This paper demonstrates that those random procedures for selecting individuals in households will lead to enormous shortcomings in the representativeness of the individual sample while the agencies will need a second or third wave to obtain a minimum of representativeness. This is a costly sampling procedure which can easily be avoided. The basic assumption underlying the study presented in the paper is that nowadays one-person households and two-person households are in the majority in most European countries. This means that individuals living in such households have a far greater chance of being drawn in random individual samples than persons living in three-or-more person households. As persons living in smaller households have different socio-economic characteristics from persons living in larger households, random samples of individuals lead to wrong results - not only in terms of socio-economic characteristics, but also in the results of practically every market research study based on individuals. The study described was performed in NIPO's Telepanel where a questionnaire was used by applying different selection procedures. The paper describes the results of a few questions about political preference by applying the various methods. Finally an extremely practical solution will be discussed. This selection method - whose value is borne out by the results - is to select within the household: in a panel situation: youngest person (from a certain age) in a cross-section situation: youngest male and if not present youngest female (from a certain age).