The women in Cedar Rapids, Iowa, were being asked to cook kidneys, brains and hearts. Kurt Lewin’s field team picked those meats on purpose. The resistance to them, its 1943 report said, was “a rather deep one, frequently combining elements of physical aversion, social status, and superstition.”
Six Red Cross home nursing groups took part, 13 to 17 women each. Three heard a nutritionist lecture: the war effort, vitamins and minerals on charts, recipes, and how well the dishes went over with her own family.
The other three got the same information in seven to ten minutes, inside a discussion. A leader asked the women how well a direct appeal would work on housewives like themselves. They talked through why their families turned these meats down, the expert answered objections, and then they voted on trying one the following week. Each session took about thirty minutes.
Seven days later, interviewers visited each woman at home.
After the vote
After the lecture, 4 of the 41 women had served at least one of the three meats, 10 percent. After the group decision, 23 of 44 had, 52 percent.
The lecture spent more time on the facts. What it never asked for was a decision. Motivation, the report said, “leads to action only if it is brought down from the level of wishes and sentiments to the level of a ‘decision.’”
In 1947, in the opening article of the new journal Human Relations, Lewin named a second problem. A change toward better group performance “is frequently short lived; after a ‘shot in the arm,’ group life soon returns to the previous level.” A successful change, he wrote, includes three aspects: “unfreezing (if necessary) the present level,” moving to the new one, and “freezing group life on the new level.”
Changes often slide back unless the old habit is loosened and the new one held in place.
The rope that pulls both ways
Work on people one at a time, Lewin argued, and their tie to a valued group standard acts as resistance. Change the standard itself, and the same tie helps the change along and steadies each person at the new level. A shared habit works like a rope between climbers: pull one forward alone and the rope drags her back; move the whole party and the rope keeps her where they now stand.
He did not credit the decision alone. A housewife who had decided to use more milk, he noted, “might place a standing order with the milkman which could automatically keep milk consumption high.” Mapping the forces that hold a level in place is the job of force-field analysis.
Nor was group decision a trick to copy: “Managers rushing into a factory to raise production by group decisions are likely to encounter failure.”
Who drew the three boxes
The tidy textbook version, three boxes and two arrows, is mostly not Lewin’s. In 2016 Stephen Cummings, Todd Bridgman and Kenneth Brown went back to the source. The passage, they found, sits in the 24th of 25 subsections of a 37-page article, with no data or diagram of its own. Lewin never wrote “refreezing”; the earliest use they could find was a 1950 conference paper by his former student Leon Festinger. Ronald Lippitt in 1958 and Edgar Schein from 1961 turned the sketch into phases a “change agent” could run, and by the late 1980s it was the foundation of change management. Lewin “is assumed to have given us the terms unfreeze-change-refreeze,” they wrote, “which is only 33% right.”
Not everyone agrees. In 2020 Bernard Burnes argued that the three steps were “a well-thought-out approach to change” built on Lewin’s field theory.
The evidence underneath is thinner than the diagram. The Iowa check came after one week. The groups were small, and the report’s summary speaks of “about one hundred and twenty women” while its results table counts 85. The decision groups voted and the lecture groups didn’t, so the study can’t separate the talk from the vote. Lewin offered the three steps as a way planned change “may be thought of,” not as a tested recipe.
The third week
Imagine a clinic manager who wants her nurses to use a new handoff checklist. The memo goes out, the training goes well, and by the third week most handoffs are back to a quick word in the hallway.
Second try: the shift talks through the near-misses they have seen at handoff and decides together which items the checklist must always cover. Then the checklist goes into the handoff form that has to be signed, a standing order rather than a reminder. Nothing guarantees it lasts. The first time, though, nothing held the habit once the memo was forgotten.
Telling isn’t adopting
The lecture has an AI-era cousin: telling people what a new tool can do. Between November 2023 and January 2024, Anders Humlum and Emilie Vestergaard surveyed 100,000 Danish workers in 11 occupations exposed to ChatGPT. A random group was shown experts’ estimates of how much time ChatGPT could save on their job tasks.
Beliefs moved: the gap between workers’ views and the experts’ shrank by 15 percent, and most of that shift held two weeks later. Behavior didn’t. By then, the informed workers were no more likely to have actually used ChatGPT. The barriers they reported most were needing training (43 percent) and employer restrictions (35 percent).
This was a short message in a survey, not a lecture, and nobody in the study tried a group decision, so this working paper doesn’t show that Lewin’s method works for AI tools. Where it differs from Iowa is who holds the freeze: the rules and training that could keep a new tool in use sat largely with employers.
The Iowa lecture groups went home with the recipes. The decision groups went home with the recipes and a vote, and 23 of those women put one of the meats on the table.