From poorly defined research questions to misinterpreted statistical results, these are the most common pitfalls we see — and practical advice for avoiding them.
Introduction
Every year, thousands of students in Nepal embark on research projects for their bachelor's, master's, or PhD degrees. Despite their hard work and dedication, many stumble on the same avoidable mistakes. Having worked with hundreds of researchers across Nepal, our team has identified the ten most common errors — and, more importantly, how to fix them before they derail your work.
1. Vague or Unmeasurable Research Questions
The most foundational mistake is starting with a research question that is too broad or impossible to measure. Questions like "What is the health situation in Nepal?" cannot be studied in a single project. A well-formed research question should be Specific, Measurable, Achievable, Relevant, and Time-bound (SMART). Instead, try: "What is the prevalence of hypertension among adults aged 30–60 in Lalitpur district, 2023?" This gives you a clear population, outcome, setting, and time frame.
2. Skipping the Literature Review
Many students see the literature review as a formality rather than a scientific necessity. A thorough literature review tells you what is already known, what gaps exist, and which methods have worked in similar contexts. Skipping it leads to duplicated work, missed methodological lessons, and weak theoretical grounding. Use databases like PubMed, Google Scholar, and HINARI to search systematically before finalising your proposal.
3. Choosing a Sample Size by Intuition
Sample size should never be determined by guesswork or convenience. Underpowered studies fail to detect true effects; overpowered studies waste resources. Use a formal sample size calculation based on your primary outcome, expected effect size, significance level (usually α=0.05), and desired power (usually 80–90%). Free tools like OpenEpi and G*Power can guide you through this process.
4. Using an Unvalidated Data Collection Tool
Creating a questionnaire from scratch without validation is a common trap. If you are measuring a construct such as depression, anxiety, or knowledge, use a pre-validated instrument (e.g., PHQ-9 for depression) whenever possible. If you must create a new tool, pilot it with at least 10–15 participants who match your target population, and check for clarity, language, and internal consistency.
5. Ignoring Ethical Approval
Ethical clearance is not optional. Any research involving human participants requires approval from an Institutional Review Committee (IRC) or the Nepal Health Research Council (NHRC). Starting data collection before clearance is a serious violation that can invalidate your entire study. Apply early, as approval can take 4–8 weeks. Include informed consent forms, data confidentiality plans, and risk–benefit assessments in your submission.
6. Poor Data Management
Data collected on paper forms often gets lost, smudged, or incorrectly transcribed. Use digital data entry tools such as KoBoToolbox or REDCap from the start. Apply range checks, skip logic, and mandatory fields to reduce entry errors. Back up your dataset to at least two separate locations (e.g., an encrypted hard drive and a cloud service) throughout data collection.
7. Misinterpreting Statistical Results
A p-value below 0.05 does not automatically mean your finding is important or clinically meaningful. Always report effect sizes (odds ratios, risk ratios, mean differences, Cohen's d) alongside p-values. Confidence intervals tell you the precision of your estimate. And remember: statistical significance is not the same as practical significance. A 0.5 mmHg reduction in blood pressure may be statistically significant in a large sample but clinically irrelevant.
8. Writing Discussion Without Comparing to Existing Literature
Your discussion should situate your findings within what is already known. For each major finding, ask: Does this agree or disagree with previous studies? If it disagrees, why might that be — different population, methods, or context? Students often write discussions that simply restate their results without engaging with the broader literature. This weakens the scientific contribution of the work.
9. Leaving Limitations as an Afterthought
Limitations are not a weakness to hide — they are a sign of scientific rigour when acknowledged transparently. Common limitations include cross-sectional design (cannot establish causality), self-reported data (recall bias), convenience sampling (limits generalisability), and small sample sizes. Acknowledge each limitation, explain its likely impact on your findings, and, where possible, describe what you did to mitigate it.
10. Citing Irrelevant or Low-Quality Sources
Grey literature, Wikipedia, and predatory journals do not belong in a scientific research paper. Prioritise peer-reviewed articles from indexed journals (PubMed, Scopus, Web of Science). When citing Nepali data, use national surveys (NDHS, STEPS), government reports, and NHRC publications. Use a reference manager like Zotero to organise and format citations consistently.
Conclusion
Avoiding these ten mistakes will significantly improve the quality, credibility, and publishability of your research. If you are unsure about any aspect of your study design, data analysis, or write-up, the Research Sathi team is here to help. Good research is not about perfection — it is about being systematic, transparent, and honest about what your study can and cannot tell us.
Dr. Suresh Poudel
Senior Research Consultant — Research Sathi
Our team brings together experienced researchers, statisticians, and academic writers dedicated to helping students and professionals across Nepal produce impactful research.