Following these recommendations, candidateitems for perceived usefulness and perceivedease of use were generated based on their conceptual definitions, stated above, and then pretested in order to select those items that bestfit the content domains. The Spearman-BrownProphecy formula was used to choose thenumber of items to generate for each scale. Thisformula estimates the number of items neededto achieve a given reliability based on thenumber of items and reliability of comparableexisting scales. Extrapolating from past studies,the formula suggests that 10 items would beneeded for each perceptual variable to achievereliability of at least .80 (Davis, 1986). Addingfour additional items for each construct to allowfor item elimination, it was decided to generate14 items for each construct.