Calculation of Cronbach's Alpha which measures the internal reliability of a tool is easy, but misuse and misconception are widely observed. Common misunderstandings of the Cronbach's Alpha include inability to differentiate between being one-dimensional and consistency, and between raw and standardized alpha. Violations of the assumptions of alpha could also cause an increase or decrease in the calculated alpha. In this paper, we showed that a very large but confusing alpha coefficient could be obtained, particularly when the one-dimensional theory is violated. Also, the use of standard error measurement as an index of the reliability has been presented. Finally, the generalizability of alpha coefficient is also discussed.
