The present article demonstrates the use of random-effects models, e.g. latent growth models (LGM), for the estimation of media effects on attitudes in longitudinal designs. On the basis of data from a three-wave panel study on the asylum law campaign in Switzerland, we show that the attitude change process can be modeled more appropriately relying on LGM compared to conventional procedures. As a consequence, applying LGM revealed moderate effects of media use on political attitudes toward the asylum law in the course of the campaign. More importantly, we demonstrate that LGM is likely to unmask media effects that were not obtained with conventional autoregressive fixed-effects models (ARM).