Encouraging Record Use for Financial Asset Questions in Aweb Survey.

We report on two experiments to encourage record use by respondents in an Internet survey. The experiments were conducted in the 2009 Health and Retirement Study (HRS) Internet Survey, administered to those in the HRS panel with Internet access, and in the 2011 HRS Internet Survey. Encouraging respondents to consult records at the relevant point in the questionnaire significantly increased reported record use (from 39 percent to 47 percent), but was insufficient to produce significant changes in the precision (amount of rounding) of the information reported. Including the encouragement in the mailed invitation to the Web survey in 2011 resulted in a lower response rate (77 percent with encouragement, 80 percent without), but increased reported record use among respondents (from 46 percent to 55 percent). In neither case was the increase in reported record use large enough to produce significant differences in the precision of the information reported between the groups with and without encouragement.