5.5 Limitation This study had several limitations. First, the important limitation was the small sample size because the drop-out rate was higher than our expectation. All potential volunteers were already screened so we could not recruit more subjects for the study. This could result in producing a type-I error as reported in McMurdo and colleagues study. A second limitation was the ability of elderly persons to comply with the protocol of eye-head coordination exercise. To induce the adaptation process, the eye-head coordination exercise must be done at the speed high enough to cause the retinal slip. The elderly, however, could not cope with our required speed affecting the efficacy of the exercise. Third, many participants had high balance performance before study leading to the ceiling effect of our outcomes later on. Our screening criteria were based on our pilot study. We could not recruit only people who had the balance problem because there would be only few people available. Then, we also used numbers of fall risk as another inclusion criteria with the assumption that people with fall risks were fall-prone. Fourth, blinding the participants and physical therapist was impossible in this type of study because our subjects lived in the same place. Fifth, the changes in the VOR gain were possibly too small to be detected by our technique. Sixth, all subjects were female elderly. We did not intend to limit our subjects to only female group. However, only few male elderly met the inclusion criteria and they refused to participate in the study. Therefore, the generalization of this study is still limited.