Description
Code every action users take per required task step from screen recordings, categorise by intent and locus, and compare action patterns between interfaces.
Time and accuracy show which interface wins but not why. This method decomposes each task into the steps it requires on each interface, then codes every action users took within each step from screen recordings, counting each action type once per step. Codes emerge as actions appear and are grouped by intent (for example navigate, verify, select) and by locus: whether the user acted on separate controls or directly on the data. Analysis looks at the range and frequency of actions, adoption of features over the task sequence, and choices among interchangeable actions, interpreted alongside time and accuracy. In the source study of a node-link visualisation versus a table tool across nineteen tasks, it exposed learnability, overlooked features and extra verification effort.
Attributes
- Discipline
- Usability & Evaluation
- Loss type
- Latent
- Stage
- Evaluation
- Data
- Mixed
- Effort
- Weeks
Steps
- Record screens during a comparative task study.
- Decompose each task into the steps it requires on each interface.
- Code every action users took within each step, counting each action type once per step, adding codes as new actions appear.
- Group codes into categories by intent and by locus of interaction (controls versus direct manipulation of data).
- Analyse range and frequency of actions, feature adoption over the task sequence and choices among interchangeable actions.
- Interpret alongside time and accuracy data, and have a second coder check a sample for consistency.
Sources
- Human Interface and the Management of Information (Yamamoto & Mori, 2020)
Also known as
- Screen-capture action coding for interface comparison
- Task-step action coding of screen recordings
Categories
- User Research
- Usability & Evaluation