3 Design & Measurement

Many avoidable problems start here. Decide what you'll measure, how, when, and against what, before anything else is locked in.

Before you move on

  • Each outcome is valid for your question, feasible to collect, and relevant to your stakeholders.
  • Each outcome can realistically change within your data collection window.
  • You have outcome and process measures, and a balancing measure if your change could make something else worse.
  • You'll use the same instrument, the same way, at every time point, and you can link responses if your design needs it.
  • Your expected sample gives you a realistic chance to see the size of change you care about.

Measurement decisions are the hardest to undo. Once data collection starts, you usually can’t add a baseline you didn’t collect, change a survey without breaking comparability, or link responses you collected anonymously.

If you do only one thing at this stage, write down each outcome with how it will be measured, when, by whom, and what you’ll compare it against. Then ask your faculty advisor to review that list.

Guides for this stage