Define the problem
Describe the real-world issue, who it affects, what is already observed, and why the unanswered part matters.
Quality check: Keep the problem specific enough to investigate—not a whole field of study.RESEARCH GUIDE / 01
Build a study that can answer the question.
Study design is the structure connecting a research question to credible evidence. A strong design makes the population, measurements, comparisons, timing, ethics, and analysis clear before data collection begins.
Core questionWhat design can answer this question with the least avoidable bias?
Best forProposals and protocols
What you should leave withA defensible study blueprint
Follow the reasoning
How the pieces fit
Start in order, but do not treat research as a straight line. Later evidence often sends you back to fix an earlier decision—and that is part of doing the work properly.
Write the research question
Name the population, main exposure or intervention, comparison where relevant, outcome, place, and time period.
Quality check: A focused question makes the literature search, variables, and analysis easier to align.Set objectives and hypotheses
Use one overall objective and specific objectives that each lead to a measurable result. Add hypotheses when the study will test relationships or differences.
Quality check: Avoid objectives that introduce concepts not present in the main question.Choose the design
Select a cross-sectional, case-control, cohort, experimental, qualitative, mixed-methods, or other design based on the question—not convenience alone.
Quality check: Descriptive questions, causal questions, and lived-experience questions require different designs.Plan the population and sampling
Define eligibility, sampling frame, sampling method, sample size reasoning, recruitment, and how non-response will be handled.
Quality check: The people you can reach are not automatically the population you want to understand.Pre-plan measurement and analysis
Define every key variable, choose valid measurement tools, identify likely confounders, and map each objective to an analysis before collecting data.
Quality check: Include ethics, data protection, bias control, limitations, and a realistic timeline.See it in context
Worked example: phone use and sleep
Research questionAmong university students in Kenya, is late-night smartphone use associated with poorer sleep quality during the academic semester?
An analytical cross-sectional survey could estimate the prevalence of late-night use and test its association with a measured sleep-quality score at one point in time.
Population
Registered university students meeting stated eligibility criteria
Exposure
Frequency or duration of smartphone use after a defined night-time hour
Outcome
Sleep quality measured with a stated, appropriate instrument
Other variables
Age, course workload, caffeine use, living arrangement, and other justified factors
Analysis
Descriptive summaries followed by an appropriate adjusted regression model
Limitation
A cross-sectional design cannot establish which behaviour came first
Before you move on
Questions your work should answer
✓
The question names a population and a measurable outcome.
01✓
The design matches the type of question being asked.
02✓
Eligibility, sampling, and sample-size reasoning are explicit.
03✓
Every objective maps to variables and a planned analysis.
04✓
Bias, confounding, ethics, privacy, and limitations are addressed.
05A sophisticated analysis cannot rescue a design that collected the wrong evidence.
Follow the next decision
Keep the research story connected.
A good question should shape the literature search; the evidence should sharpen the design; and the methods should make the result possible to understand and challenge.