A procedure is developed for determining appropriate levels of scale reliability by examining coefficient alpha in conjunction with the standardized regression coefficient for each variable. Present study aims to examine the effect of addition and deletion of items on coefficient alpha. Also, necessary assumptions for appropriate interpretation of coefficient alpha are examined. Present research suggests that deleting scale items to increase coefficient alpha may result in an under specification of the construct the scale attempts to measure. Furthermore, this research offers prescriptive and descriptive insights for appropriate use of coefficient alpha.