Unseen information can lead to various "threats to validity" when analyzing complex datasets using visual tools, resulting in potentially biased findings. We enumerate sources of unseen information and argue that a new focus on contextual visualization methods is needed to inform users of these threats and to mitigate their effects.