The main point about refreezing is that new behavior must be to some degree congruent with the rest of the behavior and personality of the learner or it will simply set off new rounds of disconfirmation that often lead to unlearning the very thing one has learned. The classic case is the supervisory program that teaches individual supervisors how to empower employees and then sends them back into an organization where the culture supports only autocratic supervisory behavior. Or, in Lewin's classic studies, the attempt to change eating habits by using an educational program that teaches housewives how to use meats such as liver and kidneys and then sends them back into a community in which the norms are that only poor folks who can't afford good meat would use such poor meat.
The implication for change programs are clear. For personal refreezing to occur, it is best to avoid identification and encourage scanning so that the learner will pick solutions that fit him or her. For relational refreezing to occur, it is best to train the entire group that holds the norms that support the old behavior. It is only when housewives groups met and were encouraged to reveal their implicit norms that change was possible by changing the norms themselves, i.e. introducing collectively a new set of standards for judging what was"ok" meat.
In summary, what I have tried to show above is that Lewin's basic model of change leads to a whole range of insights and new concepts that enrich change theory and make change dynamics more understandable and manageable. It is a model upon which I have been able to build further because its fundamental concepts were anchored in empirical reality. Intellectual knowledge of the change process is not the same as the know-how or skills that are learned in actually producing change. In the next section I examine the implication of Lewin's thinking for the practice of change management.