The semantics of questionnaire rating scales
Graham David Low · Evaluation & Research in Education · 1988
Handbooks on questionnaire design generally cover ‘logical’ aspects of question design and statistical approaches to score processing in considerable depth; only rarely, however, do they consider the equally important semantic aspects of the rating scales used. This paper focuses on five main areas: (1) labelling midpoints; (2) the scaling of uncertainty; (3) semantic aspects of unidimensionality; (4) the differential gradability of adjectives; and (5) some limitations of Likert‐type scales. The main conclusion is that, if a questionnaire is to be at all valid, the designer must check the meaning and applicability of each point, or value, on each rating scale to every question asked, however time‐consuming this may in practice prove to be.