Making spatial comparisons in visualizations with multiple views is often difficult (e.g., comparing locations is difficult across small multiples, and overlay techniques quickly become cluttered). To improve multi-view spatial comparisons, we developed Shadow Marks – visual marks that are replicated across all views – that provide a common spatial reference frame at the location where a comparison needs to be made. We evaluated Shadow Marks through three crowdsourced studies that explored different visual-analytics tasks. Study results show that Shadow Marks consistently led to more accurate answers than either a small-multiples grid or an overlay. Shadow Marks also required less effort and were strongly preferred, showing that user-controlled spatial reference frames can improve multiple-view comparison.