Abstract
General population data were sought on the question of what is the optimal number of response categories for attitude and personality scale items. When an ethnocentrism scale was administered to 100 randomly sampled residents of Johannesburg, it was found to show a reliability of 0,73 with five-point options. Scoring the extreme options as the same gave also a three point version of each item. So scored, the scale reliability dropped to 0,65. In a second study the Jackson Dominance Scale was administered to two random postal samples of the Australian State of New South Wales. When administered and scored with three response options for each item the reliability was 0,84. Collapsed to two options the reliability of the scale dropped to 0,80. When administered and scored with only two response options the reliability was 0,81. It was concluded that collapsed scoring is a useful methodology for examining the effects of varied response options and that more response options generally improve reliability.
Get full access to this article
View all access options for this article.