里克特量表是心理测量和社会调查中广泛使用的一种量表，本研究主要探讨它的最佳选项数目和最佳标签样式。以 229名北京师范大学学生作为被试，使用 Rosenberg的自尊量表、Leary的交往焦虑量表和自编量表作为实验材料，选项数目从 4到9，标签样式包括全标签和两端标签，以内部一致性信度和被试偏好作为评价指标，结果表明：（1）选项数目和标签样式都不影响量表的内部一致性信度，被试得分变异也不影响选项数目与量表信度的关系；（2）5点量表最受被试偏爱，全标签样式比两端标签样式更有助于被试理解选项的含义、区分相邻选项程度上的差异。所以在以大学生作为施测对象的里克特态度量表中， 5点计分的全标签量表是最合适的。

Likert-type scales had been employed widely in psychological measurement and social survey. The present study investigated the optimal number of response categories and label formats. Rosenberg’s Self-Esteem Scale (SES), Leary’s Interaction Anxiousness Scale (IAS) and a self-administrated questionnaire were administered to 229 students of Beijing Normal University. The number of response categories ranged from 4 to 9 and the label format contained ALL form and END form. The indices of this study were internal consistency reliability and respondent preference. The results indicated: (1) The number of response categories and label formats had no effect on the internal consistency reliability, and the variance of respondent’s scores had no influence over the relationship between the number of response categories and the consistency reliability. (2) Respondents preferred 5-point scale, and they could understand the meaning of categories and distinguish the difference of them better from all-form scales than end-form. So, 5-point all-form likert scale was the most suitable with college students as the respondents in attitude measurement.