Research methods vary across attitude–behavior, qualitative–quantitative, and natural–controlled dimensions

What it is

A three-dimensional UX research method space locates methods along independent dimensions: stated attitude versus observable behavior, meaning-preserving qualitative analysis versus numerical analysis, and natural use versus researcher-controlled context. It is not eight rigid boxes. The same named method moves with implementation: diaries contain self-report and behavioral episodes, while remote usability tests range from scripted laboratories to participants’ own devices and routines.

Why it happens

Each dimension trades different information. Attitudes preserve evaluation and meaning but depend on recall and expression; behavior records occurrence without directly revealing motive. Quantification enables comparison by compressing context, while qualitative analysis retains context without estimating a population. Control isolates variables; natural settings preserve authentic goals and constraints. Separating dimensions reveals which gap a method fills and which blind spot it introduces.

Studying it

For each candidate, specify evidence source, analysis form, and degree of contextual control from the actual procedure rather than the method label. Match information needs and acceptable loss on each dimension. Estimating real-world failure frequency calls for behavioral, quantitative, relatively natural evidence; understanding its process may add qualitative observation or controlled tasks. Combination addresses named gaps rather than mechanically balancing every axis.

Where it stops holding

The space is a planning heuristic, not a complete epistemology. It omits longitudinal design, participation, generation versus confirmation, and researcher power. Attitude and behavior are not a mental–physical binary: think-aloud speech is observable action that also expresses evaluation. Naturalism and control are not necessarily inverses either: field experiments can strongly manipulate conditions in everyday settings, while laboratory studies can preserve open-ended tasks. Sampling, measurement, and inference determine quality; coordinates do not rank methods.

Applying it

• Plot each proposed method and name its primary blind spot.
• Check whether a high-consequence claim rests on evidence from only one side; add a complementary source when needed.
• State what each strand answers and how they connect rather than calling any multi-method design triangulation.
• Compare intended and actual positions after fieldwork; revise claims when a nominally natural task became tightly scripted.

Related

• Same group: Q1.10.2 Stated attitudes and observed behavior can diverge · Q1.10.3 One question may require choices across all three dimensions · Q1.10.4 A qualitative–quantitative binary hides context
• Adjacent: Q1.03 Choosing qualitative and quantitative methods · Q3.11 Telemetry and instrumentation
• Search terms: attitudinal behavioral research · qualitative quantitative · context of use