The use of a specific non-response option on Likert-type scales

The paper argues that the use of non-response options in questionnaires relating to attitudinal research offers various advantages. First, if non-random patterns of response on individual items within a questionnaire occur, such patterns may hold valuable information for the researcher. Second, the provision of the non-response option may mean that any resultant analysis is based upon a sample with appropriate knowledge or opinions, thereby aiding discrimination. To support this view, various pieces of past research by the authors are presented where patterns of non-response are analysed. In each case it was found that the existence of the non-response option offered some new insight. Such insights could help to improve subsequent questionnaires, if used at the pilot study stage, or illustrate specific aspects of a sample's attitudes - for example, in one case, about uncertainty over a council's tourism policies. (C) 1999 Elsevier Science Ltd. All rights reserved.