Knowledge of the influence of variations in rating scale characteristics  generally is limited by the restriction of study to single scale items. This experiment  broadens the scope in this research area by studying the impact of varying numerical assignment to scale points and scale balanced-ness on the responses and internal consistency reliability of a well established multi-item instrument. Results indicate that this numerical format manipulation significantly influenced mean responses but not scale reliability. The balanced-ness manipulation also did not alter reliability, but positively unbalanced scales produced higher means than balanced scales. Implications and suggestions for future research are discussed.