Abstract
A high degree of validity for an operationalised concept is indicated if different types of data aimed to measure the same concept have strong correlations as compared to the correlations of the same type of data aimed to measure related but different concepts. With respect to the communicative ability of profoundly retarded persons, recent studies have almost exclusively used observations as the data‐collection method. This paper presents an investigation of the statistical relationships between observational data and interview data aimed at measuring the complexity of communicative behaviours and frequency of their use. The results will be discussed in terms of the validity of different modes of data‐collection. Finally, the relationships between complexity of communicative behaviours used and frequency of use of communicative behaviours in profoundly retarded persons will be discussed.