Graduate‑Level Research Session: Framing Gaps, AI Ethics, Questionnaire Design, and K‑Q Platform

Introduction

The session covered a wide range of topics that are essential for graduate‑level research, from framing research gaps with interrogative pronouns to the ethical use of artificial intelligence (AI) in dissertation writing, and from designing robust questionnaires to using the K‑Q platform for data collection, cleaning, and analysis. Participants asked numerous practical questions, and the facilitator provided detailed, step‑by‑step guidance.

Framing Research Gaps with Interrogative Pronouns

The facilitator explained how the five interrogative pronouns—what, where, when, how, and who—can be used to highlight the novelty of a study:

The key message was to frame the gap in knowledge using these pronouns, thereby making the research question compelling and novel.

Using AI in Dissertation and Thesis Writing

COPE Guidelines on Authorship and AI

• AI tools cannot be listed as authors because they are non‑legal entities that cannot take responsibility for conflicts of interest, copyright, or ethical breaches.
• Authors must disclose in the Materials and Methods section how an AI tool was used and which tool was used.
• Authors remain fully responsible for all manuscript content, including parts generated by AI.

Risk‑Based Framework for AI Use

The facilitator proposed dividing research tasks into low‑risk and high‑risk categories:

• Low‑risk tasks – summarising text, expanding sections, creating thematic groupings, etc.
• High‑risk tasks – any activity that involves data analysis or interpretation; AI is not recommended for these because current AI models lack the required sophistication.

The emphasis is on principles rather than strict laws: focus on adding value, understand when AI is appropriate, and maintain rigorous oversight.

Designing Questionnaires and Data Collection

Critiquing a Narrow Study Topic

A volunteer presented a study on “dietary adequacy and nutritional status of adult tuberculosis patients attending Jumman General Hospital.” The facilitator highlighted several issues:

• Scope too narrow – publishing in a high‑impact international journal is unlikely when the study is limited to one hospital in one district.
• Journal expectations – editors prefer papers that are anchored in strategic policy, have international relevance, and avoid overly parochial topics.

Suggested improvements

1. Broaden the focus – shift from a single hospital to a national survey (e.g., “South Africa”).
2. Change the design – convert the descriptive study into an analytical (case‑control) study that examines associations (e.g., between dietary adequacy and mortality among people living with tuberculosis).
3. Use respectful language – replace “tuberculosis patients” with “people living with tuberculosis.”

PYOT (Population, Intervention, Control, Outcome, Time) Framework

When planning a national survey, the facilitator recommended filling out the PYOT elements:

• Population – define clearly (e.g., community‑based vs. hospital‑based).
• Intervention – “not applicable” for a purely descriptive study.
• Control – “not applicable.”
• Outcome – self‑reported dietary adequacy and nutritional status.
• Time – allow at least six months for planning and 12 months for data collection (e.g., July 2024 – January 2025).

Building a Sampling Frame

• Identify community organisations, patient support groups, or existing panels that serve the target population.
• Use Google searches to locate organisations, filter for the relevant country (e.g., Uganda), and contact them for referrals.
• Construct a sampling frame by compiling contacts, verifying eligibility, and documenting the recruitment strategy in grant proposals.

Formative Research & Stakeholder Involvement

• Conduct literature reviews and interview clinicians, patients, and community leaders to identify current priorities and gaps.
• Follow the principle “nothing about us without us.”

Questionnaire Structure

1. Screeners – eligibility based on country, age, and diagnosis (blinded to avoid false positives).
2. Informed consent – generate default consent content; add assent for minors if needed.
3. Core questions – demographic items, household characteristics, nutrition, health‑seeking behaviour, etc.
4. Required fields – mark a question as required by checking the appropriate box.

Using the K‑Q Platform

Common Platform Questions

• Making a question mandatory – check the “required” box on the question settings.
• Post‑survey use – the platform can be used for any data‑collection activity, including evaluations.
• Preventing duplicate responses – disable link forwarding for closed surveys or upload a list of authorised email addresses.
• Generating unique links for multiple data collectors – use the offline mode with agent validation; each agent’s data are tagged.
• Handling disqualified participants – inclusion criteria logic automatically displays an ineligibility message and stops the survey.
• Viewing response counts – the platform reports only those who answered eligibility questions; raw click‑through numbers are not reported as scientific data.

Guidance on Specific Research Topics

Action Steps

1. Frame research gaps using the five interrogative pronouns (what, where, when, how, who).
2. Follow COPE guidelines: do not list AI tools as authors; disclose AI usage in Materials and Methods; retain full responsibility for manuscript content.
3. Apply a risk‑based framework for AI: use AI only for low‑risk tasks (summarising, theme generation); avoid AI for data analysis or interpretation.
4. Broaden narrow study topics by expanding geographic scope (national) or changing design (descriptive → analytical).
5. Complete a PYOT table before drafting the questionnaire.
6. Build a sampling frame by contacting community organisations, patient groups, or commercial panels; document the process in grant proposals.
7. Conduct formative research with literature reviews and stakeholder interviews to confirm relevance and novelty.
8. Design questionnaire screeners that are blinded to the study focus; set inclusion criteria for country and age.
9. Generate consent/assent using the platform’s default templates; edit as needed.
10. Prepare questionnaire for import: prepend “Q” to each question, “A” to each answer, end lines with “###”; use Excel to automate.
11. Add special symbols for question types (e.g., “!” for drop‑down, “L” for Likert).
12. Set logic and inclusion criteria in the platform; test both country and age branches repeatedly.
13. Preview the survey before launch; verify all logic, required fields, and consent flow.
14. Enable quality‑assurance settings: disable back button, disable link forwarding for closed surveys, require numeric answers where appropriate.
15. Launch the survey after confirming dates (e.g., start now, end 31 Dec).
16. Monitor responses; the platform will automatically clean data and provide a raw dataset if needed.
17. Download the methodology report, analysis report, and codebook directly from the platform.
18. Interpret the data quality score (duplicates, straight‑lining, missingness, speed) when writing the dissertation’s quality‑assurance section.
19. Use offline mode for field data collection: create the survey online, download to device, collect data offline, then sync when internet is available.
20. Track data collectors by using the offline mode with agent identifiers; each agent’s responses are tagged.
21. Prevent duplicate entries by disabling link forwarding or uploading an authorised email list.
22. Make questions required by checking the “required” box in the question settings.
23. Add post‑survey functionality (e.g., thank‑you messages, data download options) via the platform’s settings.

Following these steps will help researchers design robust, ethically sound studies, efficiently manage questionnaire development, and leverage the K‑Q platform for high‑quality data collection and analysis.

The session emphasized that framing research gaps with interrogative pronouns creates compelling questions, while ethical AI use requires disclosure, authorship exclusion, and a risk‑based task approach. Expanding narrow topics to national or analytical designs and applying the PYOT framework strengthens relevance and publication potential. Building a rigorous sampling frame and conducting formative stakeholder research ensures representative, ethical data collection. Mastery of the K‑Q platform—through proper import tags, logic, quality‑assurance settings, and offline capabilities—enables efficient questionnaire deployment and high‑quality data management. Following the detailed action steps integrates these principles into a systematic, reproducible research workflow.