Influential models and studies of public opinion formation identify party elite cues as prominent drivers of public policy opinion. However, there is substantial variation in effect sizes across studies, and this variation is a barrier to the generalizability, theoretical development and practical utility of party elite cues research. In this paper, I estimate the variation in party elite cue effects that is caused simply by heterogeneity in the policy issues studied. I analyze three datasets from party elite cue experiments that contain between 10 and 34 U.S. policy issues each. I estimate the variance across the unobserved population of policy issues in both (i) the basic party elite cue effect and (ii) its relationship with the putative moderators of political sophistication and need for cognition. My estimates of between-issue variation in the basic and moderator effects equate to somewhere between one-third and two-thirds of the between-study variation previously observed in the literature. This highlights that a majority of existing between-study variation in party elite cue effects could simply be caused by the studies’ limited sampling of policy issues. I conclude that sampling a larger number of policy issues would substantially improve the generalizability of studies’ estimates of party elite cue effects, thereby providing a firmer empirical foundation for theory building, testing, and for applying the results to predict future cases of party elite influence.