Method choice

Open, closed, or hybrid: choose the question before the controls.

Short answer: use an open sort to explore participant-created groupings and labels, a closed sort to test placement into supplied categories, and a hybrid sort when supplied categories matter but participants need a visible escape route for poor fits.

Method	Participant can create groups?	Core evidence	Typical bounded question
Open	Yes, all groups	Pair similarity and raw label frequency	How do people naturally group and name these items?
Closed	No	Placement counts/share in predefined categories	Where do people expect each item within this proposed set?
Hybrid	Yes, alongside supplied groups	Placement, pair similarity, and created labels	How does the proposed set work, and where does it fail to fit?

Open sort

The participant begins with an unsorted tray and creates every group. This can expose vocabulary and conceptual patterns that a predefined taxonomy would suppress. Analysis compares card pairs and preserves raw group labels. The researcher should still avoid treating frequently repeated words as automatically correct navigation labels.

Closed sort

The researcher supplies every group. This makes placement counts directly interpretable and can test an existing or proposed category set. It also constrains the answer space. A participant may place an item in the least-wrong group, so high completion is not the same as confidence or success.

Hybrid sort

The researcher supplies groups but allows participant-created groups. Hybrid can expose cards that do not fit the proposed structure and language participants prefer. It also creates mixed evidence: predefined placement and open labels need to be interpreted together rather than collapsed into one score.

Completion controls change behavior

Requiring every card or every created-group label may improve data completeness. It can also increase effort, encourage arbitrary placement, or increase abandonment. Pilot both the rule and the wording. The field lab shows completion settings and category capacities explicitly because they are part of the issued edition, not administrative trivia.

Do not choose by chart preference. A dendrogram does not make an open sort appropriate, and a placement matrix does not make a closed sort appropriate. Choose the method that answers the research question, then use only analysis consistent with that method.

Published context

See Nielsen Norman Group's card-sorting overview, Optimal Workshop's method comparison, and the field lab's exact method page. These are contextual sources, not universal rules.