Earlier studies have found that a substantial part of the contributions in public good games can be explained by subjects misperceiving the game's incentives. Using a large-scale public good experiment, we show that subtle changes in how the game is framed substantially affect such misperceptions and that this explains major parts of framing effect on subjects' behavior. When controlling for the different levels of misperception between frames, the framing effect on subjects' cooperation preferences disappears. This suggests that merely changing how tax-, fine- or subsidy systems are framed, without reducing complexity, could reduce welfare loss from misperception of incentives.