Recently, in the field of consumer decision-making, there have been numerous reports of preference reversal phenomena, wherein consumer preferences differ depending on two modes of product evaluation: joint evaluation and separate evaluation. The present researchers hypothesized that consumer knowledge would influence the preference reversal phenomenon, and tested this experimentally. In Study 1, participants were presented with products composed of two attributes—one that is difficult to evaluate independently and one that is easy to evaluate. According to the existing preference reversal literature, in separate evaluation where products are evaluated individually, products are assessed based on easy-to-evaluate attributes, but in joint evaluation, the relative influence of hard-to-evaluate attributes increases, resulting in different preferences across the two evaluation methods. The experimental results showed that preference reversal did not occur in the high consumer knowledge group. In Study 2, we examined whether the preferences of the high-knowledge group are stable regardless of context. Participants were presented with a highly superior product, and we tested how evaluations of this product differed between joint and separate evaluation. The experimental results revealed that a product rated as highly superior in separate evaluation was rated as less superior in joint evaluation. This pattern was observed in the opposite direction when a highly inferior product was presented. Through the two studies, it was found that while preference reversal does not occur in the high consumer knowledge group when evaluation contexts differ, changes in product evaluation do occur. That is, it was confirmed that even consumers with extensive product knowledge do not make stable evaluations regardless of context.