Studies of victim number effects in charitable giving consistently find that people care more and help more when presented with an appeal to help an individual compared with an appeal to help multiple people in need. Across three online experiments ( N = 1,348), Bayesian estimation revealed the opposite pattern when people responded to multiple appeals to help targets of different sizes (1, 2, 5, 7, and 12). In this joint evaluation context, participants donated more to larger groups, when appeals were presented in both ascending order (Study 1) and random order (Study 2). The pattern held whether or not participants saw an overview of all appeals at the start of the study and when a single individual was added to the array (Study 3). These results clarify how compassion fade findings typical of separate evaluations may not generalize to contexts in which people encounter multiple appeals within a short temporal window.