Among the first decisions a Web survey designer must make are those that deter­mine the size and background color of the screen. Web survey designers under­standably want to work with the largest possible canvas, and, indeed, a substan­tial and growing majority of Web users work at 1024 by 768 resolution or above. Unfortunately, a significant minority of Web users seem to prefer a lower resolu­tion 800 by 600 screen. Surveys designed for 1024 by 768 will display differently at lower resolutions, sometimes resulting in the full question not being displayed without vertical or even horizontal scrolling. The use of "fluid” screen designs that expand or contract according to the available window size are one solution, but even these designs can result in potentially important display differences, espe­cially with horizontally arrayed response sets and long scales.

Background color may be equally problematic. Designers want to make the sur­vey more appealing and use the rich colors available on the Web to achieve this. Preliminary research (Pope and Baker, 2005) has suggested that variations in screen color can influence perceptions of survey length and even reporting of risk behaviors, at least among small samples of college students. Gorn and col­leagues (2004) have found that background color affects perception of download times. Whether these results replicate to the general population is unclear.

The research will systematically vary screen size in a questionnaire using a vari­ety of standard Web question types, including successive screens of wide grids. The experiment will also vary screen background color. Measures of interest will include break-off rates, actual completion times, perceived completion times, re­porting of sensitive behaviors, and differences in response to standard question types, especially those arrayed horizontally in grids. The goal of the research is to provide clearer guidance to Web survey designers about the response impacts of these key design issues.