Due to the many advantages of online surveys, many researchers are taking advantage of this survey method. Although for many psychological instruments, previous studies have shown that online and paper-and-pencil administration formats have equivalent results, other studies have shown that some online surveys result in different score distributions from those observed through paper-and-pencil administration. In this study, we conducted surveys using Zung’s self-report depression scale (SDS) to Japanese undergraduates both online and through paper-and-pencil and examined whether there are differences between different administration formats only in the scale’s mean structures and, if so, why the difference occurs. Analysis results showed that there was the difference only in mean structures. Results also implied that the online administration format lowers item thresholds; this decrease would cause the difference between the two formats’ mean structures. Finally, we think about the future directions of this research; to examine whether similar results would be seen in other scales, other countries, and other generations.