Impact of the Number of Response Options and Scale Length on Reliability
Valerie R. Moody*, Katherine A. Judd, Jennifer L. LaChapell, Tracy H. Mulderig, and Amelia Bobzien
Dr. Teresa Heckert, Faculty Mentor
In the literature, there is inconsistency in the optimal number of response options for maximizing reliablity of typical performance tests. The recommended maximum was found to be five (Bandalos & Enders, 1996), seven (Cicchetti et. al., 1985) or ten (Thorndike et al., 1991). The usage of a neutral option has been both criticized (e.g., Masters, 1974) and supported (e.g., Guy & Norvell, 1977). Given that reliability increases with scale length, the optimal number may be influenced by test length. No published research was found on this interaction. Our study compared the reliability of five, six, and seven response options for scales of varying lengths (i.e., 5, 10, and 20 items). We found that the optimal number of options depended on the test length.
Keywords: reliability, response options, psychology, testing, evaluation, test length
Topic(s):Psychology
Statistics
Presentation Type: Poster
Session: 3-5
Location: SUB-GEO
Time: 4:15