Individual panels and housing panels can be used to reconstitute individual trajectories over a number of years. In the first case, which is illustrated by the European Panel, interviewees are tracked from one year to the next even if they move. In the second case, however, housing is resurveyed regardless of whether its occupants have moved. An individual panel is required for longitudinal analyses when the phenomenon studied is not independent of the existence of a move, when the observation period is long or when the population to be studied is mobile. Not tracking people who move is tantamount to depriving the analysis of an increasingly large proportion of the sample, which could substantially undermine the sample's representativeness. Descriptive statistics, such as the three-year rate of employment for young people initially in education, could be distinctly altered by such a decision. Conversely, cross-section statistics drawn from housing panels are of higher quality, since they are less subject to attrition. Hence, choosing between an individual panel and a household panel is practically the same thing as choosing between cross-section representativeness and the possibility of long-term tracking. A comparative study of France's rates of transition between employment, unemployment and being out of the labour force in the first four waves of the European Panel and the Employment surveys shows that the panel's loss of representativeness remains negligible after four waves and, in any case, of lesser importance than the memory effects.