9. Experimental & Interventional Research

RCTs and the wicked-problem mismatch

The randomized controlled trial is the dominant design for evaluating interventions. It is also the design most often misapplied to wicked problems. Understanding both its strengths and its limits is the start of a transdisciplinary frame for interventional research.

RCT strengths:

• Strong internal validity for well-specified interventions
• Clear comparator and causal inference under assumptions
• Standardized reporting (CONSORT) for transparency

RCT limits in wicked-problem contexts:

• Treats the intervention as discrete and manipulable; community interventions often aren't
• Treats success criteria as predetermined; community success criteria may emerge during the work
• Treats the unit as the individual; the unit is often the network or the system
• Treats variation as noise; in complex systems, variation is often signal

The transdisciplinary move is not to abandon RCTs but to use them where they fit, and use alternative designs (quasi-experiments, stepped-wedge, multiple-baseline single-case, realist evaluation, developmental evaluation) where the standard RCT can't capture what matters.

Applying design thinking to study design

Design thinking — empathize, define, ideate, prototype, test — was developed for product design but adapts well to research design. The transdisciplinary version:

• Empathize — what are the actual constraints, fears, and aspirations of the people the research is for? Spend time before you write the protocol.
• Define — articulate the problem statement in participants' language, not academic language.
• Ideate — generate multiple intervention candidates without prematurely committing.
• Prototype — build the cheapest possible version of the leading candidate that can fail informatively.
• Test — run the prototype with real users, designed to surface failure modes.

A common failure: skipping empathy and define, going straight to ideate. The result is interventions that meet academic-team intuitions and miss participant constraints.

Empathy as a methodological act

"Empathy" in research design isn't a feeling; it's a methodological act. Concretely:

• Sit with potential participants in their setting, not in your office or clinic
• Ask what makes the problem hard for them, before proposing how to fix it
• Listen for words you wouldn't have used and use them in your protocol
• Identify constraints you didn't know about (work schedules, transportation, caregiving, technology access)
• Identify aspirations beyond your outcome variable (relational goals, identity, control)

Skipping empathy doesn't save time; it shifts the cost to recruitment, retention, and translation. The empathy step is rigor, not preliminary courtesy.

Defining the problem in participants' voice

The define step is the single most underdone step in research design. The exercise: write the problem statement, then read it aloud to a participant. If they don't recognize their own concern in it, rewrite.

A useful test: the problem statement should not contain academic theory terms unless those terms are also used by participants. "Adherence" might be your word; "remembering" or "fitting it in" might be theirs. Use both, but make sure theirs appears.

Prototyping in research

A prototype is a deliberately incomplete early version of an idea, designed to fail informatively at low cost. A pilot is a feasibility test of a near-final design. They serve different purposes.

Skipping prototyping wastes intervention dollars. A research prototype might be:

• A one-session intervention with three participants, observed by the design team
• A draft of an educational tool, walked through with two practitioners
• A simulated workflow run in a single clinic for a single morning
• A read-aloud of the consent form with three potential participants

The prototype is meant to be discarded. The information you get is what changes.

Co-design with participants, not on them

Standard intervention research designs for participants. Transdisciplinary intervention research designs with them. The difference is decision rights.

Concretely, co-design involves participant or community partners holding decision rights on at least some design questions: eligibility criteria, intervention dose, delivery channel, primary outcome, dissemination strategy. Without shared decision rights, participant input is consultation, however respectfully done.

A practical structure: a community advisory board that meets monthly during design, has documented decision authority on specified questions, and includes paid community members rather than only volunteer advisors. The structural commitments are what make co-design real.

Iterative, human-centered intervention research

Putting it together, a transdisciplinary interventional study looks like this:

1. Empathy phase — pre-protocol, spent in the setting, listening
2. Definition phase — problem statement co-written with participants/community
3. Ideation phase — multiple intervention candidates generated jointly
4. Prototype phase — cheapest versions tested with small samples, with decision criteria for which to pursue
5. Pilot phase — feasibility study of the chosen design, including recruitment and retention
6. Main study phase — the formal RCT or alternative, run with continued community governance
7. Mobilization phase — translation and adoption, planned from the start

Each phase has documented deliverables and decision criteria. The total time investment is greater than a standard RCT; the probability of producing an intervention that scales is much higher.

When RCT is the right tool

RCTs are the right tool when:

• The intervention is discrete and manipulable
• The population is well-defined
• The outcome is measurable and agreed-upon
• The community is willing to be randomized
• The science is at the stage of testing, not exploring

When any of those conditions don't hold, consider alternative designs first.

Closing

Design thinking — empathize, define, ideate, prototype, test — applies to research design as much as to product design. Empathy and define are the steps most often skipped and most often missing. Prototypes are not pilots; they are cheap and discardable. Co-design requires structural decision rights, not just respectful consultation. RCTs are the right tool sometimes, not always.

Next: design thinking applied to the research workflow itself — iteration on instruments, recruitment, consent, and procedures.