Description
Survey respondents draw how a system works using a fixed icon set; drawings are coded by which icons appear and clustered into mental model classes linked to behaviour.
Drawing tasks reveal mental models but are hard to analyse at scale. This method constrains the drawing with a fixed palette of icons for the system's components, so a web survey can collect hundreds of drawings. In the source study, 772 respondents drew a communication connection in working and failed states using 48 icons. Non-serious drawings were excluded, each drawing was coded by which icon types it contained rather than how many, rarely used icons were dropped, and a two-step cluster analysis gave four mental model classes. These were related to ICT literacy (scored by multiple correspondence analysis), satisfaction with waiting times, chosen behaviour during failures and assumed causes (latent profile analysis). Mental model class varied with literacy and predicted different behaviour when the network failed. The icon set limits what can be expressed, and presence coding ignores structure such as arrows and arrangement.
Attributes
- Discipline
- Survey & Psychometrics
- Loss type
- Latent
- Stage
- Discovery, Definition
- Data
- Mixed
- Effort
- Weeks
Steps
- Design a fixed icon palette covering the system's components as users might imagine them, plus generic elements.
- Ask respondents to draw the system in a normal state and in a failure state using the palette.
- Exclude non-serious or empty drawings.
- Code each drawing by which icon types it contains, and drop icons that almost nobody used.
- Cluster drawings (for example two-step or latent class clustering) into mental model classes.
- Relate classes to literacy, satisfaction, troubleshooting behaviour and assumed causes, and illustrate each class with typical drawings.
Sources
- Human Interface and the Management of Information (Yamamoto & Mori, 2020)
Also known as
- Icon-based mental model drawing survey with cluster classification
- Mental model drawing task
Categories
- Analytics & Behaviour
- Survey & Psychometrics