Contextual inquiry

Contextual inquiry can be loosely defined as an interview that takes advantage of the user's work environment to trigger and organize user responses.

Properties

• Issues such as defining the audience and recruiting are the same as other types of interviews
• User (interviewee) has more control of the interview than a conventional interview; the researcher provides a framework
• Environment with artefacts cues responses
• Setting allows user to walk through tasks

Specific strategies

• Systematically reviewing items in an application, Web site or work environment
• Master/apprentice model
• Task walkthroughs
• Using a list of systematic questions for each artefact/goal/task

Note that Kuniavsky description of contextual inquiry borders upon basic user observation.

What to note

This list is adapted from two lists in Kuniavsky (pp. 171 - 172 and pp. 175 - 176):

• Tools
• Methods and action sequences
• Kinds of interactions
• Mental models
• Terminology
• Goals and tasks
• Values

The list of attributes for user profiling could easily apply here too.

Analysis

Contextual inquiry is a data gathing method. Collected results can be analysed using standard methods (e.g. task analysis, user profiling), but there are particular analysis methods often associated with contextual inquiry.

Contextual design (Beyer and Holtzblatt)

1. Assemble approximately 100 notes
2. Cluster notes
3. Use color-coded notes to indicate and label different levels in a hierchy

Top-down method

• Organize analysis by tasks
• Note critical elements for each task (e.g. frequency, duration, tools, criticality, obstacles)

I used this method for a study invovling the CTI intranet (notes from HCI 440 course).

Reporting results

The principal goal of user research is to identify unmet needs and present useful information for finding their solutions.

Kuniavsky lists models by Beyer and Holtzblatt:

• Flow models (information flow)
• Sequence models (actions for tasks)
• Artefact models
• Physical models (environment)
• Cultural models

Results can also be reported in the form of task analysis diagrams and user profiles.

Items for discussion

• Kuniavsky says that ideally CI should be the first step in the development cycle (before specifying problem statements, technologies, solutions, etc). Does this happen in practice? Why not?
• Differences between conventional interview, site observation and contextual inquiry; consider costs, type of information, validity