Research

Choose a research design you can defend

Branch your question into candidate designs, weigh the validity threats of each, and keep the designs you rejected for your limitations section.

See it in action

One research question tested against six methods, so you can see which design answers it.

Open the full map

How it works

Enter your question and your constraints, and the model maps the designs that fit. Each design gets its own branch, and every question under it keeps your question and constraints in context.

  1. State the question and constraints. Include your budget, timeline and access to participants. Ask which kinds of design fit: descriptive, correlational, causal or exploratory.

  2. Branch into specific designs. Under a causal approach, ask for options such as randomized experiments, natural experiments, regression discontinuity and instrumental variables. Ask for each one’s assumptions and when they break.

  3. Weigh the validity threats. Under each design, ask about internal, external, construct and statistical conclusion validity. Note which threats the design handles and which it leaves open.

  4. Choose and record why. Pick a design and write your reasons on its branch. The rejected branches become your limitations and alternative approaches sections.

Why a map instead of a methods draft

A change to one part of a design spreads to the rest. Moving from random to convenience sampling changes your analysis plan, your claims about generalizing, and the sample size you need. Ask about the change under the sampling branch, and those effects sit together in one place.

Example

Say you want to know whether a new teaching method improves critical thinking in undergraduates. A randomized experiment means assigning students to sections at random, which your university may not allow. A pre-post design with a comparison group is feasible but brings selection threats. A branch on propensity score matching shows you need baseline measures you had not planned to collect, and you add them before data collection starts.

Now try it yourself

Compare methodological approaches to study the impact of remote work on team innovation
Three methodological families apply: quantitative approaches using patent or product output metrics with difference-in-differences designs, qualitative case studies using ethnographic observation of team interactions, and mixed methods combining survey instruments with semi-structured interviews.
What are the validity threats for the quantitative approach?
Selection bias is the primary threat — teams that went remote may differ systematically from those that didn't. Measurement validity is another concern: patent counts capture only codifiable innovation, missing process innovations and incremental improvements.

Ready to try research methodology?