1. Define a question you can answer
Write a main question that identifies the phenomenon, population, context, and the description or relationship of interest. Then ask what evidence would answer it. If the evidence remains unclear, refine the question before choosing an instrument.
Consider an illustrative teacher-training study. “How frequently do teachers use formative assessment?” asks for a description. “Is training participation associated with its use?” asks about a relationship. A causal version requires further design decisions; changing the wording of a title does not provide those decisions.
2. Connect each question to an objective and evidence
Prepare a short alignment matrix before expanding the proposal. For each question, identify its objective, concept or variable, data source, and proposed analysis. The matrix helps expose questions the instrument cannot answer and information the project does not need to collect.
| Illustrative question | Evidence needed | Decision to resolve |
|---|---|---|
| What is the level of practice? | Defined indicators of the practice | How will the level be defined and measured? |
| Are two variables associated? | Both variables measured on the relevant units | Which measurement and analysis fit? |
| How do participants interpret an experience? | Accounts or interviews suited to the question | How will participants and material be selected and analysed? |
3. Describe the design beyond its label
A label such as “descriptive analytical approach” does not explain how the study works. Specify when data are collected, where they come from, whether units are followed over time, whether there is an intervention or comparison, and how procedures will be implemented.
The University of Southern California guides explain research designs and the role of collection, measurement, and interpretation. Use those distinctions to justify decisions alongside your disciplinary literature and institutional requirements. In a cross-sectional study, keep conclusions aligned with the relationships observed during the study period. A correlation alone does not establish that one variable caused another.
4. Define the population, sample, and unit of analysis
Data may come from individuals while the question concerns schools or teams. State the unit of analysis explicitly, then define the target population, the frame used to reach it, eligibility criteria, and the sampling procedure. Easy access to respondents is not evidence of population representativeness.
Document how you will respond to lower participation or missing responses. Sample-size justification depends on the objective, design, and intended analysis. A number borrowed from another study is not a substitute for a justification relevant to your own project.
5. Treat the instrument as part of the design
Connect each dimension to the definition you adopted and each item to the information you need. Review wording, response direction, score construction, and completion time. For an existing measure, check its original source, conditions of use, and suitability for your context.
Plan an appropriate preliminary trial and record resulting decisions. Distinguish wording review, evidence of validity, and measurement consistency: these address different aspects of an instrument. Explain what you assessed rather than treating one statistic as evidence of every aspect of quality.
6. Prepare for a methodology consultation
Start with a one-page decision brief. It helps focus the discussion and prioritise changes by their effect on the study design.
- The working title, main question, and subquestions.
- An alignment matrix of questions, objectives, and concepts or variables.
- The proposed population, sample, and route to the data.
- Your instrument, its source, and relevant institutional guidance.
- A list of unresolved decisions and your current stage of work.
Sources & further reading
Would you like to discuss your study?
Explore the support related to this guide, or share a short description of your needs to discuss the next step.