Classical random-route sampling designs were continuously modified in order to reduce costs on the one side and as a reaction to a new type of full time interviewer on other side. The modifications have led to a situation where a random route sample is no longer a probability sample. Big social- and market-research institutes combine the random walk with a quota approach.
In this paper three different types of random-route sampling designs are compared: a classical approach with a controlled random walk and in a different step the interviewing of a fixed gross N of interviewees, a random walk defined by the number of net N completed interviews, and a random walk combined with a quota approach. The different types of sampling are discussed and the results of fieldwork demonstrated.