Web surveys have introduced a number of changes to the survey data collecting process. Significantly more possibilities in forming the questionnaires have been presented to the researches. In this context, graphic elements can improve user experience, as well as cause unforeseen changes in the understanding of the issue. The role of graphic elements in the evaluation variables on the ordinal scales, which are extremely common in social sciences and marketing research, is particularly relevant. The goal of this thesis is to evaluate the effects of graphic elements in such scales. We have focused on the role of graphic elements in measurement of satisfaction in web surveys. We have examined relevant literature, putting a special emphasis on the cognitive aspects of filling survey questionnaires, as well as on the aspects of visual design of questions in web surveys. In the empirical part, we have conducted an experiment based on a web survey (n = 153), in which one group answered questions with evaluation variables on a five-level ordinal scale with the usual selection buttons, and the other group with graphic elements, for which we selected the usual smiley faces. The results have shown that the answers with the included graphic elements were more reliable. Due to a relatively small sample, other effects of the use of graphic elements have only been indicated: less pronounced agreement on the rating scales, lower final satisfaction with the overall questionnaire and bigger effects on the respondents with a low level of education.