2016 年 43 巻 2 号 p. 83-101
We report the results of a randomized experiment conducted in Japan as part of a mail survey of voters. In this experiment, we examined whether or not inserting a pen in the questionnaire would improve the response rate. We employed a two-by-two treatment in which we varied the price of the pen (high-priced or low-priced) and whether or not the pen was inserted in a paper-made box. We found that providing a pen increased the response rate by more than nine percentage points. We also found that a low-priced pen inserted in a box was more effective than other condisions. However, the effect of the high-priced pen was not statistically distinguishable from that of the low-priced one.