ABSTRACT
This chapter discusses the split-half reliability, a researcher administers a test but scores the items in the test as though they consisted of two separate tests. Typically, researchers do this by performing what is known as an odd-even split. Specifically, a researcher scores all the odd-numbered items and obtains a score for each examinee. The researcher scores all the even-numbered items and obtains a second score for each examinee. This process results in two scores per examinee. Then, the researcher correlates the two sets of scores, yielding what is known as a split-half reliability coefficient. An alternative to the split-half method for estimating internal consistency is Cron-bach's alpha, whose symbol is a. While the computation of alpha is complex, the concept is relatively easy to understand. Like the split-half method, it is based on a single administration of a test.
