Data collection involving human-human handover has provided enormous leaps in driving human-robot interaction research. However, all existing datasets lack information on giver and receiver preferences in handover interactions. Most previous studies have relied on small-scale human participant experiments involving a limited range of objects, where participants are often expected to share similar handover attitudes. Nevertheless, in real-world scenarios with diverse objects, it is likely that giver and receiver preferences will not always align. In this paper, we present a large-scale study of human-human handover behavior involving 96 participant dyads derived from 32 participants in total and 204 objects. Each dyad consists of 2 participants engaging in handovers, where after a giver-initiated handover, participants provide comfort ratings and binary responses indicating whether they agreed on the handover location. We also ask the receiver to demonstrate their preferred handover to gain detailed information on object pose at the handoff point. Our study captures 4-viewpoint RGB-D recordings of both giver-initiated forward handover and receiver-initiated demonstration handover. Using the collected data, we evaluate how the subjective ratings provided by participants correlate with objective measures of alignment of object orientation at handoff.