The Effect of Item Format on Reliability and Validity: a Study of Multiple Choice and True-False Achievement Tests

Publisher Website
Google Scholar
Abstract
A sample of 529 nonurban high school students each responded to one of four test forms which differed in subject matter (natural science or social studies) and item form order (true-false items before or after multiple choice). The ratio of the number of true-false to multiple-choice items attempted in the first eight minutes of testing was 3:2. The reliabilities of the multiple-choice tests were significantly greater than those of the empirically-lengthened true-false tests. The disattenuated correlation coefficients for three of the four tests were significantly less than 1.00. Conclusions and explanations are discussed.