1.

Data Collection Methods

This section can take around 50–60 minutes in your workshop. The aim is to move from “what
we do” (methods) to “how we should do it responsibly” (ethics in Section 2).
1.1 Quantitative Data Collection Methods
a) Surveys and Questionnaires
Description:
Standardized instruments administered online (Google Forms, Qualtrics), on paper, or via mobile
apps to collect numerical or categorical data from many respondents.
Use cases (education / social science):
 Measuring attitudes toward AI in teaching
 Collecting student satisfaction scores
 Capturing technology adoption levels
Strengths:
 Can reach large, geographically dispersed samples
 Easy to quantify and analyze (SPSS, R, SmartPLS)
 Standardized questions → comparability across respondents
Limitations:
 Superficial understanding of complex phenomena
 Response bias (social desirability, acquiescence)
 Low response rates if incentives and design are poor
Ethical issues to highlight:
 Need for informed consent at the beginning of the survey
 Avoid collecting unnecessary identifiers (names, roll numbers)
 Avoid coercion (e.g., “fill this or lose marks”)
 Ensure anonymity/confidentiality in reporting
Example for slides:
“Online survey of 500 faculty across India on their reluctance to disclose AI use in research.”

b) Structured Observations
Description:
Researcher uses an observation checklist (e.g., frequency counts, rating scales) in classrooms,
labs, or workplaces.
Use cases:
 Observing classroom technology use
 Tracking student engagement behaviours
 Studying customer interactions in a service setting
Strengths:
 Direct behaviour rather than self-report
 Useful when participants cannot accurately recall or articulate behaviour
Limitations:
 Observer bias and reactivity (Hawthorne effect)
 Time-consuming and context-dependent
Ethical issues:
 Whether participants know they are being observed
 When is covert observation acceptable? (Rare and high scrutiny)
 Need to protect identities when reporting sensitive behaviours
Example:
Observing how often students use ChatGPT or other AI tools during lab time.

c) Experiments and Quasi-Experiments

Description:
Manipulating one or more independent variables to see effects on dependent variables.
Use cases:
 Testing whether AI feedback vs. human feedback improves student performance
 Studying how different types of AI explanations affect trust
Strengths:
 Stronger causal inference
 Control over conditions, random assignment (in true experiments)
Limitations:
  Artificial settings may reduce ecological validity
 Hard to implement full randomization in real classrooms
Ethical issues:
 Ensuring participants are not harmed or disadvantaged by being in the “control” group
 Avoiding deception, or at least debriefing thoroughly if used
 Clear consent, especially where grades or career outcomes might be affected

d) Use of Existing / Secondary Datasets

Description:
Using data already collected by others (institutional records, national surveys, platform logs).
Use cases:
 Analysing university LMS usage logs
 Mining MOOC platform data
 Re-using large published survey datasets
Strengths:
 Cost-effective and time-efficient
 Enables work with large, sometimes representative samples
Limitations:
 No control over how data was originally collected
 Variables may not align perfectly with your research question
Ethical issues:
 Were participants informed that data might be used for research?
 Are you allowed to link datasets in a way that re-identifies individuals?
 Need for data sharing agreements, especially with institutions

1.2 Qualitative Data Collection Methods

a) In-depth Interviews
Description:
One-on-one, semi-structured or unstructured conversations exploring experiences, perceptions,
meanings.
Use cases:
 Understanding why faculty hesitate to disclose AI use
 Exploring students’ lived experiences with online learning
 Investigating ethical concerns about surveillance technologies
Strengths:
 Depth and richness of data
 Flexibility to probe and clarify
Limitations:
 Time-consuming to conduct and transcribe
 Smaller samples → limited generalizability
 Researcher’s own perspective can shape interpretation
Ethical issues:
 Managing disclosure of sensitive or incriminating information
 Ensuring confidentiality in transcription and storage
 Emotional distress: discussing failure, fear, discrimination
Example:
Interview questions like: “What worries you most about revealing that you used generative AI in
your last paper?”

b) Focus Groups
Description:
Group discussions (6–12 participants) led by a moderator to explore attitudes, norms, shared
experiences.
Use cases:
 Faculty focus group on policy needs for AI in assessment
 Student focus groups on privacy concerns in LMS analytics
Strengths:
  Participants build on each other’s ideas
 Efficient way to generate many perspectives
Limitations:
 Dominant voices can overshadow quieter participants
 Social desirability bias amplified
 Harder to maintain confidentiality among participants
Ethical issues:
 Cannot guarantee full confidentiality (others in the group may talk outside)
 Need strong ground rules (respect, no naming people, no sharing stories outside)
 Special care when power differences exist (e.g., faculty + students together)

c) Ethnography and Participant Observation

Description:
Extended immersion in a setting to understand culture, practices, meanings from the inside.
Use cases:
 Studying institutional culture around technology adoption
 Understanding informal norms about plagiarism and AI in student communities
Strengths:
 Deep contextual understanding
 Captures real-life complexity
Limitations:
 Long-term commitment
 Difficult to “bracket” researcher influence
 Large amounts of unstructured data
Ethical issues:
 Blurred roles (researcher vs. participant vs. insider)
 Managing long-term confidentiality and trust
 Dealing with observations of harmful or illegal behaviour
1.3 Digital, Online, and Big Data Methods
a) Web and Social Media Data
Description:
Collecting posts, comments, likes, usage patterns from platforms (X/Twitter, Reddit, YouTube,
learning forums).
Use cases:
 Analysing discourse about AI tools in academic communities
 Studying student discussions on online platforms about exam prep and cheating
Strengths:
 Naturalistic, large-scale data
 Time-stamped, networked interactions
Limitations:
 Platform-dependent (APIs change or get restricted)
 Data can be messy, noisy, unstructured
Ethical issues:
 Public vs. private: Is a “public” tweet ethically free to use without consent?
 Need to consider user expectations, not just legal terms of service
 De-identifying usernames when quoting posts

b) Learning Analytics and Educational Platforms

Description:
Using activity logs from LMS (Moodle, Canvas), MOOCs, apps (Duolingo), proctoring systems.
Use cases:
 Tracking time-on-task, clickstream data
 Identifying at-risk students for interventions
 Evaluating engagement with AI-based tools embedded in LMS
Strengths:
 Continuous, fine-grained behavioural data
 Supports early warning systems and personalization
Limitations:
 High risk of misinterpretation (a “long login time” ≠ learning)
 Technical and algorithmic complexity
Ethical issues:
 Surveillance concerns – do students know they are being tracked at this granularity?
 Consent and opt-out options for analytics-based research
 Risk of profiling and discrimination (e.g., labelling someone as “low potential”)

1.4 Mixed Methods

Description:
Combining quantitative and qualitative data collection in sequenced or parallel designs.
Use cases:
 Survey on AI disclosure reluctance + follow-up interviews to explain patterns
 Pre/post-test learning outcomes + classroom observations + student focus groups
Strengths:
 Breadth + depth
 Triangulation enhances credibility
Limitations:
 More complex design and analysis
 Higher demands on time and skills
Ethical issues:
 Managing multiple layers of consent (for survey and interview, etc.)
 Ensuring that integration of datasets does not make participants re-identifiable
 Communicating clearly to participants exactly what each phase involves

2. Overall Research Ethics

This section can take the next 60–70 minutes. The idea is to link the above methods to ethical
principles, processes, and decision-making.
2.1 Core Ethical Principles
Use this as an anchor framework (you can reference Belmont Report + ICMR guidelines).
1. Respect for Persons (Autonomy)
 Treat participants as agents capable of making their own decisions.
 Key practices:
 Informed consent (clear information, voluntary participation).
 Right to withdraw without penalty.
 Extra care with vulnerable groups (students, patients, low-income
populations).
2. Beneficence and Non-Maleficence
 Maximize potential benefits, minimize potential harms.
 Ask: “What risks (psychological, social, reputational, legal) might this research
create?”
 Plan support or referral mechanisms for participants in distress.
3. Justice
 Fair selection and fair treatment of participants.
 Avoid over-using “easy” groups (own students, one institution).
 Avoid burdening vulnerable populations without proportionate benefits.
4. Transparency and Accountability
 Be open about goals, funding, methods, risks, and limitations.
 Take responsibility for mistakes and proactively correct them.
 Maintain documentation (protocols, consent forms, data management plans).
5. Privacy and Confidentiality
 Distinguish between:
 Privacy: what information people choose to share
 Confidentiality: how you protect what they have shared
 Implement de-identification, secure storage, strict access control.
6. Integrity and Honesty
 No fabrication, falsification, selective reporting, or “p-hacking.”
 Report methods and limitations honestly.
  Acknowledge use of AI tools in analysis or writing where relevant.

2.2 Ethics Across the Research Lifecycle

You can walk participants through the lifecycle and pause at each stage to ask: “What can go
wrong ethically here?”
1. Problem Formulation & Design
 Are research questions socially responsible and respectful?
 Could the research reinforce stereotypes or biases?
 Is the population appropriate and fairly chosen?
2. Sampling and Recruitment
 Are people free to decline without repercussions?
 Are incentives modest and non-coercive?
 Is recruitment information accurate and non-misleading?
3. Data Collection
 Is the chosen method appropriate and proportionate to risk?
 Have participants been fully informed (aim, procedure, risks, benefits)?
 Are there procedures if unexpected issues arise (distress, disclosures)?
4. Data Management (Storage, Access, Security)
 Where is data stored? Personal laptop vs. institutional server?
 Is data encrypted at rest and in transit?
 Who has access, and is it logged?
5. Data Analysis
 Are analyses consistent with what participants consented to?
 Are subgroup analyses small enough to re-identify people?
 Is AI being used for analysis, and if so, with what safeguards and disclosure?
6. Reporting and Publication
 Are individuals or institutions identifiable in case studies or quotes?
 Are results overstated, cherry-picked, or sensationalized?
 Is ethics approval disclosed, along with consent procedures?
 7. Data Sharing, Retention, and Deletion
 How long will data be kept and why?
 Under what conditions, if any, will it be shared with other researchers?
 How will data be securely destroyed?

2.3 Informed Consent: Central Ethical Mechanism

You can emphasize that consent is a process, not just a signature.
Key elements to cover:
 Purpose of study (in plain language).
 What participation involves (time, actions, setting).
 Risks and discomforts (even if minimal).
 Benefits (realistic, not exaggerated).
 How privacy and confidentiality will be protected.
 Data retention and possible future use.
 Voluntariness and right to withdraw.
 Contacts for questions, complaints, and ethics concerns.
Example you can discuss in class:
Show a “bad” one-paragraph consent and ask them to fix it into a proper one-page sheet.

2.4 Typical Ethical Dilemmas (Good for Discussion)

You can present these as mini-scenarios and ask participants to vote or discuss.
1. Using student assignment data collected for grading as research data
 Without prior consent: ethically problematic.
 With proper information to students and opt-out: more defensible.
2. Discovering plagiarism or academic misconduct in qualitative data
 Duty to protect confidentiality vs. duty to uphold academic integrity.
 Emphasize: consent forms should pre-specify the conditions under which
confidentiality can be broken.
3. Offering extra credit for research participation
  Risk of coercion, especially in courses you teach.
 Safer to offer alternate, equally attractive non-research ways to earn the same
credit.
4. Publishing “de-identified” but recognizable case narratives
 Combining too many specific details can reveal identity even without names.
 Good practice: change non-essential details and obtain explicit permission for
case use if risk is high.

2.5 Ethics with AI and Digital Tools (Very Relevant for Your Audience)
Given your specialization, this can be a strong closing theme.
Key points:
 Using AI (e.g., ChatGPT) in data analysis or writing
 Disclose use transparently in methods or acknowledgements.
 Do not feed confidential/raw participant data into external AI tools unless
explicitly allowed and protected.
 Maintain human responsibility for interpretation and final conclusions.
 Algorithmic Bias and Fairness
 When using predictive models (early warning systems, risk scores), examine
whether they disadvantage particular groups.
 Engage in fairness-aware evaluation.
 AI in Data Collection (chatbots, automated interviewers, proctoring)
 Participants must know when they are interacting with AI vs. humans.
 Assess additional privacy risks: facial recognition, keystroke logging, webcam
monitoring.