Paired comparison data is common in psychological and consumer research, and requires specialized techniques to properly analyze. This study focuses on applying one such technique, a log-linear pattern model, to data related to audio treatment preferences on media clips. In addition, this analysis takes the unique step of accounting for dependencies between participant decisions and includes a subject-level cognitive covariate, reliable digit span. I also compare models under two different conditions: 1) with audio-only stimuli and 2) with audiovisual stimuli. This research demonstrates the complexities in application of such a model and the implication of certain trends of worth parameters. Results indicate that participants of this study struggled to make consistent results, possibly due to having difficult in discriminating between the audio treatments. However, some minor findings are that reliable digit span may be related to a certain audio treatment preference in an audiovisual context. Limitations and future directions for this style of research are discussed.