Measuring mental models: Rationales and instruments
Yan Zhang, Peiling Wang
2006
Proceedings of the American Society for Information Science and Technology
The purpose of this study is to develop and test an instrument for measuring users' mental models of Web search engines. Mental models play an important role in users' interactions with a system. Understanding users' mental models is important to the design of congruent interfaces. Rationales of measuring mental models In the field of human-computer interaction (HCI), research on mental models has produced a body of literature in the past twenty years. Despite differences in perspectives and
more »
... minologies surrounding mental models, the core of the topic concerns the understanding of the cognitive structures and processes underlying the behaviors of human beings performing computer based tasks. It is clear that interaction with a computer is often a subtask during the completion of some main tasks. The main tasks may include finding information, sending a message, producing a report, and testing a statistical hypothesis. In fact, some tasks can be done without a computer (but generally more efficiently with a computer software tool). Knowing nothing about system structure and mechanisms beyond the system's display could be frustrating when interacting with a computer system. End users must possess a mental representation of the system before they feel comfortable with it. In this article, mental model refers to users' conceptual/internal representation of the system. Mental models are incomplete, limited, naive (unscientific), unstable, fuzzy, but vitally important. Mental models enable users to interact with and learn by trial and error about systems. Our goal for understanding mental models is three-fold: (1) systems should help users build appropriate mental models by providing meaningful and context-sensitive clues. (2) we need to design congruent interfaces that can anticipate users' next moves in order to provide adequate support. (3) we need to develop learning tools to help users move from novice to advanced levels. A theory on mental models needs to address three important questions: (1) what are the users' conceptual representations of a system? (2) How do users apply such representations in interactions with systems? (3) How do users derive mental models? The first question concerns the content and structure of the mental models: what concepts do the users know and how are they related? The second question concerns the use of mental models in taking actions during interactions to perform the main task. These actual actions can be attributed to the aspects of the mental models. The third question concerns the learning process through which users build and modify their mental models. We postulate: (1) Mental models are based on prior experiences with similar systems using analogy or metaphor. Users tend to use mental models of other systems in interactions with a computer system. (2) There are gaps between users' mental models and a system's conceptual model. A system has an underlying conceptual model that is often hidden from its users. (3) There are gaps between the system's conceptual model and its user views or external models. We use user views to mean what Norman (1983) referred to as "system image", which is presented to the users as the implementation of the conceptual model. There are often more than one user views targeting different user groups. Typically, one for novices and one for advanced users. In empirical studies of mental models, a majority adopt the strategy to measure users' mental models by measuring their task performances. Users with better mental models perform better in tasks. However, tasks assigned by the researchers may not measure all aspects of mental models. In this study, we propose an alternative approach to study mental models. We aim to develop an instrument to measure users' mental models of the Web. We are interested in the Web because it is the first-stop or the only place for many users in finding information. For more and more end users, this is the first IR system they learn. Their mental models of the Web will likely affect their behaviors when using other types of IR systems.
doi:10.1002/meet.14504201270
fatcat:lfwkywzrgfgv3hxbdonxrzvvdy