This research empirically tests three models that predict the expected intention change that results when people read and think about written advertising copy. The three models were derived from the principals of information processing 4 theory. This framework predicts that change in intention is a function of the discrepancy between a message and the person's initial stance before reading the message. From information processing theory, three models are derived and empirically tested against one another. The three models make specific predictions about behavioral and cognitive processing of the content in an advertisement. The three models tested in this research are termed the Presentation Model, the Random Model, and the Importance Model. The Presentation and Random Models assume that the way in which a multiple message is read determines the processing of the content. The Presentation Model assumes that people read down the page and change intention incrementally as they read. The Random Model assumes that people read copy by giving short consideration to each message and they cycle back through the copy, i.e., browsing several times. These two models were previously derived by Hunter, Danes and Cohen (1984). In addition to the aforementioned two models, this research proposed a third model, the Importance Model. This model assumes that people read the copy and enter it into short term memory. They then rank order the messages according to the importance of the attributes about which the message communicates a benefit. The processing then occurs in descending order of attribute importance. An experimental methodology was developed to test the assumptions made by each of these models and to examine the predictive validity of each model. Four treatments were employed to test the assumptions made in deriving the models. The presentation of the messages was computer controlled and the computer was used to simulate model assumptions. Also, magnitude scales were successfully employed to measure the independent and dependent variables of the three regression. The results of the research support the Importance Models as the appropriate models. This is based upon predictive validity and regression fit. This research expanded the information processing paradigm by offering a third processing model which is a more promising predictor of intention change. Future research in the area should expand upon these three models incorporating more homogeneous groups in terms of benefits. Also, future research should include allowances for logical distortions in the processing. Other topical areas for explorations are different delivery mechanisms such as verbal presentation and creative aspects of advertisements. / Ph. D.