I would like to conclude this paper with two research problems to be studied further. First, we need to develop more good problems, especially open-ended type problems. As stated above, it is most difficult to construct open-ended type problems among the three types of open problems. There is not sufficient stock of such problems even in Japan, so we need to develop them because they are very valuable to mathematics education today. We examined the "marble problem" as the example of open-ended problem. In that problem, students are expected to mathematically make sense of the situation where marbles are scattered. This suggests that some open-ended problems can be related to mathematical modeling. It may be possible to get hints for developing good open-ended problem by referring to the research on mathematical modeling.