Abstract
The purpose of this article is to examine the effect of trust in government on voter turnout in Japan, based on the analysis of large-scale panel survey datasets. In previous works, on static approach, it had been argued by many scholars trust hardly influences on voter turnout. This approach, however, is so inadequate for an examination the effect of trust could have been estimated only to partial. In this article, I argue that not only perception in present but also in the past must be necessary for it, and, on dynamic approach, there is “continuing” and “reflected” effect in trust. The results of logit estimation and post-estimated simulation show (1) the effect of trust in the people whose perception in the past and present is same is stronger than different, and (2) even if the perception in present differs, that difference is not effective when the perception in the past also differs.
