Abstract: | In 1982 the Netherlands Central Bureau of Statistics started large-scale survey research on the supply of hidden labour. As it was not known what type of survey would yield the best results, eight survey variants were tried out. They differed in the medium of communication between the researcher and respondent (face-to-face, mail and telephone) and in interview design (direct approach, gradual approach and ‘free-form’ conversation). Several criteria were used to assess the survey variants on their suitability for obtaining statistical information on the magnitude and structure of the hidden labour market. With respect to non-response and item non-response, face-to-face methods yielded the best results. In combating incorrect responses, a gradual introduction of the topic ‘hidden income’ is to be preferred. |