The disposition effect, where people sell winning stocks too soon and losing stocks too late, is not accurately measured by the common method of comparing gains and losses. This method does show a consistent bias, but doesn't pinpoint the disposition effect itself. The study found that the size of this difference, rather than just its direction, can give clues about why this bias happens. The researchers also predict that different markets, how much information investors have, and how much attention they pay can all affect this bias.