and holding it down are equal in strength. Production levels “are quasi-stationary equilibria which can be changed either by adding forces in the desired direction or by diminishing opposing forces.”
The two routes, he argued, don’t land in the same place. Raise the level by adding push and the new level comes with “relatively high tension”; get there by weakening the opposition and it comes with “relatively low tension.” Since tension above a certain degree goes with fatigue, aggressiveness and lower constructiveness, he concluded, “as a rule the second method will be preferable to the high pressure method.”
A car with the handbrake on crawls forward, engine screaming, when you stamp on the accelerator. Release the handbrake and gentle pressure moves it.
One force holding factory speed down, Lewin noted, was the fear that a temporary increase “would bring about pressure from the supervisor or foreman permanently to keep up the higher speed.” Experiments by Bavelas in a sewing factory lifted output to a new level, and Lewin thought they worked “at least in part” by reducing the forces that kept production down.
In Geneva the push already existed. The brake was 62 seconds at the sink, again and again through a busy hour; the guideline estimated a bedside rub takes about a quarter of that. By 1997 compliance had reached 66%, and the gain came from the rub; washing at sinks stayed about level. The authors credited bedside rub with a large part of the rise, though posters and feedback ran alongside it.
Arrows are guesses
In practice, force field analysis is a whiteboard exercise: a line for today’s behaviour, arrows pushing each way, lengths for how strong each seems. Its value: it makes you list the brakes, which a plan built on pressure tends to skip.
It is a way of thinking, not a measurement. Lewin wrote that the separate forces could be determined “only under special conditions,” so the arrow lengths are judgment calls. And the forces don’t hold still. Donald Swanson and Andrew Creed, studying organizational cases, argued that a straight-line diagram of forces risks missing key complexities: “The force field is not an impermeable thing; instead, it morphs.”
Geneva has a warning too. The 2002 guideline cautioned that setting out hand rub without ongoing education and motivation “may not result in long-lasting improvement.” Releasing a brake helps most while something still pushes.
Lewin’s better-known model of how a group lets go of an old way has its own page, unfreeze, change, refreeze. Kotter’s call to raise urgency, in these terms, works on the push side.
The late reports
Imagine a finance lead who needs her team’s expense reports by the fifth of each month. She has sent reminders, then firmer ones, then put the late names on a slide at the monthly meeting. Reports came in earlier for one month, along with grumbling.
Then she sits with one of the latecomers and watches him file. The system only opens on a laptop on the office network, and receipts must be scanned. He is on the road three days a week.
Another reminder might buy another short bump and a crankier team. Accepting phone photos of receipts could stop the deadline fighting his week, though it may not be the only brake.
When the brake is a policy
Between November 2023 and January 2024, the economists Anders Humlum and Emilie Vestergaard surveyed 18,000 Danish workers in 11 occupations exposed to ChatGPT. Among workers who thought ChatGPT could halve the time of a task, only about 23% planned to use it within the next two weeks. Asked why not, 42% said they needed training and 36% said their employer restricted use.
The researchers then tested a push, randomly telling some workers what experts estimated ChatGPT could save in their tasks. Beliefs shifted. Intended use barely moved, and actual use over the next two weeks showed no increase. Workers pointed to the same restrictions and training needs.
By the team’s 2024 survey, employers had moved: about 43% of workers were explicitly encouraged to use chatbots and 30% of employees had been trained. For a typical worker, take-up was 83% where the employer encouraged use and 47% where it didn’t.
That comparison isn’t an experiment, and it rests on Danish survey answers. Encouragement also bundles a push with brakes released, so it can’t show which did the work. But the earlier test points one way: telling people about the savings was a push, and by the workers’ own account the brakes held.