Recently, the social sciences have witnessed a rising interest in dyadic design, as an efficient way to disentangle mechanisms of interpersonal influence. Despite the relevance of this design to political research, few efforts have been made to collect and efficiently analyze dyadic data. In this article, we suggest the Actor‐Partner Interdependence Model as a useful tool to test bidirectional effects in dyadic data on political attitudes and behaviors. The model explicitly assumes that members of a dyad (reciprocally identified as actor and partner) involved in political communication are interdependent and influence each other. We apply the model to estimate the effect of partner's party identification on actor's vote choice, using 1996 Indianapolis–St. Louis dyadic data. Results show that partner's party identification is significantly associated with vote choice. Moreover, we show that influence between dyad members is moderated by their intimacy and that an increased difference in socioeconomic status between dyad members tips the balance of the effect in favor of the individual with more resources. Our conclusions point to the effectiveness of APIM in modeling interdependent asymmetric relations and call for increasing efforts in collecting dyadic data and in developing proper tools for their analysis.