This study proposed a game-based flipped teaching approach and applied it to a HTML (HyperText Markup Language) course. We developed two versions of the pre-class content testing, one of which was game-based, using a "looking-through" game, and the other which was traditional, using a multiple-choice test. We conducted a teaching experiment of 9 weeks with 3 h per week. The valid sample was 39 students in the experimental group using the game-based version, and 32 in the control group using the traditional version. We hoped that the students' learning effectiveness could be increased. However, adding the game may have different impacts on different human factors and different cognitive processes of Bloom's taxonomy. To more precisely understand these problems, we divided the students into high and low prior knowledge, and divided learning effectiveness into the higher-level applying and analyzing aspects and the lower-level remembering and understanding aspects. The results showed that learners with low prior knowledge who originally lacked the foundational knowledge became interested in concentrating and then in the higher-level cognitive learning processes because of the experience of immersing themselves in the flow of the game. The learning effectiveness of their classroom participation significantly improved.