One of the major benefits of Web survey is the ability to enhance (or even replace) survey questions with a variety of visual stimuli (pictures, drawings, etc.). In addition, the rich visual medium of the Web permits the survey designer to vary color, design and layout of the survey instrument at will. However, these enhancements may come at a cost in terms of measurement error. This paper will present the results of several experiments examining the effects of visual enhancements on the answers provided in Web surveys and discuss the implications for survey design and measurement error.