Concept mapping in research: from needs analysis to design

2026-07-22

Group Concept Mapping is often used as a practical aid for a team. It is that too, but the method comes from the evaluation research tradition and has been in use there since the 1980s. For anyone who has to justify a research design, that is relevant.

What it is methodologically

Concept mapping is a structured mixed-methods approach. The input is qualitative: statements in participants' own words. The processing is quantitative: multidimensional scaling on a dissimilarity matrix, followed by hierarchical clustering.

That combination is exactly why it is usable in research. You do not lose the richness of open input, but you gain testable structure and reliability measures alongside it.

Where it fits in a research design

Needs analysis up front. Before you design an intervention, you want to know what those involved see as the task. Concept mapping provides a structured answer that has not been pre-cooked by the researcher.

Construct development. If you want to build a questionnaire for an area where no instrument exists yet, the map gives you the dimensions and the items in the language of the field.

Conceptual frameworks (conjecture mapping). For design-based research you have to make your assumptions explicit. The themes and the causal relations participants indicate form an empirically grounded basis for that.

Evaluation and baseline. Because the set-up is repeatable, you can issue the same focus prompt again later and compare the scores. That makes the map usable as a baseline for a longer project.

The appeal for research lies in repeatability: the same prompt, the same procedure, comparable outcomes.

What you have to be able to justify

A reviewer or supervisor will ask about three things.

• Reliability. Split-half reliability shows whether the map depends on who happened to take part.
• Fit. The stress value of the scaling indicates how well the two-dimensional display represents the underlying distances.
• Choice of the number of clusters. That choice is substantive, not automatic, and therefore has to be argued.

All three belong in a report, with the raw data alongside, so that someone can redo the analysis.

For Erasmus+ and similar programmes

International projects add a practical point: participants contribute input in their own language, and reporting usually has to be in English. That is no problem as long as the translation step comes after the analysis and not before, because otherwise you translate away the differences in meaning you set out to measure.

What a researcher can do in the environment is set out on the researcher page.

← All articles