Variation in results of measurement repetition of human characteristics and activities

The analysis of measurement variation in Ergonomics/Human Factors research reflects different approaches such as those prevalent in either the technical sciences or in the social sciences. The distinction is often a consequence of different types of measuring, for example measuring 'at' human beings as opposed to measuring 'through' human beings. In measuring 'at' human beings, as in anthropometrics, the measurement can be repeated several times. This allows the dispersion in results to be demonstrated via their standard deviation, as is usual in the technical sciences. In measuring 'through' human beings as, for example, in their performance in force exertion, a test-retest design is often adopted, i.e. a single repetition through several subjects in order to anticipate carry-over. In a number of research papers which were scrutinised, test-retest dispersion seems to have been analysed incompletely, ambiguously or incorrectly. In this paper a more appropriate way to specify within-subject dispersion is proposed, taking into account the dispersion patterning throughout the subjects. The relevance of studying this patterning is discussed, in terms of insight into human control, setting margins in design and for limiting the number of measurement repetitions per subject.