Recently, non-task-oriented dialogue robots that aim at communication itself have been attracting attention. These robots are required to motivate users to interact with them over a long period of time. On the other hand, the topics that can be presented by dialogue robots range from shallow and superficial topics to deep and private topics. Although a negative correlation between topic depth and dialogue motivation has been confirmed for human-human interaction, it is not clear whether this correlation applies to human-robot interaction as well. Thus, in this study, we aim to create a model of the relationship between topic depth and dialogue motivation with the robot. We conducted an online survey with 90 participants and analyzed the relationship between two variables using a generalized linear mixed model. The result showed that the deeper the topic, the more the dialogue motivation with the robot significantly decreased.