34 – Mode Effects
Survey Treatments and Response Modes: Bayesian Survival Analysis with Competing Risks
Hiroaki Minato
U.S. Energy Information Administration
As part of research on Web survey method for Residential Energy Consumption Survey (RECS), the U.S. Energy Information Administration (EIA) sponsored the 2014 students' survey practicum in the University of Maryland's Joint Program in Survey Methodology (JPSM). Under a self-administered survey called the Home Energy Use Survey (HEUS), the practicum conducted several experiments on participatory incentives and questionnaire instruments. Here, we integrate both experimental treatments and contact procedures as survey treatments. A survey treatment is a composite treatment, for example, consisting only of advance letter and first invitation letter for a Web survey. A composite treatment takes place over time and composite treatments build up over time, depending on when responses come in, if at all, and what modes (Web or paper) the respondents choose. Possible response modes also depend on a survey treatment-in the example of survey treatment above, a response by the Web form is the only possible response mode. In this paper, we examine the relationships between survey treatments and response modes in terms of counts and rates of response over survey time period as well as over survey treatments.