The presentation and visual design of the don't know category in rating scales on respondents' behavior may have ambivalent effects. The hypothesis is that attitude strength toward the topic influences the respondents' sensitivity to the graphical design of the scale. A paper-and-pencil questionnaire conducted among 307 German university students contained an experimental variation of the presentation of a don't know category and its visual design. A multinomial logistic regression shows that presenting a don't know category drives respondents toward extreme categories. The visual design of the don't know category influences the distribution of responses, all the more that the respondents' attitudes toward the item topic are weak.