The landscape of research methods
- Two axes organize almost everything: qualitative ↔ quantitative, and attitudinal (what people say) ↔ behavioral (what people do).
- The method follows the question — never the other way around.
- Say/do gaps are normal human behavior, which is why attitudinal data must be triangulated with behavioral data.
Every research method answers a particular kind of question, and most research failures start with a method chosen before the question was written.
Christian Rohrer’s map (published by NN/g) arranges methods along two axes.
Qualitative ↔ quantitative. Qualitative methods (interviews, field studies, moderated tests) answer why and how — direct observation, small samples, rich context. Quantitative methods (analytics, surveys, A/B tests) answer how many and how much — measurement at scale.
Attitudinal ↔ behavioral. Attitudinal methods capture what people say (interviews, surveys); behavioral methods capture what people do (analytics, eyetracking, usability tasks). The gap between the two is not lying — people are simply poor reporters of their own behavior. Plan for the gap: pair a survey with analytics, an interview with a task.
A third dimension is context of use: natural (field study, analytics), scripted (usability test), or not using the product at all (interview about the domain).
Choosing
Write the research question first, then pick the cheapest method that answers it:
- “Why do users churn in week two?” → qualitative + a mix of attitudinal and behavioral → interviews with churned users, session recordings.
- “Which onboarding converts better?” → quantitative, behavioral → A/B test.
- “Can people find pricing in the new nav?” → behavioral, scripted → tree test or usability test.
Try it: take three questions your team argued about recently and place each on the two axes. The method usually names itself.