Glossary · 7 · Evaluation methods
Think-aloud protocol
Also known as: Thinking aloud, Think-aloud method, Concurrent think-aloud, Retrospective think-aloud
German: Methode des lauten Denkens
In cognitive psychology and usability research, the think-aloud protocol is a method in which participants verbalize their thoughts while performing a task (concurrent think-aloud) or while reviewing a recording of it afterward (retrospective think-aloud), and the verbal reports are recorded and analyzed as data on what they attend to, how they interpret information and where they have difficulties.
- Applied cognitive psychology
- Evaluation methods
In one sentence
In a think-aloud protocol, participants say what they think while doing a task, or afterward, revealing what they notice, understand and misread.
Example
A technician installs a sensor using the draft instructions and says aloud what she does and thinks. At step 3 she says “I assume this means the left connector”, which shows that the figure is ambiguous; the writer adds the connector label.
Rests on
- Working memory: Ericsson and Simon argued that people can verbalize the contents of working memory that are already in verbal form without changing the course of their thinking much, whereas asking them to explain or justify does change it.
- Metacognition: asking for reflection adds a different, metacognitive task.
How it applies
- Testing instructions: Representative users perform real tasks with the instructions and the product while thinking aloud; the moderator only reminds them to keep talking.
- Concurrent vs. retrospective: Concurrent think-aloud gives immediate data but can slow the task; retrospective think-aloud, cued by a recording, leaves the task undisturbed but depends on memory. Van den Haak, de Jong and Schellens (2003) compared both in a usability test and found that they reveal partly different problems.
- Machinery and software documentation: Use it in usability tests of procedures, troubleshooting topics, HMI texts and search; record where readers hesitate, misread or skip.
Limits
- Thinking aloud is unnatural and harder for some participants and tasks (high workload, speech-based tasks, second-language speakers); it is unsuitable where talking distracts from a hazardous task.
- Verbal reports show what participants are aware of, not all processes; automatic, skilled actions are hardly verbalized.
- Tests at a real machine must be planned so that participants are not exposed to hazards.