We utilize and apply visual design theory to experimentally test ways to improve the likelihood that web respondents report date answers in a particular format desired by the researcher, thus reducing possible deleterious effects of error messages or requests for corrections. These experiments were embedded in a series of web surveys of random samples of university students. We seek to examine the sequential and cumulative effects of visually manipulating the size and proximity of the answer spaces, the use of symbols instead of words, the verbal language of the question stem, and the graphical location of the symbolic instruction. Our results show that the successive series of visual language manipulations improve respondents' use of the desired format (two digits for the month and four digits for the year) from 45 percent to 96 percent. These results suggest that writing effective questions for web surveys may depend as much or more on the presentation of the answer categories/spaces as the question wording itself.