How to Craft Effective Research Questions: Avoiding Common Pitfalls and Leveraging AI Tools

Introduction

The lecture focused on the practical side of developing research questions, emphasizing that understanding bad questions is essential for creating good ones. After a brief slide review, the session moved to hands‑on activities using the KIS platform.

Understanding Bad Research Questions – Six Common Attributes

1. Fuzzy Questions – Vague, lacking clear boundaries; e.g., "What is the impact of tobacco influences on youth's reactions?" fails to meet the PICO criteria.
2. Missing Question Mark – A statement rather than a question; research must be framed as an interrogative to signal intent to generate new knowledge.
3. Too‑Much‑Information (TMQ) – Overly long, multi‑part questions that obscure the primary variable and overwhelm readers.
4. Unanswerable Questions – Ask for causal language (effect, cause) that cannot be supported by the study design (e.g., cross‑sectional data cannot establish initiation or causality).
5. Recycled / Salami‑Sliced Questions – Fragmenting one solid research idea into many tiny papers, wasting reviewer time and resources.
6. Questionable Impact Questions – Well‑structured but of limited relevance to an international audience; such papers may be rejected for low impact.

The PICO/PICOT Framework

• Population, Intervention (or Exposure), Comparison, Outcome, Time, and sometimes Study design.
• A well‑defined question includes all applicable elements; descriptive studies may omit the comparison or intervention component.

Intervention vs. Exposure

• Intervention: Experimental assignment by the investigator (e.g., giving a drug).
• Exposure: Naturally occurring factor observed without researcher control (e.g., environmental tobacco smoke).
• Every intervention is an exposure, but not every exposure is an intervention.

Association vs. Effect

• Association: A bidirectional relationship; the direction of causality is unknown.
• Effect/Cause: Implies a directional, causal link, which requires appropriate analytic designs (e.g., randomized trials, cohort studies).

Why the Question Mark Matters

• A research question must be a true question (interrogation) to fulfill the two core criteria of research: intent to create new knowledge and systematic approach.
• Without a question, the activity is merely data collection for administrative purposes.

Unanswerable Questions in Context

• Example: "What is the effect of e‑cigarette use on cigarette smoking initiation among youths?" – impossible to answer with a cross‑sectional survey because initiation implies a temporal sequence.
• Solution: Re‑frame to "What is the association between e‑cigarette use and current cigarette smoking?"

Salami Slicing and Repetition

• Salami slicing: Publishing multiple papers from the same dataset by carving out tiny, incremental questions.
• Confirmatory research: Repeating studies in different contexts is acceptable when the new context (geography, population) may yield different results; otherwise it adds little value.
• Ethical concerns: waste of reviewer time, taxpayer funds, and dilution of evidence.

Questionable Impact Questions

• Even well‑crafted, locally focused KAB (knowledge‑attitude‑behavior) studies may be rejected by international journals because they lack global relevance.
• Researchers should aim for questions that can spark broader scientific conversation.

Knowledge Gaps

• Vertical gaps: Progression from basic → theoretical → efficacy → applied → public‑health knowledge.
• Horizontal gaps: Same unit of analysis but different contexts (geography, time, population, methodology).
• Use interrogative pronouns (what, where, when, who, how) to articulate these gaps.

Aligning Questions with Academic Stage

Leveraging the KIS Platform

• Research Question Generator: Input a rough idea; the tool suggests well‑structured PICO‑based questions and alternative formulations.
• Proposal Builder: Select study design (cross‑sectional, case‑control, cohort, etc.) and automatically populate sections such as background, aims, objectives, hypothesis, methodology, and statistical plan.
• Future Features: Automated analysis recommendations, abstract/manuscript generation, data upload for analysis.

Practical Writing Tips

• Keep manuscripts concise; aim for strict word limits to avoid fluff.
• Use online supplements for tables/figures when journals impose tight limits.
• Ensure a single overriding message (SOO) throughout the paper to aid media and reviewer comprehension.
• When faced with word‑limit constraints, rewrite concisely or choose a journal with a higher allowance.

Q&A Highlights

• Clarified the meaning of KAB studies (knowledge‑attitude‑behavior/practice).
• Discussed ethical concerns of salami slicing and the importance of adding genuine value.
• Explained how to choose between retrospective cohort and case‑control designs based on data availability and outcome rarity.
• Confirmed that KIS‑generated references are curated, not AI‑fabricated.
• Outlined upcoming platform capabilities (analysis suggestions, manuscript drafting, Excel data upload).

Conclusion

The session equipped participants with a checklist for spotting bad research questions, reinforced the necessity of the PICO framework, and demonstrated how AI‑assisted tools like KIS can streamline question formulation and proposal development while maintaining scientific rigor.

Avoid vague, overly complex, or unanswerable questions by applying the PICO framework, ensuring a true interrogative form, and matching the question to the appropriate study design. Use tools such as the KIS platform to generate and refine questions, but always align the scope and impact of your research with your academic level and the target journal’s audience.