Limiting the length of time that establishments stay on the survey panel by dropping a proportion of them after a certain period of time and replacing them with a new sample of establishments. Generally only done with the smaller respondents, for whom it is felt that responding to surveys imposes a significant burden.