The imaginaiy questionnaire

The experiment was designed to throw some light on the statistical problems in the analysis of questionnaire data. Previous work (unpublished) suggested that a simple choice response was partially determined by previous responses; and also that the nature of the determination was changed with changing length of series. A “null” experiment was devised in the form of a questionnaire without any questions, and the distribution of responses was studied with respect to the problems formulated.

The observations are discussed in three sections.

In the statistical discussion an alternative meaning to overall association or dissociation is advanced. This: relates association or dissociation to human behaviour in the serial response situation, rather than to qualities of the questionnaire. It is further suggested that association between specific, questions should be tested against the association in the whole questionnaire, and an appropriate treatment is indicated.

The observations depart from statistical randomness in certain ways. Answers made up almost entirely of one form of response are given less often than would be expected. Long sequences of the same type of response are relatively infrequent, and sequences of alternation of response are also rare. As the material is “null” it implies that the human concept of randomness differs from the statistical concept.

An attempt is made to define the human concept of randomness. It appears that a series of responses which has a pattern, or for which the subject can postulate a simple “cause” will not be accepted as random by the human subject. This raises problems of a perceptual and cognitive nature. It also has a bearing on the design of questionnaires. or experiments involving serial responses.