The development of the emerging technology of item writing was motivated in part by the desire to reduce potential subjectivity and bias between different item writers who attempt to construct parallel achievement tests. The present study contrasts four test forms constructed by the combined efforts of six item writers using four methods of item writing: (1) informal and subjective, (2) objective-based, (3) sentence transformation with writer's choice of distractors, and (4) sentence transformation with algorithmic creation of distractors. Three item bias detection techniques were used to examine potential sex bias in each test form: (1) delta plots of transformed item difficulty in male and female groups, (2) Scheuneman's chi-square-correct, and (3) internal consistency estimates (alpha) for each test form in each group. The basic data of the study were from 218 4th- and 5th-grade students who were randomly assigned the four 18-item multiple-choice test forms. The alternative sentence transformation method with item writers selecting the distractors proved to be somewhat superior to the objective-based test form. The study demonstrates the importance of attending to the method of item writing when potential bias is of concern in the construction of achievement tests. (Author)