Abstract: From the psychometric point of view, response styles are serious threatsto the reliability and validity of close-end psychological tests and question-naires. In this paper, we introduce some of the common response styles,and the statistical methods to control them. However, we argue that thesemethods are all inadequate because they are based on the strong assumptionthat response styles are stable. A study was conducted to investigatethe effect of the length of questionnaire on response styles. The resultsshowed that as the length of questionnaire increased, the subjects' respon-ses became less extreme, and more stereotyped. The implication of responseset instability in the conventional statistical control of response setswas discussed. We suggest that investigators avoid using lengthy question-naires in their research.

Key words: NULL