We are rarely interested in respondent characteristics that can be defined by only two states (e.g. gender), as most viewpoints and opinions can only be articulated by a continuum of choices.
Therefore we do not ask "Do you like XXX? (YES/NO)" but instead present the respondent with the question "How much do you like XXX on a scale from 1 (I don't like it at all) to 5 (I like it very much)?".
In theory, it is cognitively somewhat easier to answer a binary than a scale question, and it is also easier to construct a scale by combining binary rather than scale questions. However, by using scale questions, to begin with, we gain considerably more information (with the same amount of respondents' time and effort). We furthermore gain the possibility of calculating average values, performing dramatically more accurate comparisons between groups, testing hypotheses, performing multivariate analyses, etc.
Scale questions are, of course, presented with the single-answer (radio button) question type and not the multiple-answer (checkbox) question type.