Experts' voting behavior is conjectured to be more objective than peer voting (own group/peers) and public voters (everyone interested), who are supposedly influenced by all sorts of subjective aspects. We examine differences in voting behavior between these groups by analyzing the voting outcomes for all-star teams in American Football. This paper analyzes the impact of performance as well as non-performance markers and team effects on the voting outcome. It contains a comparative analysis across the mentioned groups to elaborate on differences. The econometric analysis uses unbalanced panel data of All-Pro and Pro Bowl player selections over 78 seasons (1951-2019). It applies panel probit regression to assess the impact of the markers on the outcome probability of winning one of the All-Star awards. We find that expert, peer, and public voting show similarities and are partially driven by the same performance and non-performance markers. However, none of the three analyzed voting systems is free from the influence of non-performance markers. We find exposure effects as well as effects from team affiliation in all of them, including in fact expert voting. Positive effects of team success are found in expert and, to a lesser extent, in peer voting. Team-specific effects are found in public voting, providing evidence for partisanship voting by fans. Our results shed doubt on the suspected objectiveness of expert voting. Furthermore, they fortify the notion of public voting being inefficient at identifying objective quality and extend the literature on voting biases among experts, peers, and the general public.