Practical explanations for difficult research tasks.

GUIDE 01

How to identify a research gap.

Begin by defining the problem narrowly enough that relevant evidence can be compared. Search multiple terms and information sources, then organise studies by context, population, design, theory and findings. A gap becomes visible when evidence is insufficient, inconsistent or systematically limited in a way that matters to the research question.

Avoid manufacturing novelty by excluding inconvenient studies or by equating “not conducted in my district” with “not known.” The more defensible question is: What important uncertainty remains after the strongest available evidence is considered?

GUIDE 02

How to write a problem statement.

A problem statement should move from the expected condition to the observed condition, then identify the unresolved issue and explain why resolving it matters. Evidence should establish that the practical problem exists, while the literature should establish what is not yet adequately understood.

GUIDE 03

How to design a questionnaire.

Start with objectives, define constructs, identify indicators, select an appropriate response format and then write items. Pilot testing should be used to detect ambiguous wording, missing response options and questions that participants interpret differently from the researcher’s intention.

Double-barrelled, leading and vague questions should be revised because measurement error begins before statistical analysis. A sophisticated analysis cannot repair a poorly measured construct.

GUIDE 04

How to interpret thematic analysis.

Thematic analysis should move beyond topic labels. A theme should capture a patterned meaning that helps answer the research question. Researchers should explain how codes were combined, why a theme was retained, which data contradict it and how the interpretation relates to existing literature or theory.

GUIDE 05

How to interpret a statistically significant result.

Report the estimate, uncertainty and practical meaning rather than the p-value alone. Ask how large the effect is, how precise the estimate is, whether assumptions were met, whether bias is plausible and whether the observed relationship matters in the real-world context of the study.