Discussion Papers

Effect of Adding Web-Mode on Unit-Response in Sample Survey:Evaluating the Effect of Adding an Alternative When No Control Group Exists (English translation of DP-2010-002)

DP Number SDP2012-003
Language 日本語・英語
Date March, 2013
Author Koji Yamamoto, Michio Naoi
JEL Classification codes
Downloads PDF

The first wave of Japan Household Panel Survey (JHPS) was designed to let
participants respond to questionnaires via the Web rather than paper if they desired. In
this study, we investigate whether introducing the Web option altered the response rate
of the survey. We emphasize that eventual Web respondents would likely have
responded even in the absence of this mode. In general, when gauging the effect of
treatments, such as policy-related treatments, a control group is usually required,
regardless of whether the design is experimental or quasi-experimental. However, in
some cases, a control group cannot be assigned into or observed.We show that the
effectof adding an alternative option can be estimated without a control group,under
certain assumptions. We use this approach to estimate increase in unit-response due to
adding the Web mode.The result shows that the options “Web response” and “paper
response” were mutually easily substitutable, and, only few respondents would not have
responded unless Web mode was available. In short, the net rise in unit-response
introduced by Web mode was only limited. Inaddition, in this paper, we show the
characteristics of those who had PC and Internet access, which enabled online
responses, and those who actually responded via Web mode.