In order to improve the quality of communication networks,weadopted a research approach that realizes individual optimal control in consideration of the context and individual characteristics of the users, such as quality of experience (QoE) and co-created quality. In this paper, we report the results of a Web questionnaire survey of 722 participants to analyze the cognitive and behavioral characteristics under network malfunction. We analyzed the features of a mental model of users by the two-step cluster method. In the analysis, the data extracted by participants regarding the communication connection was regarded as a mental model. Characteristics of the mental model were examined by the element type (carrier, content providers, server, internet lines) included in the extracted data. The result showed the mental model of the users was classified into four classes with different characteristics. Analysis of the relationship with other attributes possessed by the users showed that there is a high possibility that the class of the mental model differs depending on the degree of ICT literacy. Moreover, it was found that the differences in mental models are likely to lead to differences in behaviors with network malfunction.
