Consider what your research question is trying actually investigate.
Descriptive questions are often trying to provide a singular analysis of a condition. For example:
What are students’ perceptions of group quizzes in an introductory CS course?
This is a targetted question that isn’t trying to say that some treatment A is better than treatment B. It’s trying to ascertain the state of the world as it currently is.
Questions in this category are the more in line with a study where a [thing] is applied/given to a group of students and not given to a separate control group. Something like:
What differences, if any, exist in student attitudes regarding computer science between students’ who use X homework system and those that do not?
These questions are trying to determine if the [thing] had a measurable affect on the treatment group by comparing some set of data from this group and the control group.
Relational questions are trying to determine “if A, then B”-type relationships. For instance:
What relationship, if any, exist between students attendance in office hours and test scores for an introductory CS course?
A question such as this is looking at whether office hour attendance had an affect on test scores.