Nothing Matters: Design Testing for Reporting Zero in a Web Survey

When survey respondents have nothing to report for questions asking for a quantity, they should report a lack of quantity instead of skipping the question. Reports of zero can be collected by asking respondents to check a None box. However, the effectiveness of this design has not been supported and, with the growing prevalence of web surveys, new design options are available. This study on data quality and user experience tested a web survey and the impact of three designs for reporting zero: a None box, instructions to enter "0" if respondents had nothing to report, and a Yes/No filter question preceding a write-in quantity. We administered the online survey to an opt-in panel of respondents in the U.S. Respondents were randomly assigned one of the three designs. They received a four-question survey with expected high rates of reporting zero. After controlling for demographic variables, we found that the None Box design resulted in significantly more item nonresponse than the Enter 0 design and nominally more than the Filter Question design. Additionally, the None Box design was the least efficient as it took more time on average to answer, and respondents who received that design rated the survey as less easy to answer. These findings align with previous literature that the None Box design is problematic. This research supports the use of alternative question designs in web surveys such as entering "0" or using filter questions when respondents may not have anything to report.