The goal of test design is to devise an instrument that will provide a stable and accurate assessment of student ability in some area. One means of reaching this goal is through the use of latent trait models, which determine the relationship between the unobservable trait or ability and the observable test performance. Three common latent trait models include the one parameter (or Rasch) logistic model, the two parameter logistic model, and the three parameter logistic model. Of these, the Rasch model is preferred, due to its ease of application and the fact that test scores serve as a sufficient statistic for ability estimation. In addition, computer programs are more widely available for the Rasch model than for the other two models. Latent trait models also have a number of characteristics that make them superior to classical test models. Among them are the properties of objectivity and information, and the ability to identify biased test items. To test its application in the context of communication education, the model was used with two parallel forms of a 75-item multiple choice final examination. Given the results of the initial factor analyses, the data fit the assumption of test unidimensionality, as required by the Rasch model analysis. The model was able to recover known item parameters ("known" in the sense that they should be equal), and the shapes of the two test information functions were bell-shaped rather than flat. (HOD)